The Tabanidae of the Mitaraka expedition, with an updated check list of French Guiana (Diptera)

Abstract This paper documents the horse fly fauna collected in lowland rainforest in the southwesternmost part of French Guiana (Mitaraka). During this “Our Planet Revisited” survey nine tabanid species were recorded from French Guiana for the first time: Chrysops ecuadorensis Lutz, C. incisus Macquart, Catachlorops amazonicus Henriques & Gorayeb, Chlorotabanus flagellatus Krolow & Henriques, Cryptoylus cauri Stone, Phaeotabanus phaeopterus Fairchild, Philipotabanus stigmaticalis (Kröber), Stypommisa captiroptera (Kröber) and Tabanus amapaensis Fairchild. An updated check list of Tabanidae of French Guiana is presented, including 79 species and one unidentified Chrysops.


Introduction
The horse flies (Diptera, Tabanidae) have a worldwide distribution with almost 4,400 valid species (Pape et al. 2011). The Neotropical region has the highest species richness with approximately 1,205 species (Henriques et al. 2012), about 28% of the global tabanid fauna.
In French Guiana tabanid diversity has only poorly been studied. Except for species described by e.g., Fabricius and Macquart in the 18 th and 19 th centuries, only few species have been recorded from this part of South America and the Kröber catalogue (1934) only lists 22 species. Subsequent species lists were provided by Floch (1955) and Floch and Fauran (1955). Fairchild (1970) extended the list of French Guiana to 38 species by compiling data from the literature (including original descriptions), Floch's work, and by examining material from the Muséum National d´Histoire Naturelle (MNHN, Paris, France). In the second part of the same manuscript, through material received from A.S. Balachowsky, Fairchild described two new species and added eight new records, which further increased the number to 48 species. More recently Raymond et al. (1984) recorded another 15 species for the first time from French Guiana. Other significant inventories by Raymond (1986Raymond ( , 1987 investigating the efficiency of sampling methods also added new records and confirmed old ones. In contrast to the compiled number of species from the above-mentioned papers (63 spp.), in the most recent Neotropical catalogue merely 48 species were cited from French Guiana, with 35 restricted to French Guiana, and 13 with a wider Neotropical distribution (Coscarón and Papavero 2009).
In 2015, a biodiversity survey was conducted in the southwesternmost part of French Guiana ) that produced a substantial number of dipteran samples, including diverse Tabanidae (Pollet et al. 2015). The objective of the present paper is to document on the tabanid fauna encountered during the Mitaraka 2015 survey (French Guiana) and to present an updated check list of Tabanidae of French Guiana.

