Culicoides Latreille in the sun: faunistic inventory of Culicoides species (Diptera: Ceratopogonidae) in Mayotte (Comoros Archipelago, Indian Ocean)

Background The south-west insular territories of the Indian Ocean have recently received attention concerning the diversity of arthropods of medical or veterinary interest. While a recent study highlighted the circulation of Culicoides-borne viruses, namely bluetongue and epizootic hemorrhagic disease, with clinical cases in Mayotte (comprising two islands, Petite-Terre and Grand-Terre), Comoros Archipelago, no data have been published concerning the species diversity of Culicoides present on the two islands. Results A total of 194,734 biting midges were collected in 18 sites, covering two collection sessions (April and June) in Mayotte. Our study reports for the first time livestock-associated Culicoides species and recorded at least 17 described Afrotropical species and one undescribed species. The most abundant species during the April collection session were C. trifasciellus (84.1%), C. bolitinos (5.4%), C. enderleini (3.9%), C. leucostictus (3.3%) and C. rhizophorensis (2.1%). All other species including C. imicola represented less than 1% of the total collection. Abundance ranged between 126–78,842 females with a mean and median abundance of 14,338 and 5111 individuals/night/site, respectively. During the June collection, the abundance per night was low, ranging between 6–475 individuals. Despite low abundance, C. trifasciellus and C. bolitinos were still the most abundant species. Culicoides sp. #50 is recorded for the first time outside South Africa. Conclusions Our study reports for the first time the Culicoides species list for Mayotte, Comoros Archipelago, Indian Ocean. The low abundance and rare occurrence of C. imicola, which is usually considered the most abundant species in the Afrotropical region, is unexpected. The most abundant and frequent species is C. trifasciellus, which is not considered as a vector species so far, but its role needs further investigation. Further work is needed to describe Culicoides sp. #50 and to carry on faunistic investigations on the other islands of the archipelago as well as in neighboring countries. Electronic supplementary material The online version of this article (10.1186/s13071-019-3379-x) contains supplementary material, which is available to authorized users.

In the south-west insular territories of the Indian Ocean, recent records mentioned five Afrotropical Culicoides species on La Reunion Island (C. imicola, C. enderleini, C. bolitinos, C. grahamii and C. kibatiensis) [7], where outbreaks of BTV and EHDV are regularly observed [30], and two Afrotropical species on Mauritius (C. imicola and C. enderleini) [52]. The faunistic inventory in Madagascar is probably largely incomplete as only 14 species have been recorded [48] and the precise identification of species related to C. schultzei needs further investigation [48,53]. The Seychelles fauna for the genus Culicoides was investigated at three different times and three species were recorded (C. leucosticus, an Afrotropical species; C. kusaiensis, an Oriental species; and C. adamskii, reported only on a small Seychelles atoll) [54,55]. Recent local reports highlighted the circulation of BTV and EHDV with some clinical cases in Mayotte, Comoros Archipelago [56]. Interestingly, no data are published concerning the Culicoides species present on this island.
Herein we report on a survey of Culicoides biting midges conducted in Mayotte, in the context of previous BTV and EHDV clinical cases [56]. A recent serosurvey on the island showed active circulation of both viruses throughout the island, with at least five BTV serotypes and one EHDV serotype [56]. Our survey is the first to address the Culicoides species diversity for the island and the whole archipelago. Our field survey covered different livestock breeding and production present on the island. Our specific objective was to describe the species diversity of Culicoides in Mayotte, to assess the abundance of dominant species and to map their spatial distribution to provide important insight to the epidemiology of the Culicoides-borne viruses on the island. Together with other published checklists for Culicoides in the region (South Africa, Kenya, La Réunion, Seychelles, Mauritius, Zimbabwe), we analyzed the species-area relationship (i.e. number of species in areas of different size irrespective of the identity of the species within the areas) to estimate the species richness in Madagascar.

