The leaf-miner flies (Diptera: Agromyzidae) of Mitaraka, French Guiana

ABSTRACT Preliminary data on the agromyzid fauna (Diptera: Agromyzidae) collected during the “Our Planet Reviewed” expedition in Mitaraka (French Guiana) is provided. A total of 138 specimens representing 10 genera and 50 morphospecies are recorded. Melanagromyza Hendel, 1920 and Ophiomyia Braschnikov, 1897 were the most diverse genera with 15 and 14 species respectively, followed by Phytobia Lioy, 1864 (five species), Japanagromyza Sasakawa, 1958 and Calycomyza Hendel, 1931 (four species each), Liriomyza Mik, 1894 (three species), Nemorimyza Frey, 1946 (two species), and Agromyza Fallén, 1810, Cerodontha Rondani, 1861, and Phytoliriomyza Hendel, 1931 (one species each). Except for Liriomyza and Nemorimyza, all of these genera are recorded from French Guiana for the first time. Nearly 90% of the specimens were collected with a 6 m long Malaise trap installed for eight days on a rocky outcrop (site “savane roche 2”) during the third phase of the expedition in August 2015. Prior to this expedition, only four species of Agromyzidae Fallén, 1823 were reported from French Guiana.


INTRODUcTION
Approximately 18 000 insect species have been described from French Guiana, although about 100 000 species are estimated to occur there. About 75% of these described species belong to the order Lepidoptera Linnaeus, 1758 (moths and butterflies) and coleoptera Linnaeus, 1758 (beetles) while Diptera Linnaeus, 1758 with only 577 species and two subspecies is listed among the most poorly known orders of French Guiana insects (Brûlé & Touroult 2014). The fly family Agromyzidae Fallén, 1823, commonly known as leaf-miner flies, includes nearly 3200 species worldwide (von Tschirnhaus 2021) and has been relatively well studied in some regions of the Neotropics with nearly 500 species recorded. The fauna of French Guiana, on the contrary, remains almost completely unknown with only four species previously recorded: Liriomyza huidobrensis (Blanchard, 1926); L. trifolii (Burgess, 1880); L. sativae Blanchard, 1938 andNemorimyza maculosa (Malloch, 1913) (EPPO 2020;Martinez & Etienne 2002). These species are well known agromyzid pests and their presence in French Guiana has been reported by various agricultural or plant protection organisations (e.g. EPPO), rather than from biodiversity surveys.
During 2015, an expedition to the Mitaraka massif, a largely unexplored region of French Guiana, was organized to assemble information on the biodiversity of the area, with a focus on freshwater and terrestrial invertebrate biodiversity (Biotope 2014;Pascal et al. 2015). The expedition was part of a large-scale biodiversity program known as "Our planet Reviewed" (La Planète revisitée) led by the Muséum national d'Histoire naturelle (MNHN, France) and Pro-Natura International (Pollet et al. 2014; Muséum national d'Histoire naturelle 2016). The main objective of this program is to accelerate the discovery and description of new species by concentrating efforts in regions considered a priority for nature conservation (Muséum national d'Histoire naturelle 2016).
The Mitaraka massif is situated in the southwesternmost part of French Guiana directly bordering Surinam and Brazil ( Fig. 1). It is part of the Guiana Amazonian Park, which forms one of the largest protected rain forest areas in the world, together with the adjacent Tumucumaque National Park in Brazil (Shelley 2013). The Mitaraka massif is a completely remote, uninhabited region consisting mainly of lowland tropical rain forest with scattered isolated rocky outcrops of Precambrian granite known as inselbergs (Fig. 2) (Touroult et al. 2018). Although these inselbergs are sometimes described as being part of a larger mountain range, known as Tumuc-Humac, the truth is that this mountain range does not exist, at least not as it was visualized by French explorers. Indeed, these "mountains" are no more than isolated inselbergs varying in size from 100 to 800 m (Hurault 1973(Hurault , 2000chavance 2015).
