A cryptic new species of Chlidonoptera Karsch, 1892 from the south west protected zone of the Central African Republic (Insecta, Mantodea, Hymenopodidae)

Abstract Between 1998 and 2012, several scientific expeditions in Dzanga-Sangha Special Reserve and Dzanga-Ndoki National Park led to the collection of many Mantodea specimens from Central African Republic (CAR). Among these specimens, several males of an undescribed species were discovered. Morphologically, this species most closely resembles to Chlidonoptera vexillum Karsch, 1892 and Chlidonoptera lestoni Roy, 1975. A new lineage was revealed by DNA barcoding. Therefore, a new species is described, Chlidonoptera roxanaesp. nov. Habitus images, genitalia illustrations and descriptions, measurement data, a key to species, natural history information, and locality data are provided. These results add to the evidence that cryptic species can be found in tropical regions, a critical issue in efforts to document global species richness. They also illustrate the value of DNA barcoding, especially when coupled with traditional taxonomic tools, in disclosing hidden diversity.


Introduction
Since the beginning of the 1980s, the entomologist Philippe Annoyer had been traveling in southwestern CAR searching for butterflies and other insects. In 2008, his missions expanded and came to be called Epiphyte 2008. In 2010, a massive survey program was organised under the name SANGHA2012 Biodiversité en Terre Pygmée. On this occasion, the author joined the team to increase the study and collections of Mantodea (Moulin et al. 2017, Moulin 2018b. Several males of Chlidonoptera Karsch, 1892 were collected, mostly by light trapping. Visual searching and beating of vegetation, both on the ground and canopy, did not lead to the discovery of the associated female. The localities of these specimens are in the last remnants of primary forests of the southwestern tip of the CAR. All species belonging to the genus Chlidonoptera are morphologically similar to each other but easily discriminated from other genera. The main morphological feature of the genus is a relatively large yellow spot on the elytra located between the two black arcs of the circle. The collected male Chlidonoptera specimens were initially presumed to be C. vexillum Karsch, 1892, as they share many morphological similarities. Additional examinations of C. vexillum male genitalia compared to the recently collected Chlidonoptera genitalia led to the submission of DNA sequencing samples at Canadian Center for DNA Barcoding (CCDB) in Guelph. Many studies have used the 5' region of the cytochrome oxidase I gene (COI), more commonly referred to as the DNA barcode region, as a useful tool to discriminate various groups of insects (Cocuzza et al. 2015). The results from the DNA sequencing revealed that the specimens from southwestern CAR are different from known C. vexillum Karsch, 1892 specimens from Cameroon and Gabon. Originally described by Karsch in 1892 (Bomistria lunata Saussure, 1898 synonym) to contain a single species C. vexillum, two species were added: C. chopardi Roy, 1964 andC. lestoni Roy, 1975. During this time, Roy synonymised the East African Anabomistria werneri Giglio-Tos, 1915(Roy 1964 with Chlidonoptera, which was later confirmed by Lombardo (1997). Thus, prior to the discovery of this new species, described herein, the genus Chlidonoptera contained four species: C. vexillum, C. chopardi, C. lestoni, and C. werneri. Chlidonoptera chopardi is distributed in West Africa, C. vexillum and the new species are distributed in West Central Africa, C. lestoni is distributed in Ghana (Leston 1968, Roy et Leston 1975, with C. werneri distributed in the East. It appears that Tanzania and Kenya are the eastern limits of the distribution of C. vexillum (Ehrmann 2002, Schwarz andRoy 2019). Chlidonoptera vexillum is sympatric with C. werneri, creating confusion. Wrongly, Kirby (1904) cites Bomistria lunata Saussure, 1898, as a distinct species of C. vexillum. Chlidonoptera is classified within the tribe Hymenopodini, subtribe Pseudocreobotrina with four other genera (Mantodea Species File, http:// mantodea.speciesfile.org;Svenson et al. 2016, Schwarz andRoy 2019).
Ideally the description of a species should result from a synthesis of information that encompasses morphological, molecular, biological, biogeographical, physiological, ecological and bibliographical data; however, this compendium of information is lacking for the great majority of species.

