Berchmansus elegans (Neuroptera: Chrysopidae): Larval and adult characteristics and new tribal affiliation

The genus Berchmansus Navás, which was previously assigned to the tribe Leucochrysini, consists of three very rare species, all described from the Neotropics and all poorly known. Our report (1) provides the first description of a Berchmansus larva, the first instar of Berchmansus elegans (Guérin Méneville), (2) illustrates and redescribes the B. elegans adult, with emphasis on male and female genitalia, and (3) examines the larval and adult characters vis-à-vis the tribal affiliation of the genus. Given that the B. elegans adult and first instar share many apomorphies with other belonopterygine genera, this species belongs in the cosmopolitan tribe Belonopterygini, rather than the New World tribe Leucochrysini. Although Berchmansus larvae have not been collected in the field, we suspect that, like other belonopterygines, they are associated with ant nests. B. elegans exhibits a number of highly modified and unusual structures, some of which (#1 to #5) are not reported for any other chrysopids. Specifically: Males have (1) a unique, quadrate, dome-like hood above the gonarcus and (2) large, coiled parameres on the gonosaccus. First instars have (3) a greatly enlarged subapical seta on the flagellum, (4) a transverse row of long, hooked setae along the dorso-anterior margin of the pronotum, and (5) setose laterodorsal tubercles on the mesoand metathorax, with (6) multi-pronged, hooked setae. 221 * Dedication. We dedicate this paper to two highly distinguished neuropterologists, both of whom did pioneering work on Belonopterygini: Professor Maria Matilde Principi (University of Bologna), who recognized the uniqueness of Italochrysa and provided superb descriptions and drawings of both the adult and larva; and Professor Bo Tjeder (University of Lund, deceased 1992), whose admirable studies documented the characteristics and much of the diversity exhibited by Old World members of this chrysopid trib e. also report that our specimens of male B. elegans differ notably from Brooks & Barnard’s (1990) descriptions and illustrations of the species*. Our study here presents (1) a description of the B. elegans first instar, (2) redescriptions of B. elegans males and females, with emphasis on generic-level head, wing, and genital characters, and (3) a review of B. elegans larval and adult characters vis-à-vis the tribal affiliation of the genus. Specifically, we compare the B. elegans larvae and adults with previously described leucochrysines and belonopterygines (Larvae: Principi, 1946; New, 1983, 1986; Díaz-Aranda & Monserrat, 1995; Tauber, 2004; Mantoanelli et al., 2006; Adults: Tjeder, 1966; Adams, 1977, 1979, 1987; Brooks & Barnard, 1990; Tsukaguchi, 1995; Tauber, 2004). Finally, we propose (4) the transfer of B. elegans to the tribe Belonopterygini, and we identify (5) a suite of fifteen first instar characters that should be included in the phylogenetic analysis of higher-level lineages within the Chrysopidae. MATERIAL AND METHODS Procedures and terminology The rearing, preservation, and descriptive procedures were presented earlier (Tauber et al., 1998, 2000). Nomenclature for adult characters (head, mouthparts, wings and genitalia) follow those in current usage (Snodgrass, 1935; Tjeder, 1966; Adams, 1977; Brooks & Barnard, 1990; Tauber, 2003). Nomenclature for larval morphology follows that of Rousset (1966) for the head, and Tauber et al. (2000) and Tauber & de Léon (2001) for the thorax and abdomen. Primary setae are designated with an “S” followed by a number, and in the case of appendages, a letter (or letters). Because chrysopids (all instars) consistently have three setae on the anterior margin of the clypeus (cl), we designate these setae as primary cephalic setae S13-cl, S14-cl, and S15-cl (see Fig. 1). Measurements of both adult and larval head width were made across the widest part of the head and include the eyes. The ratio (head-width : eye-width) was the distance between the eyes across the vertex : mean width of left and right eyes (see Brooks & Barnard, 1990). Wing width was measured at the widest part of the wing. Larval body length was measured along the curved midlateral line of distended specimens; head length was determined along the midline from the tip of the clypeus to the 222 Figs 1–5. Berchmansus elegans, first instar. 1 – head dorsal; 2 – mandible; 3 – maxilla; 4 – labial palpus; 5 – antenna. epi – epicranial marking; fl – flagellum; fr – frontal marking; gen – genal marking; ped – pedicel; sc – scape; sp – spur; S1–S7, S11–S15, Vx – primary setae. The pair of unlabelled setae below the labial palps in Fig. 1 are secondary setae. * After this manuscript went to press, we had the opportunity to examine the types of B. elegans (male holotype and two female paratypes; Koninklijk Belgisch Instituut voor Natuurwetenschappen, Brussels). Consequently, we have confirmed the identity of our specimens as B. elegans. We also examined the specimens that Brooks & Barnard (1990) referred to as “B. elegans” but that differed significantly from ours. We note that Brooks & Barnard’s figures of “B. elegans” actually refer to a new species of Berchmansus that closely resembles B. elegans externally, but has significantly different male genitalia and is slightly smaller. We thank J. Constant and D. Goodger for making the specimens available to us, and S. Brooks for calling our attention to the differences between his


