Diversity of mantids ( Dictyoptera : Mantodea ) of Sangha-Mbaere Region , Central African Republic , with some ecological data and DNA barcoding

This study aims at assessing mantid diversity and community structure in a part of the territory of the Sangha Tri-National UNESCO World Heritage Site in the Central African Republic (CAR), including the special forest reserve of Dzanga-Sangha, the Dzanga-Ndoki National Park. The study area is located in the biome of the dense tropical rainforest of the Congo Basin, the second largest area of tropical forest behind the Amazon basin, and one of the last regions on earth where pristine forest ecosystems persist. Between 1984 and 2012, eight collecting campaigns were conducted by P. Annoyer, M. Loubes and S. Danflous, with the aim of documenting invertebrate diversity in this remote and poorly studied area. Mantids were collected in different habitats and microhabitats using four harvesting techniques: visual hunting (threshing, mowing), active search in the trees, autonomous UV collecting (Remote Canopy Trap) and classical light trapping with mercury light. Community patterns were assessed using rarefaction curves and diversity estimators (ACE). Specimens (n = 25) were also found at the National Museum of Natural History (MNHN) of Paris. Seventy-one species of Mantodea are recorded from Sangha-Mbaere Region in the CAR from 1232 specimens. This is the first synthesis published about Mantodea in CAR. Some new species are present in the genera Cataspilota, Galepsus and Chlidonoptera. Taxonomic revisions of these genera are in progress. Some genera need global revision like Plistospilota, Miomantis, Entella, and Galepsus. This type of work is the first on the mantids in Central Africa since the various inventories carried out in the sixties and seventies, which were mainly based on lists of species. The contribution of data on ecology and biogeography is a novelty.


Introduction
Currently, over 2,400 valid species of Mantodea have been recorded worldwide (Giglio-Tos 1927, Ehrmann 2002, Otte and Spearman 2005, Otte et al. 2017).Since 2005, several generic revisions have been published (Roy 2009, 2010, Roy 2013a, 2013b, Roy and Stiewe 2014, Tedrow et al. 2014, Svenson et al. 2015).In Africa, only surveys by R. Roy, in the years 1960 to 1980, provided distribution records of Mantodea from several African countries (Ivory Coast, Gabon, Ghana, Guinea, Senegal, etc.); and those of A. Kaltenbach in 1996 and 1998 provided records from South Africa.During all these years, many research stations were in operation, and many samples were collected there (La Maboké-Boukoko in CAR, for example).They are, for the most part, unpublished.More recently, a team published results from a survey in Rwanda (Tedrow et al. 2015).According to this research, the number of African Mantodea is estimated to be over 1000 valid species.
The rainforests of Central Africa are the largest rainforest block after the Amazon basin (UNEP 2008).This is one of the last undisturbed biological environments.Evergreen forests of Central Africa, with an area of about 1.62 million square kilometers (36% of the country surface), represent an exceptional natural heritage (Basset 2001, Basset et al. 2003, Lowman and Rinker 2004, Vande Weghe 2004, Dial et al. 2006, UNEP 2008).They host a significant proportion of the world's biodiversity, and play an important role in climate regulation and carbon sequestration (Ngatoua 1997, Aveling andDebonnet 2010).
Sangha-Mbaere Region is located in the Southwest of the Central African Republic (CAR) (Fig. 1).It is named in reference to the Sangha and Mbaere Rivers, which both represent the main elements of the regional hydrographic network.This region covers a total area of 19 412 km² and is extensively covered by tropical rainforests in the southern triangular-shaped part of the region.It contains several natural lakes of different size (Massa 2013, de Wasseige et al. 2014).Two main types of forests have been distinguished: a) dryland forests present open, mixed canopy dominated by Sterculiaceae and Ulmaceae, generally with a dense understory vegetation of Maranthaceae and Zingiberaceae (Aveling and Debonnet 2010, Massa 2013), b) semi-evergreen forests contain swamp-forest areas along the rivers, and closed-canopy, monodominant Gilbertiodendron dewevrei formations, with trees over 50 meters high (UNEP 2008).The northern part of the Region, north of Nola, is currently undergoing a dramatic process of deforesta-tion.In the extreme southwest, Dzanga-Ndoki National Park is part of the Sangha Tri-National (TNS), a UNESCO World Heritage Site which covers an area of 4 420 000 ha and includes three national parks: Loboké (Cameroon), Nouabalé-Ndoki (Congo) and Dzanga-Ndoki (CAR) (Ngatoua 1997, Massa 2013).Another National Park, the Mbaéré-Bodingué National Park, was recently created near Mbaere River, including the beautiful Ngotto Forest (Brugiere et al. 2005, Ndarata Massanguet 2012).This part of the region is a very rich sanctuary of biodiversity, which is the focus of recent scientific investigations.
Here, we report on the diversity of the Mantodea (Insecta: Dictyoptera) found within Sangha-Mbaere Region since 1984.Couturier et al. (1986) described that some insect collections remained in the research center of Boukoko and La Maboké, CAR, after de-parture of expatriate researchers.Unfortunately, poor conservation of these specimens did not permit us to study them.Currently, the vast majority of Mantodea from CAR are conserved at the Muséum national d'Histoire naturelle of Paris.Nevertheless, mantids in this collection are mainly native to the Lobaye, CAR, region.These specimens are disregarded in the current work, but will be the focus for a future publication.
Our review covers all species found on decades of collecting trips through Sangha-Mbaere Region.We discuss the ecological traits and taxonomic status of the Mantodea of this under-sampled region.Some specimens are not identified to species.They will be described in other papers, which will include generic revision (e.g.Cataspilota, Chlidonoptera, Plistospilota).Some others need more taxonomic research (e.g.Miomantis, Entella, Galepsus).

