A New Fishfly Species (Megaloptera: Corydalidae: Neohermes Banks) Discovered from North America by a Systematic Revision, with Phylogenetic and Biogeographic Implications

The taxonomy of Megaloptera from the Nearctic region is fairly well known and their faunal diversity has been largely surveyed, even in relatively remote regions. However, the evolutionary history of Nearctic Megaloptera is still poorly known with phylogenetic and biogeographic studies lacking. In this paper, we report a new fishfly species of the endemic North American genus Neohermes Banks, 1908, increasing the total number known of species to six. This new species (Neohermes inexpectatus sp. nov.) is currently known to occur only in California (USA) and is apparently confined to the Northern Coastal Range. The new species resembles the three Neohermes species from eastern North America based on the relatively small body size and the presence of female gonostyli 9. However, our phylogenetic analysis using adult morphological data recovered the new species as the sister species to the remaining Neohermes, which includes two species from western North America and three from eastern North America. According to the present interspecific phylogeny of Neohermes, with reconstructed ancestral areas, the initial divergence within the genus was found to take place in western North America, with a subsequent eastward dispersal. This likely lead to the modern distribution of Neohermes in eastern North America with the closure of the Mid-Continental Seaway, which separated western and eastern North America in the Mid-Late Cretaceous (100–80 MYA) and finally disappeared at the end of the Cretaceous (70 MYA). The uplift of the Cordilleran System probably accounted for the divergence between the eastern and two western Neohermes species.