INTRODUCTION
The green lacewing genus Berchmansus Navás, 1913, provides a prime example that illustrates the essential contributions of larval morphology and comparative biology to the classification and phylogeny of Neuroptera (e.g., Stange & Miller, 1990;Díaz-Aranda & Monserrat, 1995;Aspöck et al., 2001;Monserrat et al., 2001;Tauber, 2003;Grebennikov 2004). Berchmansus contains three neotropical species that are very rare in collections and that have received very little attention. Navás (1913) erected the genus with a short description of external adult features. While presenting a brief summary of work in progress, Adams (1978) erected the Chrysopinae tribe Leucochrysini and assigned Berchmansus to it; he did not provide a definition of the tribe or additional information on the genus or the included species.
In their review of the green lacewings, Brooks & Barnard (1990) restructured Leucochrysini to include taxa known only from the New World, and they retained Berchmansus (with its three species) within the Leucochrysini. Because of the paucity of specimens and the tenuous nature of the diagnostic characters, they were unable to confirm the monophyly of the genus. The genus is distinguished primarily by the position of the first radial crossvein, a feature that, like many aspects of the venation in the Chrysopidae, is probably variable within and among chrysopid taxa. They also described a unique component in the male genitalia of Berchmansus elegans (Guérin Méneville, 1844), an arcuate structure ventral to the gonarcus, and they provided a brief description of the Berchmansus adumbratus Navás, 1913, female genitalia.
In 2002 and 2003 we obtained eggs from fieldcollected adults of B. elegans; these eggs hatched and we preserved first instars; we were unable to rear subsequent stages. Features of our first instar specimens led us to suspect that B. elegans is more closely affiliated with the cosmopolitan tribe Belonopterygini than the New World tribe Leucochrysini. Consequently, we re-examined the adult morphological features that had led to the earlier assignment of Berchmansus to Leucochrysini. On the basis of both larval and adult characters, it is now evident that B. elegans' affiliation lies with Belonopterygini. We also report that our specimens of male B. elegans differ notably from Brooks & Barnard's (1990) descriptions and illustrations of the species*.

