Cenozoic origins of the genus Calliarcys (Insecta, Ephemeroptera) revealed by Micro-CT, with DNA barcode gap analysis of Leptophlebiinae and Habrophlebiinae

Mayflies (Ephemeroptera) are among the oldest pterygote insects, with the earliest fossils dating back to the Late Carboniferous. Within mayflies, Leptophlebiidae are a highly diverse and widespread group, with approximately 140 genera and 640 species. Whereas taxonomy, systematics, and phylogeny of extant Leptophlebiidae are in the focus of extensive studies, little is known about leptophlebiid fossil taxa. Because fossil remains of Ephemeroptera in sedimentary rocks are relatively rare, inclusions of mayflies in amber are a unique source of information on their evolution and diversity in the past. Leptophlebiidae found in Cenozoic resins mostly belong to the subfamilies Leptophlebiinae (in Eocene Baltic amber) and Atalophlebiinae (in Miocene Dominican and Mexican ambers). In the present contribution, we confirm the first finding of the genus Calliarcys from Eocene Baltic amber by using Micro-CT, which allowed confirming its generic placement by visualizing diagnostic key characters otherwise hidden by a cloud of turbidity. Additionally, we present first molecular data on the extant species Calliarcys humilis Eaton, 1881 from the Iberian Peninsula and the barcode gap analysis for Leptophlebiinae and Habrophlebiinae.

www.nature.com/scientificreports/ the fused styliger plate as additional apomorphic character shared by the Atalophlebiinae lineage (including the Terpides lineage + Castanophlebia). Kluge 7 in a revised phylogenetic classification of Atalophlebiinae s.l. split off three new subfamilies: Terpidinae and Castanophlebiinae were established for 19 species previously assigned to Atalophlebiinae. For Calliarcys, the monotypic subfamily Calliarcyinae was proposed 7 .
Based on a set of 20 morphological characters of nymphs and adults, Leptophlebiinae were confirmed as sister group to the remaining Leptophlebiidae in the first strict cladistic analysis. Calliarcyinae were revealed as sister group of the clade Habrophlebiinae + Atalophlebiinae s.l. (= (Atalophlebiinae s.str. + Castanophlebiinae) + Terpidinae) 2 . However, this concept proposed by Kluge 7 and later corroborated by Godunko et al. 2 remained controversial: Bauernfeind and Soldán 8 rather followed Kluge 4 , placing Calliarcys provisionally within Leptophlebiinae, at the same time pointing to some characters of Calliarcys common with Habrophlebiinae.
All these higher phylogenies of Leptophlebiidae discussed above were solely based on morphological evidence. In a first molecular analysis using two nuclear markers, O'Donnell and Jockusch 9 did not even recover the monophyly of Leptophlebiidae (see also 2 ). Atalophlebiinae were also not recovered as monophyletic, only each Leptophlebiinae and Habrophlebiinae were reported as monophyletic 9 . Contrary to these findings, the monophyly of Leptophlebiidae was later confirmed by Ogden et al. 10 based on a large, combined set of morphological and molecular characters. Analysing 153 leptophlebiid taxa within 53 genera using the two molecular markers COI + 28S, Monjardim et al. 11 yet proposed a new higher phylogeny of this family. Basically, there were three monophyletic branches recovered: Atalophlebiinae s.l. was confirmed, though with uncertain position of Castanophlebiinae, and the third branch was represented by Habrophlebiinae + Leptophlebiinae. Unfortunately, Calliarcys was not included in this molecular study. More recently, Gatti et al. 12 also used a molecular dataset to investigate historical events linked to the vicariance and dispersal of Atalophlebiinae in the second phase of the Gondwanan breakup. Fossils were used to calibrate the time of origin of phylogenetic nodes, but until now no fossil representatives of Calliarcys have been available.