Introduction
We present a phylogeny of Neohermes based on adult morphological characters to reveal the systematic position of the new Neohermes species in the genus. Furthermore, we reconstructed the ancestral distribution areas in the phylogeny of Neohermes. The result sheds new light on the biogeography of Megaloptera.
Diagnosis. Adult (Figs 1-5). Medium to large-sized (male forewing length 26-50 mm). Body generally grayish, reddish or blackish brown. Antennae sexually dimorphic; male antenna moniliform with whorl of long setae, ca. 3/4 length of forewing, female antenna filiform with much shorter setae than male, ca. 1/2 length of forewing. Distance between lateral ocelli~3.0 times as long as width of median ocellus. Wings narrowly elongated, with a few or numerous dark spots along longitudinal veins on forewing. Both anterior and posterior branches of Rs forked distad, a crossvein usually present between secondary distal branches of posterior Rs branch; three crossveins present between R and Rs; MA simple; MP with two simple branches; 1A 2-branched, with posterior branch sinuate; 2A 2-branched, with anterior branch sinuate and fused with stem of 1A for a short distance on forewing, and with posterior branch short and straight; base of MA on hindwing long, oblique, simply connecting to Rs. Male genitalia: Tergite 9 subquadrate, thick apodeme present along anterior and posterior margins; sternite 9 broad with arcuately convex posterior margin; gonocoxite 9 present, fused with lateral arms of fused gonocoxites 10; ectoproct stout, sometimes strongly curved ventrad or with additional ventral lobes, inner portion with numerous black rhabdoid-shaped setae; cercus rounded and feebly protruded; gonocoxite 10 strongly sclerotized, lateral arm subtriangular, median plate usually ovoid or blade-shaped with a notch at tip; a pair of membranous sac-like lobes protruding beneath anteroventral portion of fused gonocoxites 10; a setose patch present beneath anus. Female genitalia: gonocoxite 8 broad; gonocoxite 9 nearly rectangular, with or without gonostylus 9; ectoproct subtriangular, with round and feebly prominent cercus.
Larva. Head yellow with dark markings or blackish brown with a few yellow markings. Prothorax dorsally yellow with dark markings; meso-and metathorax dark brown with a few yellow markings. Antenna with distal two segments combined as long as second segment. Labrum nearly rectangular, ca. 2.0 times as wide as long. Terminal pair of abdominal spiracles opening at tips of rather short and coniform projections. See more details in Evans [26].
Biology. The biology of Neohermes is relatively better studied than many other fishfly genera, especially with detailed information on the life history of N. californicus [26]. The larvae of N. californicus and N. filicornis usually inhabit rocky intermittent or permanent warm streams, while the larvae of N. concolor were reported to live under leaf litter near a spring seep [26,27]. The life cycle for N. californicus was estimated to take two years in permanent warm water streams, but it is extended in intermittent streams where the water flows only during the winter months [26]. In drought streams, the larvae burrow into the stream bed and form a small chamber for diapause which may last for more than six months [26].
Remarks. Due to similar wing marking patterns and wing venations, Neohermes is closely related to Protochauliodes, and the females of these two genera are frequently misidentified with each other due to lack of distinct morphological differences [26]. The male antennae are different between these two genera and could be the most readily diagnostic morphological feature. When the female antennae are examined closely, the flagellomere is ovoid, more pilose in Neohermes but subtrapezoid and less pilose in Protochauliodes. Evans [26] noted that the presence/absence of a crossvein beyond the fork of posterior branch of Rs is another good character to distinguish Neohermes and Protochauliodes in most cases. However, it should be noted that this crossvein is absent only in the North and South American species of Protochauliodes, but it is present in the Australian Protochauliodes species. In addition, the female gonostylus 9 is absent or slender digitiform in Neohermes, but it is always robust in Protochauliodes.
Neohermes Head dark brown, with yellowish brown clypeus; vertex with yellowish brown scars. Compound eye blackish brown; ocelli pale yellow, medially margined black. Antenna blackish brown, with distal half of flagellum slightly paler. Mouthparts brown; mandibles with distal half reddish brown, distal three segments of maxillary and labial palpi much darker.
Female. Body length 20-27 mm; forewing length 29-33 mm, hindwing length 25-29 mm. Fused gonocoxites 8 (Fig 2F and 2G) subtrapezoidal in lateral view, posterior margin strongly convexed, medially truncate in ventral view. Gonocoxite 9 ( Fig 2F) broadly ovoid, posteroventrally with a tiny gonostylus 9. Ectoproct ( Fig 2F)   Remarks. The holotype of N. inexpectatus sp. nov. bears an identification label "Protochauliodes minimus (Davis, 1903) det. N. Penny, 1999" given by Norman D. Penny (California Academy of Sciences) who is an eminent expert of New World Megaloptera and Neuroptera. However, the male moniliform antennae with whorl of long setae in this new species suggest its definite affiliation in Neohermes. Possibly, the misidentification was due to the similarly small body size and external appearance of the new species with various Californian Protochauliodes species in the same collection.
Compared with the other western North American fishfly species that are present in the collections with relatively large number of specimens, only one male and two females are known for the new Neohermes species. All the specimens of N. inexpectatus sp. nov. were collected from 1950s to 1970s and no specimen has been collected since then. Based on the collecting data, N. inexpectatus sp. nov. might be restricted to the Northern Coastal Range, and the flight period lasts at least from June to October. 3. Forewing with indistinct dark markings (Fig 1); male ectoproct with apex distinctly curved ventromesad (Fig 2A); male fused gonocoxites 10 with median plate short, rectangular, entirely concealed beneath sternite 9 in ventral view ( Fig 2D); female fused gonocoxites 8 subtrapezoidal in lateral view, with posterior margin strongly convexed (Fig 2G-2E); female gonocoxite 9 posteriorly without subtriangular projection ( Fig 2G);

Phylogenetic analysis
The heuristic search in NONA yielded single most parsimonious tree (MPT) (length = 22, consistency index = 0.90, retention index = 0.88) (Fig 6), and the analysis in TNT also yielded the same phylogenetic tree (length = 22). The three antennal features support the monophyly of Neohermes. Within Neohermes, N. inexpectatus sp. nov. was assigned to be the sister to the remaining five species that form a monophyletic group supported by the prolonged median plate of male gonocoxites 10 with convex lateral margins (chars. 11:1 and 12:1). In this group, the species from western and eastern North America respectively form a monophyletic clade. The synapomorphies supporting N. californicus + N. filicornis are the male ectoproct distally with a ventral projection (char. 8:1) and the loss of female gonostylus 9 (char. 17:1). Whereas, two female characters, i.e. the gonocoxite 9 posteriorly with a small sub-triangular projection (char. 16:1) and the presence of broad membrane between gonocoxites 8 and abdominal segment 9 (char. 18:1), were assigned as the synapomorphies for the clade comprising the three eastern species. N. angusticollis and N. matheri are sister species with two synapomorphic characters: the prolonged sub-triangular median plate of the male gonocoxites 10 (char. 13:1) and the strongly concave posterior margin of female gonocoxites 8 (char. 15:1).