Methods
Mayotte is an overseas department of France in the south-west part of the Indian Ocean, located in the northern Mozambique Channel. The island, constituted of a main island (Grande-Terre) and a smaller one (Petite-Terre), belongs geographically to the Comoros Archipelago (Fig. 1). The soil type is mostly related to the volcanic origin of the island with massive soil erosion caused by heavy tropical rainfall on unprotected and deforested areas. The latest survey (2010) totaled 5700 cattle farms with 17,150 heads of cattle (less than 5 heads of cattle per farm on average), and 2200 sheep and goat farms with 12,600 animals (less than 6 animals per farm on average), highlighting the importance of smallholder farming. Breeding practices are mostly traditional with tethered cattle (72%) and small ruminants (51%) of local breeds. There is a single horse-riding center on Grande-Terre.
Thirteen sites were prospected, and collections were made from single night collections from 4th to 11th of April 2016, and 5 sites (YT1, YT2, YT10, YT28, YT30) from 20th to 27th of June 2016 (Fig. 1, see Additional file 1: Table S1). The timeframe in April was chosen to match the end of the rainy season when Culicoides populations are supposed to be the highest (dry season from April to November and rainy season from December to March). Selected sites were cattle (YT2, YT4, YT8, YT10, YT14, YT15, YT25, YT29, YT31), mixed farms with cattle and sheep and/or goats (YT1, YT6, YT11, YT13), sheep farms (YT9), goat farms (YT3, YT30) and the unique horse-riding center (YT28) to represent the different ruminants or equids present on the island. Culicoides trapping was done using a black light suction trap (Onderstepoort Veterinary Institute design, powered with a 12 V car battery) set up from before dusk until after sunrise, and positioned in the vicinity of the animal holdings (inside the shelter, along the fence, the closest tree for tethered animals) (Additional file 1: Table S1).
Specimens of Culicoides were stored in 70% alcohol until identification. Large samples were subsampled following a modified procedure described by Van Ark & Meiswinkel [57]. For each large sample, a 3 ml subsample was entirely sorted out and used to calculate the estimated total catch. All individuals were identified to the species level and sexed using a stereomicroscope. Morphological identification was performed using the available literature for the Afrotropical region [44,49,58,59] and the expertise of KL. Biting midge specimens are deposited in the collection of Cirad, UMR ASTRE, Ste Clotilde, La Reunion, France (accession code: YT), and are available upon request to CG. Maps were generated using ArcGis ® software (version 10.3).
To determine the sampling efficiency at the end of the rainy season, species accumulation curves were plotted according to a randomization procedure using the R vegan package version 2.5-1 and by fixing the number of permutations to 100 [60]. A species accumulation curve derives as a plot of cumulative number of species discovered as a function of sampling effort. Each species is considered regardless of its abundance or rarity. The number of non-sampled species was extrapolated by estimating different richness indices (Chao, Chao bias-corrected, first order jackknife, second order jackknife and bootstrap estimators).
One of the reasons why islands are important in ecology and biogeography is that they are relatively isolated areas and therefore excellent natural laboratories to study the relationship between area and species diversity [61]. To estimate Madagascar's Culicoides species diversity, we plotted the area-species curve of the south-west Indian Ocean using literature, our dataset for Mayotte and official island sizes. All analyses were performed using R (https ://www.r-proje ct.org) [62].

Results
During the two collection sessions, 17 farms were prospected for 18 collection sites. Thirteen sites were prospected in April and 5 sites were prospected in June 2016. In one of the farms prospected in June, collections were carried out in two sites because of the presence of cattle (YT10) and goats (YT30) on two separate fields. The other sites prospected in June were the unique horseriding center (YT28), one isolated goat and cattle farm in the north of Grande-Terre (YT1) and one site on Petite Terre (YT2) (Fig. 1).
At least 17 species were recorded during the two sessions (Table 1). One damaged specimen (absence of wings) collected in site YT28 was not identified. In site YT3, DNA of one specimen which could not be identified based on morphological features was extracted and the cox1 gene amplified. Unfortunately, the amplification failed. Out of the 17 species, 16 were known species distributed in the Afrotropical region (Table 2) and one was an undescribed species named Culicoides sp. #50 [49,63,64]. This latter species was collected in 9 sites (Fig. 1, Table 1). Ten species were collected both in April and in June: Culicoides albopunctatus, C. bolitinos, C enderleini, C. imicola, C. leucostictus, C. rhizophorensis, C. similis, C. subschultzei, C. trifasciellus and the undescribed species. Four species were collected only in April: C. accraensis, C. distinctipennis, C. milnei and C. moreli. Three species were only collected in June: C. dekeyseri, C. kibatiensis and C. nevilli plus one unidentified specimen (Fig. 1, Table 1). The species accumulation curve highlighted that the collection effort was sufficient to cover the species richness on Mayotte at the end of the rainy season ( Fig. 2). Moreover, the different indices used to calculate extrapolated richness and compare it to our dataset showed that we could have missed from 1 (bootstrap) to 5 species (second order jackknife). Overall, this indicates that the inventory for the rainy season was notably robust and comprehensive.