Procedures and terminology
The rearing, preservation, and descriptive procedures were presented earlier (Tauber et al., 1998(Tauber et al., , 2000. Nomenclature for adult characters (head, mouthparts, wings and genitalia) follow those in current usage (Snodgrass, 1935;Tjeder, 1966;Adams, 1977;Brooks & Barnard, 1990;Tauber, 2003). Nomenclature for larval morphology follows that of Rousset (1966) for the head, and Tauber et al. (2000) and Tauber & de Léon (2001) for the thorax and abdomen. Primary setae are designated with an "S" followed by a number, and in the case of appendages, a letter (or letters). Because chrysopids (all instars) consistently have three setae on the anterior margin of the clypeus (cl), we designate these setae as primary cephalic setae S13-cl, S14-cl, and S15-cl (see Fig. 1).
Measurements of both adult and larval head width were made across the widest part of the head and include the eyes. The ratio (head-width : eye-width) was the distance between the eyes across the vertex : mean width of left and right eyes (see Brooks & Barnard, 1990). Wing width was measured at the widest part of the wing. Larval body length was measured along the curved midlateral line of distended specimens; head length was determined along the midline from the tip of the clypeus to the 222  Berchmansus elegans, first instar. 1 -head dorsal; 2 -mandible; 3 -maxilla; 4 -labial palpus; 5 -antenna. epi -epicranial marking; fl -flagellum; fr -frontal marking; gen -genal marking; ped -pedicel; sc -scape; sp -spur; S1-S7, S11-S15, Vxprimary setae. The pair of unlabelled setae below the labial palps in Fig. 1 are secondary setae. * After this manuscript went to press, we had the opportunity to examine the types of B. elegans (male holotype and two female paratypes; Koninklijk Belgisch Instituut voor Natuurwetenschappen, Brussels). Consequently, we have confirmed the identity of our specimens as B. elegans. We also examined the specimens that Brooks & Barnard (1990) referred to as "B. elegans" but that differed significantly from ours. We note that Brooks & Barnard's figures of "B. elegans" actually refer to a new species of Berchmansus that closely resembles B. elegans externally, but has significantly different male genitalia and is slightly smaller. We thank J. Constant and D. Goodger for making the specimens available to us, and S. Brooks for calling our attention to the differences between his specimens and ours. cervix; mandible length was measured along the center of the mandibles' curved dorsal surface. All measurements were made with image analysis software (Image-Pro Plus, Version 3).

Specimens examined
Adults

Distribution and seasonal occurrence
B. elegans is reported from Brazil, Surinam and Costa Rica (Penny, 2002). We collected adults in the Brazilian Atlantic Forest in March, April, May, and November, and the species is reported from Costa Rica in May (Penny, 2002).

Habitat and field observations
All of the B. elegans adults that we found in the field were resting on the undersides of leaves on trees and shrubs in secondary growth forests. Leaves in these forests tend to have numerous dark blotches that are visible against the sky; the pattern of these blotches provides considerable crypsis for B. elegans adults, with their spotted wings.
In general, when seeking B. elegans adults in the field, we found only one or two specimens. However, on one occasion (May 1, 2003), we discovered a large cluster of adults resting on the undersides of leaves of trees in a relatively cool, shady, protected area of approximately 100 square meters within a mixed-species secondary growth forest. We observed these adults in the field through the afternoon until dark; they remained virtually motionless with their wings held flat to the substrate and their abdomens curved to one side or the other. Occasion-
On May 5, we returned to the same area. Again adults were present; we collected eight specimens (four males, four females) and left many others. On this occasion, the adults appeared to be slightly higher in the trees than on the previous trip four days earlier. On both trips, we searched for eggs (on leaves, branches, tree trunks, near the ground, etc.), but none were found. One month later (June 1), our four-person team spent the whole afternoon searching the trees where adults were previously found; the effort resulted in no adults, eggs or larvae in the area. We found only one adult ca. 50 meters away from the site.

Laboratory observations
Three females oviposited in the laboratory. Eggs were laid singly on long stalks. The maximum number of eggs produced in any one day was four; usually only one or two. The largest number of eggs obtained from a single female was fourteen over a period of five days. Hatching occurred within eight to 12 days of oviposition (temperature variable).
Newly hatched larvae remained on the chorion for several hours; subsequently they walked around the vial with a steady gait or rested on the cotton plug. Larvae were provided water and a mixture of protein hydrolysate of yeast, honey, and sugar, and a variety of prey (e.g, several species of aphids, whiteflies, mealybugs, spider eggs, honeybee larvae and pupae, flies and caterpillars). None were observed to feed on the prey, and all died within four to five days without moulting.
As is typical of many chrysopid taxa, the larvae exhibited trash-carrying behavior. That is, they grasped small pieces of debris with their jaws, curved the upper parts of their bodies dorsally, and placed the debris on the dorsum of their thorax and abdomen. The debris generally remained affixed to the larva's dorsal surface.