In this contribution, we aim to broaden the knowledge of this enigmatic genus, describing the first fossil representative Calliarcys antiquus sp. nov., from the Eocene Baltic Amber. Its description is also the first establishment of a mayfly species by Micro-CT investigation. This method has by now frequently proven its potential to address the problems linked to taxonomy and systematics of fossils embedded in Mesozoic and Cenozoic resins (see, e.g. [13][14][15] ). Its application in mayfly systematics however has been limited so far, with the visualisation of genitalia in the redescription of two extant mayfly species 16 and the report of a potential case of phoresis in a fossil mayfly 17 . With the Micro-CT investigation of Calliarcys antiquus sp. nov., we intend to reconstruct and describe its genitalia to confirm its generic placement, and to specify diagnostic generic characters based on both extant and extinct species. In addition, we provide the first DNA barcode data for the extant Calliarcys humilis as well as provide a gap analysis of the DNA barcode data for all the extant Leptophlebiinae and Habrophlebiinae.

Systematic Paleontology
Subphylum Revised diagnosis of adults (modified from 2,8 ). (i) Two pairs of connected intercalary veins of different length in cubital field of forewing; (ii) costal process of hind wing situated nearly at half length, well developed, slightly asymmetrical or symmetrical, bluntly pointed or rounded apically; (iii) penis lobes straight, simple, and tubular; (iv) tip of penis lobes slightly bent inwards, without blade-like process apically; (v) posterior margin of ventral forceps base deeply concave medially, with V-shaped medial incision and two long, submedian projections rounded apically and directed caudally; (vi) segment I of forceps longest, at least 3 × longer than segment II, without inner process or appendages; (vii) terminal and subterminal segments of forceps shorter than first segment; terminal segment shortest.
Calliarcys antiquus sp. nov. Godunko Derivation of name. The genus name "Calliarcys" is a compound noun of masculine gender derived from the Greek κάλλη (kalli, for "beauty") and ἄρκυς (arkys, latinized arcys, for net). The species epithet "antiquus" (masculine adjective, Latin for "ancient") refers to the ancient origin of the fossil preserved in Eocene resin.  Fig. 1a-c). Lateral sides of thorax and tip of abdomen covered by a cloud of turbidity (so-called "Verlumung", see Fig. 1d). Piece of amber with numerous cracks, organic debris, and multiple resin layers, thus details of male genitalia are hardly visible from dorsal and ventral sides in optical view ( Fig. 1f-g). Additionally, the ventral side of styliger plate is partly covered by the preserved cercus and medial terminal filament, so shape of styliger plate and details of penis lobes are hardly distinguishable. Hind wings present, but details of venation are poorly discernible. All these hidden details in regular view however could be made visible by Micro-CT reconstruction (see below). www.nature.com/scientificreports/ Colours. Preserved colour of specimen is yellowish to brown, with blackish-brown spots laterally on thorax and abdomen. Eyes pale, dirty yellow to light brown. Facial keel intensely brown, darker than eyes. Mesonotum darkest, dark brown, with blackish maculae laterally; wing surface covered by artificial, small brownish and blackish maculae [as result of fossilisation]; narrow brownish strip along outer margins of forewings. Legs unicoloured brown. Abdomen with translucent terga III-VII; abdominal sterna and three last abdominal segments intensively brown; traces of dark brown maculation along lateral margins of abdominal segments (Fig. 1b-c).
Head (Figs. 1d,e, 3a,b and 4a-c). General colour yellowish-brown to dark brown. Relatively small, brown facial keel. Antennae brown. Ocelli indistinct, relatively small, without conspicuous coloration. Upper portion of compound eyes well developed, large, widely rounded, contiguous medially; facets of compound eyes hexagonal. Border between upper and lower portions of compound eyes well distinguishable.
Wings  www.nature.com/scientificreports/ well-developed. iRS well-developed, connected to RSp by 9-10 cross veins, slightly approximated to RSa1. Asymmetric MA fork, forked after 0.44 to 0.46 of its length; MA1 and MA2 connected to iMA by 5 to 7 cross veins. MP slightly asymmetrical, forked after 0.18 of its length; MP1 and MP2 are basally connected by a single cross vein; iMP relatively long, connected to MP1 and MP2 by 4-6 cross veins on each side. Cubital field with four intercalaries connected by several cross veins; CuA and CuP connected by a single cross vein proximally; iCu2 longest, connected with CuA distally; Cu-A angle smoothly curved; CuP approaching A1; no cross veins in anal field (Fig. 2a).