A total of 194,734 individuals were collected in the 18 sites during 20 collection nights (Table 1). Indeed, due to social protests resulting in road barricades, two sites (YT6 and YT14) could not be reached on the morning after the trap was set up but only the day after. As the lights and fans of these two traps were still running correctly, we assumed that the mean of the total catch of the two nights of each trap best represented single night catch estimates. Overall, 98.29% of individuals were females (191,401) and 1.71% were males (3333). Taking into account 18 collection nights, a total of 187,302 individuals were identified during the two sessions (mean and median catch per night 10,406 and 1436, respectively) with 98.25% of females (184,026.5) and 1.75% of males (3275.5).
As expected, over 99.5% of the individuals (186,390 individuals) were caught in the 13 sites sampled in April and only 912 individuals were collected in June. Percentage of females in April was 98.26 in June (183,137.5 females) and 97.48 in April (889 females).
During the June collection session, the abundance per night was low, ranging between 6 and 475 individuals. Mean and median abundance were 182 and 204      Fig. 3). Culicoides bolitinos was the most abundant species in 3 out of the 5 sites; C. rhizophorensis and C. trifasciellus were the most abundant in YT10 and YT30, respectively. Culicoides trifasciellus was also present in 2 other sites but in low numbers (1 and 6 individuals in YT1 and YT2, respectively). The number of species collected per site varied between 2 and 9 ( Fig. 1, Table 1).

Discussion
Faunistic inventory of mosquitoes was recently made in the south-western islands of the Indian Ocean [5,6] but no such work has been done before for Culicoides species. Yet, completing a species checklist is the first fundamental step before any further epidemiological studies on vector species can be launched. Indeed, tropical oceanic islands are highly appropriate for surveys to attempt to complete a species checklist for a given group as they are isolated and endemism may be rampant [61].
Our study reports for the first time livestock-associated Culicoides species and records at least 17 described Afrotropical species with one undescribed species (Culicoides sp. #50). The reference trap for Culicoides collection was used because it can catch the highest diversity and abundance regardless of the season [65]. However, as it is a light-trap, species with diurnal activities may have been missed. Traps were localized in farms with cattle or small ruminants to investigate the species in close contact with hosts for BTV and EHDV. Therefore, we could have missed species breeding in other ecosystems such as sandy beaches, mangroves or sylvatic species. However, the species accumulative curve plot allows to consider our study as a good picture of the diversity of livestockassociated species at the supposed abundance peak (end of the rainy season). As the statistical analysis was carried out on species richness at the end of the rainy season, this work could be completed by another inventory during the dry season (only five sites were prospected in June).
Considering the size of Mayotte (374 km 2 ) and its volcanic origin, the species diversity is high (Reunion island, five species for 2512 km 2 ; Mauritius, two species for 1865 km 2 ). All the species recorded in Mayotte had been previously found in South Africa [49], while the species presence and diversity composition was quite different from that recorded on neighboring island territories. In Mauritius, the last update inventory mentioned two species, C. imicola and C. enderleini [52], which needs to be confirmed with a large-scale survey. Different surveys recorded five species on La Réunion island [7,66] and allow to consider this figure as robust. Unfortunately, the Culicoides diversity is clearly underestimated in Madagascar [53] and unknown on the other islands of the Comoros Archipelago. Our estimation of species number in Madagascar reached 71 species (Fig. 5). Although entomological surveys were undertaken in the Seychelles early in the past century [54], only two inventories were published for the islands. The oldest one [54] reported two species, C. leucosticus, an Afrotropical species, and C. kusaiensis, an Australasian species extending to Southeast Asia: Melanesia; Micronesia; Australia (Queensland) [67]; Thailand [68]; China [67]; and Malaysia [54]. This species list was further completed with a description of a new species from the Aldabra island, Seychelles, named C. adamskii [55]. All these collections in the different territories are incomplete and certainly require updating. The high species diversity recorded in Mayotte might be explained by the close geographical connection to the African continent which may facilitate Culicoides dispersal [69]. Indeed, the Comoros Archipelago has strong links to the African continent through current and past trade and human migrations.