Sampled region
The study area includes the UNESCO World Heritage site Sangha Trinational, the Dzanga-Sangha Special Reserve (6,865.54 sq km) and the Dzanga-Ndoki National Park (1,143.26 sq km) (Moulin et al. 2017). These national parks and reserves aim to protect the second largest rain forest on the earth. Altitude ranges from 300 to 620 meters above sea level. The whole zone is on alluvial sands. Along streams, forest clearings are present with marshy depressions. There are three types of forest within the study area: mainly dryland forest, a semi-evergreen forest that contains swamp-forest areas along the rivers, and a closed-canopy, monodominant Gilbertiodendron dewevrei forest. The dryland forest is an open, mixed canopy that is dominated by Sterculariaceae and Ulmaceae; often associated with it is a dense understory of Marantaceae and Zingiberaceae. Along the Sangha river, there are stands of Guibourtia demeusei (Vande Weghe 2004, http://www.dzanga-sangha.org/).

Collection and preparation
Collection was predominately made by light trapping with 250-Watt bulbs. A few individuals were found at or around the lamp; or on the tents of the camp, attracted by the diffuse light of the incandescent bulbs. The specimens were placed in cyanide vials and then kept dry on layers of cotton and blotting paper. Some specimens were kept alive in cubital screen enclosures to capture live images. Some males were pinned after genitalia preparation was made and a leg was preserved in ethanol for DNA barcoding with tissue samples deposited in CCDB in Guelph.

DNA barcoding
DNA barcoding, the analysis of a standardised segment of the mitochondrial cytochrome c oxidase subunit I (COI) gene, was performed on a representative selection of specimens (n = 25). Tissues were sent to CCDB at the University of Guelph for DNA extraction, polymerase chain reaction (PCR), and sequencing. DNA was extracted from dry legs using a routine silica-based 96-well extraction automation protocol (Ivanova et al. 2006). The 658bp region of COI proposed for use as a 'DNA barcode' (Hebert et al. 2003) was amplified with the PCR primers C_LepFolF/C_LepFolR (Hebert et al. 2004). Data are currently managed under the following projects: "Mantodea of Gabon -Project 1 [ ECO-TROP 2014]," "Mantodea of Gabon -Project 2 [ECOTROP 2011]," "DNA Barcoding Mantodea -Collection N. Moulin" Barcode of Life Data Systems (BOLD, Biodiversity Institute of Ontario, Canada; http://www.boldsystems.org). Kimura-2-parameter (K2P) distances were calculated using the BOLD 4.0 interface (Ratnasingham and Hebert 2007). Sequences were then analysed and trees constructed using the BOLD 4.0 interface.

Deposition of the specimens
Specimens are, currently, in the Research Collection of Nicolas Moulin (Montérolier, France) and Philippe Annoyer Personal Collection (Sainte-Croix-Volvestre, France). Types will be deposited at the MNHN (Paris, France).
Abbreviations used in this paper:

Descriptive conventions and character systems
The species treatment within this study provides a brief diagnosis and criteria descriptions stemming from the anterior surface of the head, the dorsal surface of the pronotum, the legs, the wings, and the abdomen. Foreleg spine nomenclature follows Wieland (2008Wieland ( , 2013 and morphological terminology, including genitalia, follows that of Brannoch et al. (2017) where diagrams of spine arrangements can be viewed. Measurements. Specimens were measured using a Leica S8APO stereomicroscope with a caliper. All measurements in this study were taken with a caliper and are expressed in millimetres. A total of 22 measurement classes were captured, as in Tedrow et al. (2014), including: 1. Body length = length of body from central ocelli to posterior tip of abdomen (intraspecifically variable measurement, primarily for general size estimation). 2. Forewing length = from proximal margin of axillary sclerites to distal tip of the discoidal region. 3. Hindwing length = from proximal margin of axillary sclerites to distal tip of the discoidal region. 4. Pronotum length = from anterior margin to posterior margin. 5. Prozone length = anterior margin of pronotum to center of supra-coxal sulcus. 6. Pronotum width = from the lateral margins at the widest point, the supra-coxal bulge. 7. Ratio pronotum = ratio between pronotum width and length. 8. Pronotum narrow width = from lateral margins of the pronotum at the narrowest region of metazone. 9. Head width = from lateral margins of the eyes at the widest point. 10. Frons width = from lateral margins of the frons, inferior to the antennal insertions, at the widest point. 11. Frons height = from upper margin abutting central ocellus to lower margin abutting clypeus. 12. Prothoracic coxae length = from pronotum to trochanter. 13. Prothoracic femur length = from proximal margin abutting trochanter to distal margin of genicular lobe. 14. Mesothoracic femur length = from most proximal margin abutting the trochanter to the distal side of the terminal spine insertion site. 15. Mesothoracic tibia length = from most proximal groove near joint with the femur to the distal side of the terminal spine insertion site. 16. Mesothoracic tarsus length = from proximal joint to the apex of the ungues curve. 17. Metathoracic femur length = from most proximal margin abutting the trochanter to the distal side of the terminal spine insertion site. 18. Metathoracic tibia length = from most proximal groove near femoral joint to the distal side of the terminal spine insertion site. 19. Metathoracic tarsus length = from proximal joint to the apex of the ungues curve. 20. Anteroventral femoral spine count = all inner marginal ridge spines, except the distal terminal spur. 21. Anteroventral tibial spine count = all inner marginal ridge spines, except the distal terminal spur. 22. Posteroventral tibial spine count = all outer marginal ridge spines but except the distal terminal spur.
The measurement of the total body length produces a measurement only useful for general assessment of body size rather than species description. Since head position, abdominal expansion, and wing position are all variable, total body length should only be used as a rough measurement to initially discriminate between the small and large Mantodea species when performing identifications.
Imaging. Alive specimen was captured with a NIKON D700 by Philippe Annoyer on 3 December 2010 near the base camp in Dzanga-Ndoki NP. Habitus images were taken with a Konica Minolta Dynax 5D. All images were taken over an 18% grey card background for white balance standards, excluding the image of the C. lestoni paratype from the MNHN. Images were processed in GIMP 2 to adjust levels, contrast, exposure, sharpness, and to add scale bars. Minor adjustments were made using the stamp tool to correct background aberrations and to remove distracting debris. Plates were constructed using Publisher 2016.

Taxonomic placement