Comparative larval morphology
Few data on belonopterygine and leucochrysine larvae are available for morphological comparison with B. elegans. Among the 15 genera of Belonopterygini, larvae are known for two (Italochrysa Principi, 1946, andCalochrysa Banks, 1943) (Principi, 1946;New, 1983New, , 1986Díaz-Aranda & Monserrat, 1995). And, among the seven genera of Leucochrysini, larvae are known for two subgenera of only one genus, Leucochrysa McLachlan, 1868 (subgenera Leucochrysa and Nodita Navás, 1916) (Tauber, 2004;Mantoanelli et al., 2006). Thus, the basis for comparison with B. elegans is small. Nevertheless, given the increased understanding of larval variation among chrysopids that has developed over the past 25+ years, sufficient data exist for developing testable hypotheses about the generic and higher-level phylogenetic relationships of B. elegans.
To initiate the discussion, we propose a series of traits that we consider to have phylogenetic significance and that differentiate the known first instars in the two tribes (Belonopterygini and Leucochrysini). The character states for each of the tribes are derived from descriptions and/or figures contained in the publications listed. To assess the tribal affiliation of Berchmansus, we indicate the character state of the B. elegans first instar within each category. present on mesothorax and metathorax, bearing about nine long or medium-length LDS; unique (Fig. 7).

Comparative adult morphology
Among the chrysopid tribes, adults in the Leucochrysini have received the least attention from modern systematists (the primary synthesis of adult characters being that of Brooks & Barnard, 1990). Synapomorphies for the genera and the tribe are not well defined, and the monophyly of the tribe is in question.
In contrast, some regional belonopterygine faunae (adults) have received modern systematic treatment (primarily by Tjeder, 1966;also Tsukaguchi, 1995). Consequently, the adult characteristics of genera in the tribe are reasonably well defined, a number of synapomorphies that have been proposed for belonopterygine adults, and the tribe is considered to be monophyletic (Brooks & Barnard, 1990).
Unfortunately, in many cases, the characteristics that Brooks & Barnard (1990) described for the Belonopterygini overlap with those proposed for leucochrysine adults. Therefore, for our comparison with B. elegans we chose those that appear to differentiate the two tribes. Below, we discuss each.
B. elegans fits the description for Belonopterygini (Figs 19,20). However, it is noteworthy that our observations of the palps on numerous leucochrysines show considerable variation among taxa, and that Tjeder (1966) and Tsukaguchi (1995) provide numerous examples of variation among belonopterygine genera and species. Thus, the value of this character in diagnosing the tribes is questionable.
2. Mandibles. Leucochrysini: asymmetrical, left mandible with distinct, pointed tooth; Belonopterygini: broad, more or less symmetrical, untoothed, with small basal tooth on both mandibles, or with broad basal tooth on left mandible.
Although the B. elegans mandibles (Fig. 18) are asymmetrical (the left one has a round notch on the inner margin), they are not toothed. Thus, B. elegans resembles the belonopterygine condition more than the leucochrysine.
In B. elegans the toruli are relatively large (Fig. 16), at least as large as those illustrated for Dysochrysa Tjeder, 1966. Moreover, they contact the postfrontal suture as Tjeder ( , Fig. 1133) showed for some belonopterygines.
In B. elegans the antenna is ~0.75 the length of the forewing; thus it does not fall within the leucochrysine condition.
5. Flagellar segments. Leucochrysini: three or more times as long as broad; Belonopterygini: at most two times as long as broad.
In B. elegans, the length : width ratio of the flagellar segments ranges from 1.1 to 2.2 (Fig. 17); thus, it more closely resembles the belonopterygine condition.
In B. elegans, most antennal segments fit this description, but some of the proximal flagellar segments have only three rows of setae (Fig. 17). In this regard, B. elegans is similar to the belonopterygine genus Dysochrysa (see Tjeder, 1966, p. 335).
In B. elegans, c1 is ~0.7× the size of c2 (Fig. 21); thus it more closely resembles the typical leucochrysine condition. Nevertheless, it is noteworthy that in at least two belonopterygine species in two genera [Nesochrysa seyrigi (Navás, 1934) and Belonopteryx arteriosa Gerstaecker, 1863] c1 is shorter than c2 (see Brooks & Barnard, 1990). Thus, the value of the character in differentiating the tribes is open to reassessment.
In B. elegans, a pair of long, coiled, sclerotized structures arises from the gonosaccus (Fig. 23); within the abdomen and on inflated specimens, these structures lie beneath the gonarcus in the position where parameres occur. These coiled structures are similar to the parameres in some other neuropteran families (e.g., Berothidae and Rhachiberothidae; e.g., Tjeder, 1959;Aspöck & Aspöck, 1980Aspöck & Mansell, 1994), but they are very unusual among chrysopids (see Brooks & Barnard, 1990). Given their position and the otherwise broad range of variability among chrysopid parameres (e.g. see Tjeder, 1966;Brooks & Barnard, 1990), we interpret these structures to be an elaborate form of parameres. [It is noteworthy that Brooks & Barnard (1990) did not report or illustrate structures like these in their treatment of B. elegans (see discussion below).] 9. Female -praegenitale. Leucochrysini: praegenitale [sclerotized structure separate from and anterior (ventral) to the subgenitale lobes] absent; Belonopterygini: praegenitale usually present.