Hind wings hyaline, translucent, relatively wide, 3.74-3.81 × shorter than forewings, with width/length ratio 0.54. Longitudinal venation is light brown; cross venation is well developed and relatively numerous, www.nature.com/scientificreports/ whitish yellow to yellow, and is poorly recognizable. Costal process of hind wings widely rounded, semilunar and strongly symmetric, located in middle of wing length, slightly rising above inner margin ( Fig. 2a-b). Numerous cross veins between C and RA; RS fork well developed, slightly asymmetrical, iRS connected to Rsa and RSp by 7 cross veins; MA and MP not forked, 8 cross veins between MA and MP; cubital triad well-developed; veins of cubital field not touched; AA not forked ( Fig. 2a-b).
Legs. Femora and tibiae brown, tarsi paler, yellowish brown to yellow distally.  www.nature.com/scientificreports/ ganglia ( Fig. 1b-c). Preserved right cercus brown proximally, yellow to light yellow distally; terminal filament completely preserved, slightly longer and darker than cercus. Genitalia (Figs. 1f-g, 2c, 3 and 5) only visible laterally, ventral side of body is covered by cercus and terminal filament. Due to the position of the specimen in the resin as well as cracks and streaks in the resin, styliger plate, base of forceps, and penis lobes are hardly visible in optical view (Fig. 1a,b,f-g). Styliger plate brown, deeply incised mediocaudally by a V-shaped incision; two median, nearly finger-shaped projections, markedly protruding above the anterior margin of the styliger; projections relatively large, rounded apically ( Fig. 2c and 5d-e). Forceps base dark brown to brown; distal segments of forceps slightly paler. Forceps 3-segmented; ratio of forceps segments: 1.00: 0.21: 0.18; segment I strongly elongated, with distinct rounded hump basally, without prominent appendages; segment II relatively short with convex inner margin, slightly longer than segment III; distal segment of forceps (i.e. segment III) nearly oval, rounded apically (Figs. 1f-g, 2c and 5). Simple, straight and tubular, slightly club-shaped, penis lobes separated basally and touching apically; tip of penis only slightly bent inward, not obliquely truncated, without appendices or processes (Figs. 1f-g, 2c and 5).

Female imago, male and female subimago and nymph. Unknown.
DNA barcode data for the extant Calliarcys humilis and gap analysis for Leptophlebiinae and Habrophlebiinae. DNA barcoding proved that seven specimens of C. humilis belong to one BIN (BOLD:AEG9529), with two other specimens from Spain previously deposited in BOLD as private data. We identified six haplotypes within our sequences and the maximum K2p distance between the specimens did not exceed 1.86%. The sequences of C. humilis cluster among representatives of the Leptophlebiinae subfamily.
We analysed 104 species from the subfamilies Leptophlebiinae and Habrophlebiinae, for which we found only 39 species (37.5%) with public barcodes and BINs assigned. Number of sequences per species varied from one (e.g. Paraleptophlebia gregalis, Neoleptophlebia vaciva, Habrophlebiodes zijinensis, Habroleptoides modesta, Habroleptoides pauliana) to the maximum of 147 for Neoleptophlebia mollis with the mean value of 8.4 barcodes per species. We observed a maximum intraspecific distance higher than 3% in case of 19 species. The highest distance was observed among undetermined individuals of Leptophlebia, Paraleptophlebia, Habroleptoides (24.3%, 28.47%, and 28.25% respectively). A maximum intraspecific distance lower than 3% was observed in case of 13 species (Tables 3 and S1).

Discussion
Taxonomy. We attribute C. antiquus sp. nov. to Leptophlebiidae based on the characteristic appearance of its mesonotal suture, which is transverse centrally, distinctly curved posteriorly, and stretching backward, and based on the furcasternal protuberance, which is clearly separated along its entire length. In addition to these www.nature.com/scientificreports/ characteristics, the presence of an asymmetric MA in the forewing in combination with a strongly curved CuP and distinct cubital intercalaries also indicate a systematic position of this extinct species within Leptophlebiidae (Figs. 1c, 2a, 3; Table 2).