Inselbergs ("Island mountains" in German) have recently received increased interest for biodiversity research as these "islands" are known to serve as refuges for unique flora and fauna (Vitt 1993;Porembski & Barthlott 2000;Vlasáková et al. 2008;Jocque & Giupponi 2012). The unique diversity and structure of their plant communities have been studied in both tropical and temperate regions (Porembski & Barthlott 2000;Sarthou et al. 2003;Porembski 2007). contrary to the rich flora of the surrounding moist forest, inselbergs are considered "micro-environmental deserts" (Phillips 1982) where vegetation grows on exposed granite rock with very little or no soil cover that is often exposed to extreme weather conditions. In French Guiana, the characteristic vegetation of inselbergs is called "savane-roche" (Gasc et al. 1998;Sarthou et al. 2003). More than 600 plant species have been reported from these savane-roche vegetations. Most of these species belong to the following plant families: Orchidaceae Juss., Rubiaceae Juss., Melastomataceae Juss., Poaceae Barnhart, cyperaceae Juss., Bromeliaceae Juss., Araceae Juss., Myrtaceae Juss., and Euphorbiaceae Juss. (Gasc et al. 1998), with some variation from one inselberg to the other (Sarthou et al. 2003). In French Guiana, inselbergs are located in undisturbed rain forest and their flora includes some species that are rare and restricted to these rocky outcrops in this part of the Neotropics (Sarthou et al. 2003 The objective of this paper is to document on the diversity of agromyzid species collected during the Mitaraka expedition in a southern lowland rainforest with special attention to the fauna of a minor inselberg and to provide a first insight into the composition of French Guiana Agromyzidae.

ReseaRch team and timefRame
Due to the dense vegetation and remote aspect of the Mitaraka massif, a landing area (the drop zone) was cleared by the French army prior to the arrival of the research crew by helicopter. A base camp was also established for the duration of the expedition. Three different teams of researchers were involved during three different sampling periods in 2015: February 23 to March 11, March 11-27 and August 12-20 (Touroult et al. 2018). The main mission was conducted during the rainy season (February-March 2015) by 32 researchers, each studying a specific taxonomical group, e.g. higher plants, earthworms, birds, fish, or particular insect taxa. A third shorter visit by a smaller team of 10 researchers to the site during the beginning of the dry season focused on insects only, and was organized  (Table 1). A wide array of methods was used to collect invertebrates (Pascal et al. 2015;Touroult et al. 2018) and multiple traps of particular types were employed e.g. 280 pan traps and 33 SLAM traps, but not all trapping methods were used during all three sampling periods of the expedition (e.   (Table 1). Further details on sampling methods can be found in Touroult et al. (2018). specimen pRepaRation and identification All the specimens collected benefited from the Access and benefit sharing agreement of the Our Planet Reviewed program (APA 973-1). A total of 223 invertebrate samples were examined, with non-pan trap samples being sorted to insect orders and families at the SEAG office, and pan trap samples sorted by MP. This process produced nearly 2 200 subsamples, containing 28 Diptera families and superfamilies ) that were disseminated among experts worldwide, including the six samples of Agromyzidae (Table 1) sent to SB. The Agromyzidae samples were initially stored in 70% ethanol and subsequently dried using hexamethyldisilazane (HMDS), mounted and labelled (see Table 1). Morphospecies identifications were based on external morphological characters and male genitalic characters. Male genitalia were examined by detaching the abdomen from the remainder of the specimen, clearing it in 85% lactic acid heated in a microwave oven for 2-3 intervals of 30 seconds each, separated by a cooling period, or cleared outside the microwave (later in the project) following procedures outlined in Boucher (2019). The abdo-men was subsequently transferred to a glycerin solution for further dissection and examination and was ultimately stored in a microvial pinned below the corresponding specimen. The format of the morphospecies names used here are: genus name + species number, preceded by the prefix Mit (for Mitaraka).
Six species were collected during the first sampling period, two species during the second period, and most species (43 species out of 50) during the last period of the expedition (Fig. 6), with very little or no species overlap between the different periods (Fig. 9).