Material and methods
Specimen sampling.-Since1984, several trips were made to the Sangha-Mbaere Region to collect insects (with a focus on Lepidoptera: Rhopalocera).In 2008, the expedition, named Epiphyte 2008, visited the Dzanga-Sangha Reserve.In November 2010, the one-month long expedition was a reconnaissance mission which paved the way for a larger expedition of two months, called SANG-HA "Biodiversité en Terre Pygmée", in January-March 2012.The last two expeditions (2010 and 2012) were organized in the Dzanga-Ndoki National Park.
Most specimens were collected using light traps consisting of a white sheet illuminated by a 250W mercury vapor bulb.In 2010 and 2012, light traps (Mercury Vapor lamps, economics lamps, neons) were complemented by other sampling procedures, including visual collecting during tree climbing, night and day research in understory vegetation, remote canopy trap, and aerial interception trap (Fig. 2).Historically, scientific research on the praying mantids made use of their attraction to light (Mercury Vapor).Lamps producing ultraviolet light are suspended in front of vertical white sheets.However, mostly males are attracted and very few females, and only from the very few species that are attracted to light.To complete the sampling, mowing, beating vegetation and visual searching for individuals during the day were also performed.Using these techniques, many nymph and female individuals were collected.
Specimens were killed using cyanide vapour jars, dried by wood fire smoke and preserved in paper bags or on cotton wrap.All specimens were then studied in the laboratory of Nicolas Moulin.
Taxonomic assignations.-Identificationswere done following the updated taxonomy in the Mantodea Species File website (Otte et  al. 2017) and in recent works on praying mantis molecular phylogenetics (Svenson and Whiting 2009, Wieland 2013, Roy 2014, Svenson et al. 2015).The identifications were made on the basis of habitus morphology (Figs 3-4), color patterns, foreleg framing and genitalia morphology.
Ecological analysis.-Anoccurrence table consisting of 71 species and 1232 individuals was constructed, where the locality, habitat and sampling method were documented.This table was then used to compute observed richness (calculated as the total number of species observed in a given locality or habitat, or for a given sampling method) and an estimation of the theoretical species richness using the ACE diversity estimator using the 'Vegan' Package for R v. 2.7.2. (Oksanen et al. 2008).Additionally, the completeness of the sampling at a regional scale was assessed by computing a rarefaction / extrapolation curve using the iNEXT package (Chao and Jost 2012).On the ground, distinctions were made between sampling in herbaceous vegetation (<1 m), shrubs (between 1 m and 3 m) and trees (between 3 m and canopy).
DNA barcoding.-DNAbarcoding, the analysis of a standardized segment of the mitochondrial cytochrome c oxidase subunit I (COI) gene, was performed on a representative selection of specimens (171, selected by diversity criteria).Tissues were sent to the Canadian Center for DNA Barcoding (CCDB) at the University of Guelph for DNA extraction, polymerase chain reaction (PCR) and sequencing.PCR was performed using the PCR primers C_ LepFolF/C_LepFolR.Data are currently managed under projects: "Mantodea of Gabon -Project 1 [ECOTROP 2014]", "Mantodea of Gabon -Project 2 [ECOTROP 2011]", "DNA Barcoding Mantodea -Collection N. Moulin" at Barcode of Life Data Systems (BOLD, Biodiversity Institute of Ontario, Canada; boldsystems.org).Kimura-2-parameter (K2P) distances were calculated using the BOLD 4.0 interface (Ratnasingham and Hebert 2007).Sequences were then analyzed and trees constructed using BOLD 4.0 interface.