Ancestral area reconstruction
Western North America was reconstructed to be the ancestral area for the initial divergence of Neohermes (Fig 6). Whole North America was assigned to be the ancestral area for the split between two western species (N. californicus and N. filicornis) and remaining three eastern species. Again, western North America was assigned to be the ancestral area for the divergence between N. californicus and N. filicornis, while eastern North America was the ancestral area for the sequential divergence of the three eastern species.

Discussion
Neohermes inexpectatus sp. nov. represents the only new species of Megaloptera discovered in North America in the past 20 years. It is a remarkable supplement to the fauna of the North American endemic fishfly genus Neohermes since the last species of this genus was described 50 years ago. Surprisingly, the known distribution area of N. inexpectatus sp. nov. is just ca. 60 kilometers away from San Francisco; all examined specimens were collected in 1950-1970s and its current status and habitat condition is unknown. More importantly, the new species, as assigned in the present phylogenetic analysis, is sister to the remaining Neohermes species. Therefore, the present finding of a new species is key to understanding the evolution of this enigmatic genus. In the phylogeny of world fishfly genera reconstructed by Liu et al. [29], Neohermes belongs to a monophyletic group, which also includes Protochauliodes (distributed in eastern Australia, Chile, and western North America), Taeniochauliodes Esben-Petersen (endemic to South Africa), and Nothochauliodes Flint (endemic to Chile). This group is supported by the partial coalescence between the stem of 1A and anterior branch of 2A in the forewing. Based on morphology, Neohermes is considered to be the sister group to Protochauliodes, whose status however is still questionable [29]. Penny predicted that the ancestral Protochauliodes might have originated in Gondwana and dispersed northward to western North America [30]. However, both present and previous ancestral area reconstructions [29] involving this genus suggest that either western North America or a broader area comprising western North America and austral regions (e.g. eastern Australia and South America) might be the range of the ancestor of Neohermes + Protochauliodes. On one hand, if these two genera diverged in western North America, the aforementioned northward movement of ancestral Protochauliodes will be falsified. Thus, the occurrence of some Protochauliodes species from austral regions implies that the divergence between Neohermes and Protochauliodes might have taken place before the breakup of Laurasia and Gondwana during the Early-Middle Jurassic. On the other hand, if Neohermes and Protochauliodes were separated in a broad area covering western North America and some austral landmasses, this divergence event is likely to be correlated to the splitting between Laurasia and Gondwana. In this case, the western North American species of Protochauliodes might have diversified after the northward dispersal of their common ancestor from Gondwana. The most significant event of faunal exchange between South and North America (post separation ca. 175 MYA) was facilitated by the formation of the Central American land bridge approximately 3.5 MYA [31]. However, compared with the Mesozoic origin of these two genera, such Pliocene dispersal of Protochauliodes needs further consideration.
Neohermes is the only fishfly genus distributed in both eastern and western North America. Based on our ancestral area reconstruction, the initial divergence within Neohermes was estimated to be confined to western North America, and a dispersal event was recognized from western North America to western plus eastern North America, which was the distribution area for the common ancestor of the five Neohermes species except N. inexpectatus sp. nov. The current reconstructed geological history of North America from the Mesozoic to the present shows that the Mid-Continental Seaway separated western and eastern North America in the Mid-Late Cretaceous (100-80 MYA) and finally disappeared at the end of the Cretaceous (70 MYA). This was presumably followed by an Early Cenozoic period of floral and faunal exchange between the eastern and western Nearctic [32]. A compatible pattern of the above biogeographic and the geological scenario could be that the initial divergence within Neohermes, perhaps also the origin of this genus, took place when western North America was isolated from eastern North America during the Mid-Late Cretaceous. Subsequently, the closure of the Mid-Continental Seaway at the end of the Cretaceous facilitated the eastward dispersal of Neohermes. However, as discussed above, if Neohermes and Protochauliodes diverged around or even before the splitting of Laurasia and Gondwana, one cannot deny the past distribution of ancestral Neohermes in eastern North America preceding the divergence of modern species. Likewise, the modern eastern species might be re-introduced by dispersal after assumed extinction of some ancestral species in eastern North America. Hence, dating the origin of Neohermes is crucial to clarify the early speciation process of this genus, which however has to be done in future molecular-clock model-based studies.
The separation between the eastern and western modern species of Neohermes appears to be due to the uplift of the Cordilleran Mountain System, which was presumably associated with massive vicariance in Nearctic biotas [32]. Nevertheless, there had been two major orogenic phases of the Cordilleran System, respectively occurring in the Late Eocene (35 MYA) and the Late Oligocene (25 MYA). Between these two orogenic events, the early Cordilleran System had been completely wiped out during the Early Oligocene (30 MYA) [32]. Without a molecular divergence time estimation, we can only predict that the eastern and western extant species of Neohermes diverged no later than the second formation of the Cordilleran System.
Considering the biogeography of the two western species, i.e. N. californicus and N. filicornis, their distributional range is respectively along the Sierra Nevada and the Pacific Coast Ranges, suggesting possible allopatric speciation correlated with the isolation of these two close but independent mountain chains. Inferring the biogeographic scenario associated with the diversification of the three eastern species, however, is more difficult because no clear geographic barrier has yet been found among the distribution areas of these species. The only notable issue is that the separation of N. matheri from Mississippi and N. angusticollis from Georgia, South Carolina and North Carolina might be attributed to the faunal discontinuities repeatedly uncovered between Atlantic and Gulf Plains [33], with a drainage system, which includes the Apalachicola River Basin and the Flint and Chattahoochee Rivers, as the oftenrecovered boundary for the vicariance [34].