The placement of C. antiquus sp. nov. within the genus Calliarcys can be verified by the presence of four free intercalaries in the cubital field of the forewing and the well-developed costal process at half length of the hind wing (Fig. 2b). Furthermore, the generic attribution is clearly justified by the characteristic shapes of styliger plate and genitalia as revealed by tomography: (i) simple and tubular penis lobes, deprived of a blade-like process at the tip; (ii) posterior margin of the styliger plate with V-shaped medial incision, and two long submedian projections; (iii) segment I without inner process or appendages, and (iv) forceps segments II and III shorter than the first segment, terminal segment shortest (Figs. 3c, 5; Table 2).
The body colour is not a reliable character to separate the fossil and both extant species: Adults of C. humilis and C. van are both relatively dark to black, with a darker thorax, C. antiquus sp. nov. is generally light brown to brown, without any conspecific pattern on its abdominal terga. The colouration of the compound eyes can also not be used for species separation, but the shape of the upper eye portion differs in the males. In contrast to C. humilis and C. van with well-separated compound eyes (Figs. 6a, 7a-b, 8 and 9a,b), the upper portion in C. www.nature.com/scientificreports/ antiquus sp. nov. is clearly contiguous dorsally and also larger (Figs. 1e, 3a and 4b). In C. humilis, the upper portion is smallest, well separated, and the distance between the eyes is nearly as long as the width of the compound eye (Figs. 8 and 9a-b; for C. van see also Figs. 1, 4 and 5 in 2 ). The shape of the wing venation of all representatives of Calliarcys is rather similar, but in C. antiquus sp. nov., the cross venation of both wings is well-developed ( Fig. 2a-b). Its male imago has predominantly forked veins in the pterostigmatic area of the forewing and the longest vein iCu2 (Fig. 2a), in contrast to simple pterostigmatic veins of C. humilis and C. van with the longest third vein [iCu3] in the cubital field (Figs. 10a, and 11a). While pterostigma and area between Sc and RA of the forewing are without any colouration, but translucent and hyaline in C. antiquus sp. nov. (Fig. 1b-c), this region is frosted and milky to yellowish white in both extant species The distal part of the hind wings in C. humilis and C. van seems to be narrower and elongated (Fig. 12), in contrast to a widely rounded distal half of the hind wings in C. antiquus sp. nov. (Fig. 2a-b). In C. humilis, the costal process is asymmetrical, nearly acute and step-like, situated in the proximal half of wing (Figs. 11b, Fig. 13a; for other drawings of C. humilis hind wings see Fig. 18 in 18 and Fig. 18 in 19 ), while it is nearly symmetric and rather at half length in C. antiquus sp. nov. and C. van 2 . Finally, the cubital venation in the hind wing is welldeveloped in C. antiquus sp. nov., but diminished in the extant representatives.
Peters and Edmunds 19 keyed out two generic adult characters of the genus Calliarcys, in particular, the presence of two hooked pretarsal claws of each pair of legs, with an 'opposing hook' . In fact, all representatives of this genus possess dissimilar pretarsal claws, with one hooked and one triangular, somewhat rounded apically claw. Additionally, Nikita J. Kluge (see http:// www. insec ta. bio. spbu. ru/z/ Eph-phyl/L_ Calli arcys. htm) reported in both extant species the presence of a narrowed costal field in the apical half of the hind wing; we may add that in C. antiquus sp. nov. the costal field is also strongly narrowed (Figs. 2a-b, 10b, 11b and 12).
The male genitalia of the three species differ considerably from each other. While the penis lobes of C. humilis are elongated, relatively slender, and markedly stretched inward at their tip, the lobes of C. van and C. antiquus sp. nov. are relatively robust and shorter, with the tip only slightly bent inward (Figs. 2c, 5, 7c-d and 9c,d; for comparison see also Fig. 76 in 18 , Fig. 76 in 19 , Fig. 313 in 8 , and Figs. 6, 7 and 12-14 in 2 ).