Only two species were represented in more than one sample: Japanagromyza Mit-1 was retrieved from samples 229 and 230, and Melanagromyza Mit-3 from samples 218, 229, and 230 (Table 2). Among the 50 species identified, 27 species were represented by females only and most species (38/50) were represented by only one or two specimens (Table 2).
Three collecting methods successfully caught Agromyzidae, but the 6 m long Malaise trap (MT(6M)) was by far the most successful with a total of 48 species (Fig. 8). Three agromyzid samples (186,189,230) were collected with this 6 m Malaise trap (Table 1) Fig. 8). The Malaise trap produced 46 unique species, while the SLAM and PVB traps collected one unique species each (Liriomyza Mit-3 and Ophiomyia Mit-14, respectively) (Fig. 8).

DIScUSSION
Given this relatively short survey (40 days total) and the low number of specimens examined (138 specimens), alpha diversity of Agromyzidae in Mitaraka appears quite high with a total of 50 species in 10 genera. Surprisingly, most of these agromyzids were collected with a Malaise trap set up for only eight days. A recent survey in a very small patch of cloud forest in Zurquí de Moravia, costa Rica, yielded 812 specimens of 117 species of Agromyzidae in 12 genera, but this was the result of a whole year of sampling (Brown et al. 2018).
With 67 specimens and 15 species, Melanagromyza was the most abundant and most diverse genus in Mitaraka. It is also the second largest genus in the Neotropical region with 93 described species. Its diversity is substantially lower in temperate regions of the world. As an example, only 37 species are reported from the USA (Shi & Gaimari 2015). With only three species, Liriomyza, the most speciose Neotropical agromyzid genus with 105 described species, was not well represented in Mitaraka. The other most diverse genera in the Neotropical region are Calycomyza (73 described species) and  Melanagromyza Hendel, 1920;Ophiomyia Braschnikov, 1897;Liriomyza Mik, 1894;Japanagromyza Sasakawa, 1958;Calycomyza Hendel, 1931;Nemorimyza Frey, 1946;Phytobia Lioy, 1864;Phytoliriomyza Hendel, 1931  Ophiomyia (50 species). The latter genus was represented in the Mitaraka samples with 14 species, the former with only four species.
The highest agromyzid specimen and species numbers were retrieved from a single sample (code 230), which was collected with a 6 m long Malaise trap on one of the minor inselbergs (site MIT-E-savane roche 2). This site was described as a rocky outcrop (471 m a.s.l.) covered with herbaceous vegetation (including Pitcairnia L'Hér.), scattered shrubs and exposed rocks (Touroult et al. 2018;Pollet et al. 2018) (Fig. 4). Pitcairnia is a speciose genus of Bromeliaceae present mainly in Mexico, central and South America and the caribbean (Saraiva et al. 2015), with some species known to be restricted to inselbergs (Sarthou et al. 2001). This plant genus has never been recorded as host for any Agromyzidae, and only one species of Agromyzidae is known to feed on Bromeliaceae, Melanagromyza rosales Woodley, a leaf-miner in Bromelia pinguin L. in costa Rica (Woodley & Janzen 1995). However, none of the Mitaraka Melanagromyza specimens correspond to this species. Only two species (Melanagromyza Mit-3 and Japanagromyza Mit-1) were present in more than one sample, with Melanagromyza Mit-3 collected in two distinct habitats (savane-roche and drop zone (cleared moist forest)). Surprisingly, other sites similar to the "MIT-E-savane roche 2" investigated at Mitaraka (including other minor rocky outcrops (390-470 m a.s.l.) and major rocky outcrops (540-570 m a.s.l.)), did not yield any Agromyzidae. This is most probably due to the fact that no 6 m Malaise trap was operational on these inselbergs, and that no agromyzids were actively searched for.