Hymenopodidae, Acromantinae, Otomantini
Genus Chrysomantis Giglio-Tos, 1915 Most recent genus revision (Roy 2013a).Genus Cataspilota Giglio-Tos, 1917 Note.-This genus is being reviewed.Sampling method efficiency.-Lighttrapping was the most efficient sampling method, with almost 1000 specimens captured, while day collecting only resulted in the collection of 250 specimens (Fig. 5).As for orthopterans, we observed that many mantises were attracted more efficiently by diffuse light (camp lamp in tents, classic neon, low energy consumption lamps) than by the 250W UV light traditionally used in light traps for moths and beetles.As expected, the proportion of females and nymphs was higher during the day than with light traps, and interestingly the number of species collected with both methods was not significantly different (Fig. 6).Sampling completeness, represented by the % of estimated richness that has been observed in the samples, was higher for light trapping (ca.92%) than for day collecting (ca.74%), suggesting that additional sampling effort with this last method would be necessary to reach an accurate estimation of the mantid community diversity.Figure 7 shows the number of species caught specifically for each method.

Cataspilota calabarica
DNA barcoding versus traditional taxonomy.-Atotal of 94 sequences of more than 200°bp were obtained from 119 specimens analysed.Except in a few cases, the delimitation of taxonomic units on the basis of genetic information contained in the barcodes was congruent with the species limits acknowledged by traditional taxonomy (Fig. 8).
Diversity patterns of praying mantids.-Therarefaction curve obtained for the whole data set (Fig. 9) indicates that the sampling effort was sufficient to provide a representative picture of Mantodea diversity at a regional scale.As exemplified by this figure, doubling the number of collected specimens would only result in the addition to the checklist of a small number of new species.Figure 10 shows diversity patterns in the three major geographical units of the study region, i.e.Dzanga-Ndoki National Park, Dzanga-Sangha Special Reserve and the rest of Sangha-Mbaere Region.The higher observed and estimated species richness was found in the Dzanga-Ndoki NP, where the number of species was close to the number observed at the regional scale, meaning that most of the regional species pool was present in this area (Fig. 10).
Vegetation structure.-Figures11 to 13 present the number of individuals and the diversity of mantids collected along a height gradient from the herbaceous layer to the top canopy.Most of the mantids captured were males because the main method of surveying was through the use of light traps.However, both in top canopy and in the tree stratum, a higher proportion of females were collected by active search and light trap, maybe because UV light was placed nearer to the microhabitats where females resided.The number of specimens found in the soil and herbaceous strata was lower than in other strata because the study area is largely situated in a forest context (Fig. 11).Both observed and estimated species richness were the highest between the shrub layer and the top canopy).However, a significant number of species was found in the herbaceous surroundings and edges of the swamps and lakes (Fig. 12).Each stratum of the vegetation harbors species in a specific way but the species also overlap somewhat in different vegetation types (Fig. 13).