The submedian projections of the styliger plate are of similar shape in C. antiquus sp. nov. and C. humilis, namely elongated, slender, and finger-like, in contrast to broad, nearly triangular projections in C. van. An Table 2. The summary of morphological characters of the male imagines to distinguish extant and extinct representatives of the genus Calliarcys. Distinct differential characters are marked in bold.

Characters (male imago)
Calliarcys antiquus sp. nov.  Calliarcys antiquus sp. nov. clearly differs from all described Eocene taxa of Leptophlebiinae listed in the Supplementary material S5 (Supplementary Information 5) by the shape of penis lobes, the proportions and shape of forceps segments, the structure of cubital venation of forewings, and the presence of the more prominent costal process of the hind wings (see Figs. S1 and S2; for the list of extinct Cenozoic Leptophlebiinae see Table S4). DNA barcoding. DNA barcoding is a well-established and powerful tool for gaining information on the taxonomic status of various organisms and conducting assignment for problematic specimens, immature life stages, and tissue samples [22][23][24] . Biomonitoring involving eDNA, DNA barcoding and metabarcoding starts to be implemented as a standard procedure 25,26 . Effectiveness of such methods depends on the reliability of barcode reference libraries for particular taxa, which are still far from complete 27,28 . Among freshwater macroinvertebrates, the Ephemeroptera, Plecoptera, and Trichoptera (EPT) faunas are most important in biomonitoring [29][30][31] . Thus our addition of properly identified specimens of Calliarcys humilis as one of two living representants of this genus is valuable. Interestingly, according to COI, C. humilis groups among various taxa within Leptophlebiinae, providing a first hint for the taxonomic position of the genus (see Supplementary material S4 in Supplementary  Information 2). Additionally, our gap analysis showed a surprisingly high underrepresentation (39 from 104 species) of barcoded species within the subfamilies Leptophlebiinae and Habrophlebiinae (Table 3). It must be stressed that those 19 species, where the maximum K-2p intraspecific distance is higher than 3%, may represent either cases of previously undetected cryptic diversity or point to the misidentification of some individuals. The latter is most probable for at least 10 species, where we noticed BINs shared between al least two species. Another, less likely option would be that one or more species within one BIN is not a real species, but e.g. just a morph or a hybrid. It is worth to stress that such misidentifications in open databases like GenBank and BOLD still persist, even if there are attempts for better data curation, and for removing obvious mistakes [32][33][34] . Given the number of EPT taxa and their importance for ecosystem health assessments, an inaccurate taxonomic assignment of DNA barcodes can seriously hamper the reliability of biomonitoring of water ecosystems based on molecular data. Thus pointing out such mistakes is crucial to improve the situation.

Materials and Methods
Material. The holotype of fossil Calliarcys antiquus sp. nov. described in this study is housed in the collection of the State Museum of Natural History, Stuttgart (SMNS) under the inventory number BB 2515 (holotype; male imago). The piece of resin originates from an unknown Eocene deposit of Baltic amber.
The investigated specimens of Calliarcys van belong to the type series (holotype and paratypes of male imagines) collected in Turkey in 2011 by Tomáš Soldán and Jindřiška Bojková (see also Godunko et al. 2     www.nature.com/scientificreports/ Morphological and morphometric studies. Some paratypes of C. van and some of the imaginal material of C. humilis were mounted on slides with Liquide de Faure (soluble in water). The material of extant specimens was observed using Olympus SZX7 stereo microscope and Olympus BX41 microscope for microslides. Observation and drawings of C. antiquus sp. nov. were made by using Micro-CT-based reconstruction and by a camera lucida on a Leica M205 C stereo microscope. Multiple photographs with different focal depths were taken with a Leica DFC450 Digital Camera through a Leica Z16 101 APO Macroscope using Leica Application Suite v. 3.1.8. The photo stacks were processed with Helicon Focus Pro 6.4.1 to obtain combined photographs with extended depth of field and subsequently enhanced with Adobe Photoshop Classic.
The measurements of individual body parts were taken either by using an ocular grid or inferred from the photographs taken with a calibration scale. The measurements of C. antiquus sp. nov. are given in Table 1. Identical morphometric parameters were used also for other Mesozoic and Cenozoic mayfly fossils (see e.g. 35,36 ).