Culicoides imicola was collected in eight sites and with a maximum abundance of 116 individuals per night. This is an unexpectedly low abundance and patchy distribution of C. imicola. Culicoides imicola is usually the most frequent and/or abundant species in the fringe of the African continent (Mediterranean basin and south of Africa where the climate is characterized by dry summers and rainy winters) [49,65,70,71], while less abundant or frequent in other regions of the continent [42,43,72]. This could be related to soil type or other environmental variables. In our survey, this could be explained by the limited number of hosts and low cattle density and abundant sylvatic environment around the sites. The most abundant and frequent species was C. trifasciellus. This species has often been mentioned in Central Africa [37] or Kenya [73] studies and recently as a cave breeding species in Gabon [74]. It has been reported as an anthropophilic species [37,49,73] but no such nuisance was noticed during our fieldwork or reported by farmers. Moreover, C. trifasciellus is the vector species of Onchocerca gutturosa, a cattle microfilaria. We have no data attesting the presence of this parasite on the island, but it has been recorded on the African continent [75,76].
To our knowledge and based on literature, there are no data on the vector role of C. trifasciellus for BTV or EHDV (Table 3). Culicoides bolitinos and C. enderleini, the 2nd and 3rd most abundant species respectively found in Mayotte, are known for their vector role [49]. The vector competence of C. bolitinos has been demonstrated in the laboratory for several viruses [77][78][79][80] and its host preference towards horses and ruminants has been documented [42,58,81,82] which makes this species a major vector species for BTV, AHS and EHD virus in the Afrotropical region. Culicoides enderleini is highly suspected to be implicated in BTV transmission based on laboratory susceptibility studies and isolations of BTV in South African Culicoides populations [78,80].
Culicoides leucostictus and C. rhizophorensis were frequently and abundantly collected in our survey. In a study by Venter et al. [78], BTV isolation was successful from one pool of C. leucostictus while it is a common and widespread species in South Africa, being dominant near birds [49,83,84]. Indeed, the species was not attracted by horse or sheep baits in a recent host vector contact study in Senegal [81]. Both species are reported to breed in swampy, saline areas and salt-marshes environment in South Africa, like those created by periodical flooding with sea water due to tidal activity [49,83]. Indeed, C. rhizophorensis was notably collected in our survey in farms close to the coast.
Culicoides sp. #50 is reported for the first time outside its known distribution range, i.e. South Africa [49,63]. This species was mentioned for the first time in the Kruger National Park in South Africa reared from the dung of elephant and plains zebras [64]. Our record updates the known species distribution and its biology as no big wild mammals are present in Mayotte.
The assumption that competence for orbiviruses might be widespread in the genus Culicoides encourages further assessment of the role of each species in relation to its abundance and seasonality [78]. Meanwhile, the potential involvement of numerous species in virus transmission, each exhibiting different bionomics and phenology, greatly increases the complexity of the epidemiology of Culicoides-borne viruses. Because of the limited number of livestock on the island and low ruminant density, species usually associated with livestock farming in the Afrotropical region were either collected in small numbers (C. imicola, C. bolitinos and C. milnei) or absent (C. kingi). We could also not exclude the assumption that local ecological conditions (soil composition) are not favorable for these species. The relatively large number of C. leucostictus and C. rhizophorensis might be due to the presence of natural larval habitats around the prospected farms. Overall, no clear spatial pattern was observed regarding species diversity or abundance.
Culicoides species delimitation is commonly known to be complicated by large morphological variations [85]. Recently, systematics and taxonomy of the Afrotropical species of Culicoides using molecular tools [41,48,86]