The following characters led to place the new species within Chlidonoptera genus: mantids of medium size and bright colours, very similar to Pseudocreobotra genus; but the tips of the lower frons and clypeus very short and blunt, the protuberance of the vertex shorter. The eyes are bulging but rounded. Less expanded pronotum, shorter than anterior coxa: prozone more compressed, higher with two acute conical tubers in front of supracoxal sulcus, no tubercles on the metazone. Wings are beyond the abdomen in both sexes. Forewings of females more dilated from base to apex and hindwings almost opaque, yellow with dark veins; males only the basal part with this coloration, the rest hyaline. Forewings with a large eye spot, a yellow spot near the shoulder and apex on light colour. Anterior femurs are thin. The external spines of the anterior coxa are not swollen at the base, four discoidal spines and four posteroventral femoral spines. Femurs of the meso-and metathoracic legs have a subapical and posteroventral lobe. Laterally lobed present on the abdominal segments.
Known species of the genus Chlidonoptera were compared to the males found in southwestern CAR. Distribution of known individuals of C. werneri, the structure of the genitalia and the morphology described in Roy (1964) and Lombardo (1997) exclude it as a candidate species. Similarly, distribution, structure of the genitalia and morphology described in Roy (1964) excluded C. chopardi as the species. On the other hand, the distinction between C. vexillum and C. lestoni is much more complicated (Roy and Leston 1975; for reference, the imaged types can be seen at http://specimens.mantodea.com). Morphologically, the three species are very similar. Only the structure of the posterior end of the sclerite L4A of the ventral phallomere (hypophallus) enables to distinguish them. The COI-DNA barcoding of 19 Chlidonoptera specimens enabled the differentiation of the new species from C. vexillum collected in Gabon (Moulin 2018a) and Cameroon.  Saussure, 1898. In 1900 gave measurements for females of C. vexillum and males of B. lunata, without putting them in synonymy. The genus was then misspelled, 'Clidonoptera.' W.F. Kirby (1904: 292) continued to conserve the two species, C. vexillum and Bomistria lunata, with also a misspelling in the Sjöstedt citation, 'Chlinidonoptera.' F. Werner (1908: 52), making the point between Chlidonoptera vexillum and Bomistria lunata with supporting illustrations. But, in 1915, Giglio-Tos clarified the situation: B. lunata of Saussure is the male of C. vexillum and as the female B. lunata of F. Werner would be a new genus with a new species, Anabomistria werneri Giglio-Tos, 1915. The location of A. werneri was listed only as 'Africa' (Giglio-Tos 1927: 563). In his great synthesis work, Genera Insectorum, Beier (1934: 26) listed C. vexillum and A. werneri with a description of their morphological features. He stated that A. werneri is from East Africa. Then, in 1964 (p. 939), he confirmed the locality of these species in Hymenopodidae and Hymenopodinae. That same year, R. Roy (1964) synthesised data about Mantodea from the Ivory Coast forest, wherein a new species of Chlidonoptera was described, C. chopardi (p. 764); the male genitalia of which were compared with those of C. vexillum. At the same time, the author reconsidered the genus Anabomestria and logically placed A. werneri in the genus Chlidonoptera. M. Beier (1968: 6, fig. 6b) illustrated the right forewing of A. werneri's female but the taxonomic change of genus made by Roy four years earlier was not taken into account. Later, Chlidonoptera lestoni was described (Roy and Leston 1975: 329) from Ghana. In that same work, C. chopardi was also cited. A comparison of the posterior process (pda) of the ventral phallomere was illustrated for C. chopardi, C. lestoni, and C. vexillum. It was assumed that C. lestoni was close to C. vexillum but distinct; this was not like that which D. Leston wrote in 1968. F. Lombardo (1997