Methods
In 2015 the "Our Planet Revisited" or "La Planète revisitée" Guyane 2014-2015 expedition, also known as the "Mitaraka 2015 survey", was conducted in French Guiana (Pollet et al. 2014. This was the 5 th edition of a large-scale biodiversity survey undertaken by the French Museum of Natural History in Paris and the NGO Pro-Natura international (both in France). Both organizations jointly run the "Our Planet Reviewed" programme which aims to rehabilitate taxonomical work that focuses on the largely neglected components of global biodiversity, i.e., invertebrates (both marine and terrestrial). Basic arthropod taxonomy and species discovery were at the heart of the survey, although forest ecology and biodiversity distribution modelling, nevertheless, were also part of the project. The expedition was conducted in the Mitaraka Mountains, a largely unknown and uninhabited area in the southwestern-most corner of French Guiana, directly bordering Surinam and Brazil (Fig. 1). It is part of the Tumuc Humac mountain chain, extending east in Amapa region and west in southern Surinam. The area consists primarily of tropical lowland rain forest with scattered inselbergs, isolated hills that stand above the forest plains .
From 22 February to 11 March 2015, a team of 32 researchers explored the area, including 12 invertebrate experts. During a second period (11 -27 March), a second equal-sized team took over and a third smaller team returned to the site from 12 to 20 August 2015. MP was the coordinator of the collected Diptera, and was also the only Diptera worker actively involved in this survey. Invertebrate sampling was carried out near the base camp, on the drop zone (an area near the base camp that had been clearcut entirely to allow helicopters to land) and, in particular, along four trails of approximately 3.5 km that started from the base camp in four different directions (Fig. 6). During the first period (22 February to 11 March 2015) more than 21 different collecting methods were applied, with a total of 401 traps operational within a perimeter of 1 km². This array consisted primarily of pan traps (n = 280), Charax butterfly traps (n = 50), square Malaise traps (SLAM) (n = 32), Flight Intercept Traps (FIT, n = 13) and Butterfly banana traps (n = 12), but also a light trap (Figs 7-10). In the second and third periods, pan traps were no longer included. A total of 217 invertebrate samples (often pooled yields of different traps of the same type) were examined, including 93 sweepnet samples, and 27 and 62 samples collected by SLAM and coloured pan traps (24 blue, 22 yellow and 16 white traps), respectively. As MP mainly focused on Dolichopodidae during active collecting, sweep net samples only rarely contained tabanids. Relevant metadata on the samples (e.g., exact locality and geographic coordinates, date or time period, collection method, and collector(s)) are provided in Appendix 1.
Non-pan trap samples were sorted to insect orders and families at the SEAG offices (http://insectafgseag.myspecies.info/fr), while pan trap samples were treated similarly at MP's home lab. Dipteran subsamples (mostly per family) were subsequently disseminated among experts worldwide, in the case of Tabanidae to TKK and ALH. The identification of the tabanid species was conducted by ALH and TKK using taxonomical reviews and identification keys (Barretto 1950, Fairchild and Philip 1960, Fairchild 1976, Wilkerson and Fairchild 1982, Fairchild 1983, 1984, 1985, Gorayeb and Fairchild 1985, Fairchild and Wilkerson 1986, Burger 1996, Henriques and Gorayeb 1999, Henriques 2006, Turcatel et al. 2010, Krolow et al. 2015, original descriptions, and direct compari son to reliably identified species from the Invertebrates Collection of the Instituto Nacional de Pesquisas da Amazônia, Manaus, Brazil (INPA) and the Entomological Collec tion of the Universidade Federal do Tocantins, Porto Nacional, Brazil (CEUFT). All collected material was stored in 70% alcohol during the expedition, being dry mounted on pins only about 11 months later in the laboratory. Preservation in alcohol usually affects the recognition of diagnostic features, which often no longer allows identification to species level.
In order to build an updated check list, species distribution records were compiled from the following literature: Fairchild (1976Fairchild ( , 1983Fairchild ( , 1984, Henriques and Gorayeb (1993), Henriques and Rafael (1993), Fairchild and Burger (1994), , Coscarón and Papavero (2009), , Turcatel et al. (2010), Krolow et al. (2015), and Henriques (2016). Doubtful country records are indicated by "?". Next to previously published records, all records from the Mitaraka 2015 survey are included in the check list. Each of these records is represented by the sample code and the number and gender of the collected specimens. Detailed information on the samples is given in Appendix 1. First records for French Guiana are explicitly indicated.

Results
A total number of 255 tabanids of 24 species was collected during the Mitaraka 2015 survey. The subfamily Tabaninae is clearly the best represented with 19 species, followed by Chrysopsinae with three species, and Pangoniinae with two species. Of the 24 species only one belonging to Chrysops cannot be identified at a specific level. Female specimens were dominant in the samples, accounting for 233 specimens. Nineteen of the 22 males were collected at the light trap. The 6 m long Malaise trap that was installed over a river proved to be most productive, and collected nearly 2/5 of the specimens (see Table 1). Also SLAM traps, light traps, and flight intercept traps yielded at least 10 different species. In sharp contrast to this, neither blue nor yellow or white pan traps produced one single tabanid. In palm forests and forests along rivers, only Bolbodimyia brunneipennis Stone, Dichelacera marginata Macquart and Pityocera cervus (Wiedemann) were encountered. Fifteen different species were encountered on or near the drop zone and 16 species in the Malaise trap over the river. D. marginata seems   to prefer humid sites near open water as only one specimen was collected on the drop zone compared to 15 in wet forests and 34 along the river. This investigation revealed ten species recorded for the first time from French Guiana (see check list). After also screening previous records in the literature, an updated check list of 80 species of Tabanidae is presented here.