Discussion
The data obtained during this work provide new insights into our knowledge of the diversity of praying mantids in the Sangha-Mbaere region and by extension in CAR.At present, mass data from the 1960s exist for La Maboke-Boukoko research station in the Lobaye region but are not yet published.The work presented here also allowed the establishment of a vast reference library of DNA barcodes (COI) for the mantids of Central Africa.
The different sampling methods carried out in the different field surveys provided us with a level of sampling effort that was rarely attained in past studies.It is worth reporting that 6 species were only observed during the day and 20 others only at night, which clearly illustrates the importance of combining different sampling approaches.
The lower richness observed in the Dzanga-Sangha SR could be explained by a difference in forest composition with, for instance, a higher proportion of secondary forests compared with Dzanga-Ndoki NP.Even if a higher sampling effort in Dzanga-Ndoki NP could have resulted in a higher number of observed species, it could hardly explain the difference in the estimated rich-ness between the two regions.The low diversity found in the rest of Sangha-Mbaere Region is certainly linked to a lack of sampling intensity in this area.
In the genus Chlidonoptera, barcoding diagnoses separate species from specimens with very slight differences in habitus and genitalia.Individuals from Sangha-Mbaere are closely related to Chlidonoptera vexillum, which was described from Cameroon and is known from other regions of Central Africa, and Chlidonoptera lestoni which to date is known only from Ghana.A similar case was found within the genus Cataspilota, where DNA barcodes confirmed slight but consistent morphological differences observed between specimens from Sangha-Mbaere and those of Cataspilota lolodorfana from Cameroon.DNA barcodes also support the existence of a new Galepsus sp., which is clearly distinguishable from all other species within this genus on the basis of habitus and genitalia.The case of Plistospilota sp.deserves more attention, as morphological identifications within this genus are often difficult, whereas barcoding seems to clearly separate different BINs.For instance, specimens from Sangha-Mbaere could belong to either Plistospilota maxima or Plistospilota validissima but it was impossible to determine due to the very short and weakly informative original descriptions of both of these closely related species.Two other cases that merit consideration further are represented by Polyspilota aeruginosa, which is subdivided in several BINs while no morphological differences could be found to support these divisions, and by Stenopyga ziela that was grouped with specimens of Stenopyga ipassa in a single BIN, whereas both species are unambiguously differentiated either by their habitus or genitalia.
Although reaching exhaustivity in a biodiversity survey for a group of tropical arthropods remains an unattainable objective, our work produced an extensive list of species with a quite high sampling completeness.The active search for specimens in the vegetation proved to be essential in order to have a comprehensive vision of mantid fauna in a given environment.However, the complexity of the vegetation matrix in a tropical rainforest context did not allow us to study the whole volume of available habitats and microhabitats.More sampling would probably result in the discovery of more species restricted to specific ecological niches.Canopy surveying was also very important because many mantids have ecological niches that can be restricted to this stratum.Thus, the verticality of more than 40 m in an old secondary or in a primary forest is useful to distinguish species with different ecological features.In conclusion, we recommend not restricting future studies to the sole use of light trapping, a by far too much restrictive approach to produce a representative picture of the structure of mantid assemblages.

Fig. 1 .
Fig. 1.Sangha-Mbaere Region in Central African Republic.A. Location of Tri National Sangha area in Central Africa; B. Location of principal sampling stations and largest towns.Black dots: Main city in Sangha-Mbaere Region.White dots: Main collection areas.Sources: http://www.diva-gis.orgadminstrative area, Sangha-Mbaere Region; C. Illustration of the habitats studied; D. Primary forest of CAR, Dzanga Bai with forest elephants.

Fig. 2 .
Fig. 2. Illustration of some sampling methods used during surveys in CAR. A. Beating sheet; B. Sweep net; C. Tree climbing; D. Remote Canopy Trap.

Fig. 5 .
Fig. 5. Number of mantid specimens collected with the two main sampling methods: day capture (Day) and trapping with UV lights at night (UV); black chart = females; grey chart = males; white chart = nymphs.

Fig. 6 .Fig. 7 .
Fig. 6. Specific richness of Mantodea collected with two main sampling methods; day capture (Day) and trapping with UV light at night (UV); grey chart = observed richness; dashed chart = estimated richness (ACE index) and associated standard deviation.

Fig. 9 .
Fig. 9. Rarefaction curve of Mantodea from Sangha-Mbaere Region (black curve) with an extrapolation in a hypothetical situation where sampling effort would be doubled (dashed curve); grey area = confidence interval.

Fig. 12 .
Fig. 12. Specific richness in the different vegetation strata; grey = observed richness; dashed chart = estimated richness (ACE index) and associated standard deviation.
Deposition of specimens.-Allspecimens are, at the moment, in the "Insectes du Monde" collection (Ph.Annoyer, Sabine, France) and in the Research Collection of Nicolas Moulin (Montérolier, France).