Identification key to species of Chlidonoptera using males
The key to the morphological criteria of Chlidonoptera species can only distinguish C. chopardi, C. werneri and a complex of species, named vexillum group, including C. vexillum, C. lestoni and C. roxanae sp. nov.

1
The smallest species, 23-26 mm (male); Prolongation of the vertex non bifid; forewings with yellow costal area; green discoidal area on almost two-thirds of the basal area, with two yellow spots and two black arcs as in vexillum group, but closer together; hindwings hyaline with pink coloured base; The three species of the vexillum group are difficult to differentiate without using male genitalia. There is a size gradient of the posterior process of the ventral phallomere from the smallest to the largest, from C. lestoni to C. roxanae sp. nov. through C. vexillum, in proportion to the body size. Genitalia of C. lestoni and C. vexillum are represented in Roy & Leston (1975: fig. 9) and in Roy (1964: fig. 7).
The distributions of the different species of Chlidonoptera are shown on the map in Figure 1. Figure 4 Chlidonoptera vexillum:

Chlidonoptera werneri (Giglio-Tos, 1915)
Anabomistria werneri: Giglio-Tos 1915: 108;Beier 1934: 26;Beier 1968: 6. Chlidonoptera werneri: Roy 1964Lombardo 1997: 6;Ehrmann 2002: 95;Otte and Spearman 2005: 87. Figure 4 Chlidonoptera chopardi : Roy 1964: 764;Roy 1965: 595;Ragge and Roy 1967: 586;Gillon andRoy 1968: 1039;Roy and Leston 1975: 297   Natural history. According to the collection locations of different individuals in the canopy, this species is considered to be arboreal. Both nymph and adult specimens, are presumed to reside on the inflorescences of trees. In tropical forests, these flowers are often located at the top, above the canopy, so that pollinators have access to pollen and nectar. In the present study, is only males were captured with a light trap, and were rarely captured during the day. Females Chlidonoptera specimens that were observed by climbing trees or by beating vegetation (Figure 2).
Pronotum: Presenting no special features in comparison with C. vexillum and C. lestoni. Pronotum slightly longer and wider than in other species with always two tubercles slightly directed forward, just above the supracoxal sulcus. Crenellated edges with tubercles of variable sizes. Greenish prozone in the centre and whitish on the sides. Green metazone except on the margin.
Forelegs: Legs very similar in their morphology and coloration to those of the other species previously cited. The anterior femora always with four discoidal spines, four posteroventral femoral spines, and 10-12 anteroventral femoral spines. Anterior tibia has 12-14 anteroventral tibial spines and 14-17 posteroventral tibial spines.
Meso-and metathoracic legs: Legs very similar in their morphology and coloration to those of the other species previously cited.
Wings: Forewing 23.6-30.2 mm in length, featuring the usual colour pattern for the genus, with a yellow spot contained between the two black arcs in a relatively large circle. Hindwings 24.9-27.3 mm long, hyaline, with basal region more or less yellow with red-brownish veins.
Abdomen: It presents no special features in comparison with C. vexillum and C. lestoni. Laterally lobed abdominal segments. Subgenital plate more or less asymmetrical as in the other species; supraanal plate and cerci without special features.
Genitalia: Same type of C. vexillum with the posterior process of the ventral phallomere longer and thicker than in C. vexillum and a ventral phallomere longer (Figures 5, 6).
Etymology. This species is named in honour of my oldest daughter, Roxane, who was growing in her mother's womb, while I was deep in the primary forest of the Central African Republic, for field work in February 2012.
DNA barcoding. Nineteen sequences were obtained from the 25 specimens sampled (Figure 7). C. roxanae sp. nov. and C. vexillum are distant enough from each other (9.4% between them), to allow us to consider them as two different species. BINs (Barcode Index Number) have been attributed to them: BIN: BOLD: ACX2872 for C. roxanae sp. nov. (mean intraspecific divergence 0.19%) and BIN: BOLD: AAZ5470 for C. vexillum (mean intraspecific divergence 0.76%). No fresh specimens of C. lestoni were obtained for barcoding. Nuclear mitochondrial pseudogenes (numts) sometimes lead to the creation of different BINs, a problem which was not encountered here with the differences on the genitalia and the larger general morphology. PCR did not work for six specimens, presumably due to their condition, as they had to be relaxed in order to be mounted, or due to the preserving liquid. These specimens came from Cameroon, Gabon, Republic of the Congo, Tanzania, and Uganda.

Discussion
A larger size, differing genitalia, barcoding analysis, and an isolated geographical location, allowed us to distinguish C. roxanae sp. nov. from other Chlidonoptera species. Since it was not possible to find the male specimen of Chlidonoptera cited in the publication of Roy (2018) about Mantodea in the La Maboké area of the Central African Republic, it is not possible to rule on the species. Geographically, that specimen seems to fit with C. roxanae sp. nov., without certainty, like those in the Sangha-Mbaere region (Moulin et al. 2017). For the specimens from Central East Africa (Democratic Republic of the Congo, Tanzania, Uganda), where PCR did not work, fresh material will be required to perform additional barcoding and to confirm a C. vexillum identification. It is well-known that DNA barcoding revealed cryptic species of Australian Phasmida among specimens organised at the level of morphospecies (Velona et al. 2015). Molecular analyses are of particular importance for a morphologically conserved group of organisms such as Mantodea (but not between genera) or Phasmida.

Acknowledgments
Many thanks to Philippe Grandcolas, Frédéric Legendre, Roger Roy and the Muséum national d'Histoire naturelle for giving me access to the Muséum's specimens. I would also like to thank the Government of the Central African Republic for permitting access for scientific research. I also thank all of the team of the Epiphyte 2008 and SANGHA2012 'Biodiversité en Terre Pygmée' expeditions. DNA barcoding was supported by the International Barcode of Life project led by Paul Hebert at the Biodiversity Institute of Ontario (Guelph, Canada) and by the funds dedicated to the research of the author's company.
Furthermore, I wish to thank the managers and contributors of the Mantodea Species File (MSF, http://mantodea.species.file.org) for the valuable work they have been undertaking for dictyopterists worldwide. Thanks to Simon Poulain of the MNHN for providing photographs of Chlidonoptera lestoni paratypes, to Sean Brogan for the photograph of a female from Cameroon, to Philippe Annoyer for that of a male from CAR, and to Christian Schwarz for data of Chlidonoptera vexillum from Kenya and data of Chlidonoptera lestoni from Ghana. Finally, thanks to the reviewers of the manuscript for their helpful comments and suggestions, which helped improve the quality of this work.