Chrysops sp.
Comment: Two specimens of this morphotype were captured, but it was not possible to identify them with safety by the lack of recent taxonomic works of this genus.
Distribution: Trinidad and Venezuela to Argentina (Chaco).

Distribution: Guyana, French Guiana.
Remarks: this species was not recognized as Stypommisa by Fairchild and Wilkerson (1986), and neither transferred to another genus. For unclear reasons, it was omitted in the Fairchild and Burger catalog (1994), but listed as Stypommisa by Coscarón and Papavero (2009).

Record excluded from French Guiana
Tabanus unipunctatus (Bigot 1892) was cited from French Guiana by Fairchild (1970). However, in the Fairchild and Burger catalog (1994) the species distribution was corrected to: Mexico to western Colombia. Probably the 1970 Fairchild record refers to T. fumomarginatus.

Discussion
French Guiana is part of the Guiana shield in northern Amazonia, bordering with Suriname in the west and Brazil (Amapá State) in the east, between the Maroni and Oiapoque rivers (Guitet et al. 2013). The Amazon rainforest covers more than 90% of this French department, while savannas and mangroves are present only along the coast (Guitet et al. 2014).  In their check list of insects of French Guiana, Brûlé and Touroult (2014) registered about 15,100 valid species names allocated in 20 orders and 322 families. According to the authors, Diptera is one of the poorest studied groups, with only 577 known species, including 6 endemic species, 50 species described from French Guiana, and 2 dubious records.
A high insect endemism in French Guiana is not very likely, because the country does not have strong geographical barriers with its neighbouring countries, Suriname and Brazil (Amapá) (Brûlé and Touroult 2014), and the same habitat types (or life zones) are present in each of these regions. This seems to be suggested by the observation that Suriname and Amapá share 49 and 42 species of Tabanidae (excluding the species with a large distribution) with French Guiana, respectively (Coscarón and Papavero 2009).
As expected, most species (76 sp.) observed in French Guiana belongs to the Amazonian tabanid fauna. Of its 80 species, 32 species have a large distribution in the Amazon basin, 30 species are shared by French Guiana with Suriname and/or Amapá state, and another 13 species with Guyana and/or Pará state. Three species have an even more extensive distribution range beyond French Guiana. Only one species might be endemic and another could not be identified, possibly a new species of Chrysops.
The distribution records of Coscarón and Papavero (2009) were analysed, and it is estimated that approximately an additional 43 species have a high probability of occurring in French Guiana (Table 2). All estimated species have records from Suriname (11 spp.), Amapá (10 spp.), or both regions (2 spp.), or have a wide distribution in the Amazon region (20 spp.).
With respect to the collecting methods, although interception traps (including Malaise traps and SLAM) are a passive method and without attractive power, they are among the most effective methods for capturing female tabanids, because the females are strong and frequent flyers, travelling great distances daily looking for a blood meal. The six meters Malaise trap is extremely effective for Tabanidae, and on some occasions several hundreds of specimens have been collected during one day (Gressitt and Gressitt 1962). According to Brown (2005), the Malaise trap method is especially effective to collect Neotropical Diptera, and Tabanidae seems to be one of 22 most abundant families in Malaise trap samples.
While the females are satisfactorily collected by interception traps, the males are rarely found in these traps, mainly because they are nectarivores, and thus do not need to travel far in search of warm-blooded hosts. As a result, male tabanids are also poorly represented in collections and even often unknown. Their rarity in interception traps might also be related to the effect of flowering periods, their preference to fly in higher tree strata or by their flight in restricted areas waiting for females to mate . In contrast, males are commonly attracted to light, and the use of luminous attractant for collecting horse flies usually attracts much more males than females,  (Frost 1951, Anthony 1960, Philip 1982, Fairchild 1986, Henriques and Rafael 1999. Taking into account the large number of interception trap types employed during the Mitaraka (with only one operational light trap), female specimens were dominant in the samples as expected, accounting for 233 specimens, mostly collected by interception trap types, such as the 6m long Malaise trap (n = 98), SLAMs (n = 48), and flight intercept traps (n = 44). On the other hand, 19 of the 22 males were collected at the light trap, although, curiously, the trap collected more females than males (39 females vs 19 males) (see Table 1).