Morphological terminology and nomenclature of wing veins used throughout the text follow Kluge 7,37 and Bauernfeind and Soldán 8 ; thoracic morphology is analysed using following contributions 37-40 . Micro-CT scans and image reconstruction. The piece of amber containing the fossil was fixed to the sample holder with plasticine and scanned using a Bruker SkyScan 1172 microtomograph (Bruker-micro CT, Kontich, Belgium) with a Hamamatsu 80/100 X-ray source and a VDS 1.3Mp camera. The setting parameters were as follows: voltage = 69 kV; current = 89 µA; isotropic voxel size = 4.05 µm; image rotation step = 0.2°; 360° of rotation scan without filter. This resulted in a scan duration of 5 h:04 min:04 s and 1802 2D shadow projections (X-ray images).   www.nature.com/scientificreports/ explanation of the procedure was previously published 41 . However, the reconstruction revealed that no internal structures were visible and only a thin cuticular layer is preserved. The latter showed an X-ray transparency similar to that of the surrounding amber matrix. This prevented an appropriate volumetric visualisation of the insect. Thus, we followed a procedure (for details see Supplementary Information S1). Amira 6.7.0 (Thermo Fisher Scientific, Waltham, MA) 42 DNA was extracted from one or two legs of each species used, depending on the size of the samples, using Genomic Mini Kit (A&A Biotechnology, Gdansk, Poland) according to the manufacturer's protocol. COI fragments were amplified using LCO1490-JJ/HCO2198-JJ primers 44 , using reaction conditions after Hou et al. 45 .
The success of the reaction was verified using a standard agarose gel. PCR products were purified following the procedure described by Rewicz et al. 46 . Bidirectional Sanger sequencing was outsourced by Macrogen Europe (Amsterdam, The Netherlands). The identity of the obtained sequences was verified with BLAST 47 . Doublestranded sequences were checked, aligned, and trimmed to the standard length 658 bp using Geneious 10.2.6 software package 48 . We also checked for the absence of frameshifts, double peaks, and stop codons using the Geneious 10.2.6 software package 48 . Sequences were deposited in GenBank (https:// www. ncbi. nlm. nih. gov/ genba nk) under accession numbers OM158449-OM158451 and OM158454-OM158457. Relevant voucher information, photos, taxonomic classification, and DNA barcode sequences are also publicly accessible through the dataset DS-RGCAL (https:// doi. org/ 10. 5883/ DS-RGCAL) in BOLD (www. bolds ystems. org) 49 . Obtaining BINs (Barcode Index Numbers) for sequences deposited in BOLD provided additional verification of species identification, as BINs can be treated as tentative equivalents of species 50 . To conduct gap analysis and the general molecular diversity, we searched BOLD public repository for members of subfamilies Leptophlebiinae and Habrophlebiinae. Additionally, the first author prepared a checklist of 104 described species from those subfamilies, and we searched the BOLD repository again (https:// v4. bolds ystems. org under accession https:// doi. org/ 10. 5883/ DS-RGCAL) (see Supplementary Information 3 and 4; Tables S1 and S2). The obtained unchecked database was curated, and we discarded all specimens either without COI sequence, with stop codons, contaminations or other flags added. We also discarded specimens with COI sequence shorter than 500 bp and without BIN assigned. After such curation, the final dataset contained 941 sequences (see details in: Tables S1 and S2; data set DS-RGCAL). The intraspecific mean and maximum genetic distances were calculated based on the Kimura 2-parameter model (K2P; 51 ), using the analytical tools of the BOLD workbench (Barcode Gap Analysis, Distance Summary) and MUSCLE alignment method 52 . The Barcode Gap analysis was performed for individuals identified to species level, while the Distance Summary Tool was applied to specimens determined only to the genus level. We did not evaluate if the identification of the specimens to the species level was correct. Only in the case of unidentified specimens with BIN, which was matching to BIN already assigned to identified species, we added them to identified species. To illustrate the molecular diversity of the analysed taxa, we prepared BOLD TaxonID Tree for COI sequences from our dataset DS-RGCAL (see Supplementary Information 2).