The success of the 6 m Malaise trap reflects previous knowledge that Malaise traps are an effective method for collecting flying insects, including flies (Lamarre et al. 2012;Borkent et al. 2018). But this method is also known to be female-biased for Agromyzidae (Scheirs et al. 1997), which concurs with the high proportion of agromyzid females collected during the expedition. As compared to the very productive Malaise trap on the minor inselberg in August 2015 (43 species, 123 specimens), the other habitat examined with this trap in March (samples 186, 189) proved much less species rich, with only five species and six specimens. Reasons for this difference might include: season, habitat type and host vegetation, position of the trap, or a combination of these factors. Among the other collecting methods used, the SLAM trap is not commonly applied for Agromyzidae, but when installed on land (as it was the case here) it works in a similar way as a Malaise trap, although being smaller (Touroult et al. 2018). This method collected few specimens and added only one unique species (Liriomyza Mit-3) (Fig. 8) in phase 1 of the project. The SLAM traps that were installed for nine days during the last sampling period (August) in a range of sites, including MIT-E-savane roche 2 (sample 229), were much less successful than the 6 m Malaise trap (sample 230) installed during the same period in the savane roche 2 site. The reasons for this difference remain unclear as this trap type was successful for collecting other Diptera families (Krolow et al. 2017;Pollet et al. 2018;Vieira et al. 2019). The PVB (automatic light trap with blue LED) also added one unique species (Ophiomyia Mit-14) (Fig. 8) represented by one female specimen. The automatic light traps were installed to collect nocturnal flying insects from various insect orders, but did not target Diptera (Touroult et al. 2018). Agromyzidae are known to be diurnal and are not normally collected by light traps so this specimen is possibly an accidental visitor to the trap. Despite being established as a successful device to collect Diptera, including Agromyzidae (Scheirs et al. 1997), the 280 pan traps used did not yield a single agromyzid in Mitaraka. This absence of Agromyzidae could be explained by the heavy rainfall which also affected yields of Dolichopodidae Latreille, 1809 . Although savane roche 2 and at least 29 other sites were investigated by MP using a sweep net during the first phase of the expedition, no Agromyzidae were gathered because an on sight collecting approach was used to mainly target Dolichopodidae. Sweeping was also performed during the last phase of the expedition (Touroult et al. 2018), but again this approach was unsuccessful for Agromyzidae, most probably for the same reason.
All species reported here are identified as morphospecies as a first step in the species recognition process. Species identification of small flies like the Agromyzidae (averaging 2-3 mm in wing length but sometimes as small as 1 mm) can be challenging, even for trained taxonomists. This is particularly true for species living in inadequately studied regions like French Guiana for which no reference work nor identification keys are available and where many undescribed species are expected. Although morphospecies identification is often performed by non-specialists (or "parataxonomists") for rapid biodiversity assessment (Derraik et al. 2002), it was here performed by a trained agromyzid taxonomist (SB) using elaborate external non-genitalic and genitalic characters when possible, which certainly adds to the accuracy of the morphospecies determination. Nevertheless, females often remain difficult to differentiate from each other or to assign to a male morphospecies.
Further taxonomic work is in progress and will allow the assignment of some of the morphospecies to described species, while others will be confirmed as new species and will be described. Some morphospecies will remain unidentified until further material is available, and this will include most of the 27 species represented by female specimens only.
Although additional work is required for species level identification, we can confirm that the agromyzid species collected during the Mitaraka expedition do not correspond  (6) conseil général de Guyane, the Direction de l'Environnement, de l'Aménagement et du Logement and by the Ministère de l'Éducation nationale, de l'Enseignement supérieur et de la Recherche. It was realized in collaboration with the Parc amazonien de Guyane and the Société entomologique Antilles-Guyane (SEAG). MP participated to this expedition as member of the first team (22 February-11 March 2015), hereby supported financially by MNHN and Pro-Natura international. We would also like to thank in particular Pierre-Henri Dalens and his team for collecting the major part of the agromyzid material during the August mission. Also Olivier Pascal (MNHN), Julien Touroult (MNHN), Stéphane Brûlé (SEAG), Xavier Desmier, Maël Dewynter and the Service géomatique du Parc amazonien de Guyane for providing pictures. Finally also many thanks are due to Owen Lonsdale and an anonymous reviewer for their useful comments on an early version of the manuscript.