Conclusions
The present new fishfly species from North America is a remarkable finding considering that the taxonomy of Nearctic Megaloptera has been extensively studied previously. Our phylogenetic study demonstrates that N. inexpectatus sp. nov. is the sister to other Neohermes species and the biogeographic analysis suggests a Mesozoic origin of the genus. It would be of great interest to date the origin and interspecific diversification of such a relictual aquatic insect group using a molecular approach so as to better understand the evolutionary history of this genus, as well as the Nearctic Megaloptera. In addition, populations of N. inexpectatus sp. nov. could be endangered or even close to extinction owing to the scarcity of collecting records, especially recently. It is therefore necessary to do intensive surveys around the known distributional area of N. inexpectatus sp. nov. in order to determine its current conservation status.

Ethics statement
All insect specimens for present study were acquired from authorized museum collections but no newly collected in the field. No species in the genus Neohermes are included on the ''List of Protected Animals in U.S.A.".

Material examined
The type specimens Genitalia preparations were made by clearing the apex of the abdomen in cold, saturated KOH for 8-10 h. After rinsing the KOH with acetic acid and water, the apex of the abdomen was transferred to glycerin for further dissection and examination. Following examination genitalia were stored in fresh glycerin in a microvial pinned below the specimen. The terminology of the genitalia follows that of Liu et al. [35].
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub: 8E74FFA1-D10B-4D8B-ACE3-AC59F08E7097. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Phylogenetic analysis
Eighteen morphological characters listed below were coded for six ingroup taxa and two outgroup taxa, with 16 binary and two multistate (S2 File). (1) simple. The basically bifurcated MA of the forewing is considered to be plesiomorphic [29]. State 1 herein is present in Protochauliodes and all Neohermes species.
Dysmicohermes Munroe and Protochauliodes Van der Weele were selected as outgroups because Dysmicohermes belongs to the basal-most fishfly lineage and Protochauliodes is considered to be the sister group of Neohermes [29]. The analysis was performed using NONA ver. 2.0 [38] with a heuristic search. The heuristic search was used with maximum trees to keep setting to 10000 and number of replication setting was 100. We also analyzed the datasets using TNT ver. 1.1 [39] with an initial New Technology search set to 100 (using a driven search with sectorial search, ratchet, drift, and tree fusing; finding the minimum tree 10 times). The branch support values were calculated with the function implemented in TNT (TBR from existing trees, retain trees suboptimal by 10 steps). All characters were treated as unordered and with equal weight. Character states were mapped on a most parsimonious tree (MPT) using Win-Clada ver. 1.00.08 [40], showing only unambiguous changes.

Ancestral area reconstruction
As the ingroup species are disjunctively distributed in eastern and western North America, we defined these two areas (i.e. western North America (a) and eastern North America (b)) as the areas of endemicity used for the ancestral area reconstruction. In addition, southern South America (c) and eastern Australia (d) were also defined as the endemic areas because the outgroup Protochauliodes also occurs in these two areas besides western North America. Ancestral areas at internal nodes were inferred using a dispersal-vicariance (DIVA) optimization model [41] implemented in the program RASP2.0 Beta [42]. The DIVA optimization was conducted with default settings with the maximum number of areas in ancestral ranges constrained to three.