Tribal affiliation and prey association
On the basis of the numerous shared character states described above, it is clear that B. elegans belongs in the tribe Belonopterygini. At this point, B. elegans does not appear to fall into any previously described belonopterygine genus, and the monophyly of Berchmansus is in question. We do not have larval specimens of the other two species of Berchmansus, and we have not yet examined the very few known adult specimens of these species. Thus, we cannot confirm their tribal affiliations or their relationships with B. elegans, and until we can do so, we will refrain from naming a new genus for B. elegans.

Species-or genus-specific traits
In comparing B. elegans with species in other belonopterygine genera, certain traits stand out as unique or unusual. Specifically, the adult has a remarkable quadrate hood above the gonarcus and highly modified parameres on the gonosaccus. In addition, the first instar has a greatly enlarged subapical seta on the flagellum, a row of submedian setae on the anterior margin of the pronotum, setose laterodorsal tubercles on the meso-and metathorax, and multi-pronged setae on the thoracic and abdominal tubercles. Most of these characteristics are reported for no other chrysopids. One of our goals is to assess whether these traits are diagnostic at the species or higher levels; an answer to this question would be of considerable value in understanding the diversity and phylogeny of New World chrysopids.

Variation in published descriptions of B. elegans
It is also noteworthy that several features of the male genitalia that Brooks & Barnard (1990) described for B. elegans differ markedly from those in our specimens. For example, Brooks and Barnard did not mention or illustrate the coiled structures (parameres), described above. Also, the quadrate dorsal hood that we observed in our specimens is absent from their illustration, but they report an arcuate structure ventral to the gonarcus on their specimens that is not present on ours. Two explanations can account for these differences: First, the genitalia of chrysopid males typically undergo substantial changes during maturation (Adams, 1977(Adams, , 1987Tauber, 2003); it is possible that sexual maturation in B. elegans includes striking alterations in male genital structures. In this case, Brooks & Barnard's (1990) specimens and ours may encompass a single, very distinctive, but developmentally variable species (B. elegans). We have not identified teneral males among our specimens, so this possibility cannot be tested now.
A second possibility is that Brooks and Barnard's specimens and ours represent different, but cryptic, species in which the distinctive external adult features are similar, but the genitalia differ markedly. To stabilize B. elegans' taxonomic status, and thus resolve the identity of Brooks and Barnard's and our specimens, will require us to examine their specimens, as well as the type specimens of B. elegans (see footnote p. 2). Meanwhile, to preclude possible confusion over the identity of our specimens, we are depositing vouchers in several museums (see Material and Methods).