Natural Leishmania (Viannia) infections of phlebotomines (Diptera: Psychodidae) indicate classical and alternative transmission cycles of American cutaneous leishmaniasis in the Guiana Shield, Brazil

From 1996 to 1999 multi-trapping methods (Center of Diseases Control, CDC) light traps, light-baited Shannon traps, and aspiration on tree bases) were used to study the phlebotomine fauna of the “Serra do Navio” region of the Brazilian State of Amapá, which is part of the Guiana Shield. Fifty-three species were identified among 8,685 captured individuals. The following species, associated with the transmission of American cutaneous leishmaniasis in Amazonian Brazil, were captured: Nyssomyia umbratilis (3,388), Psychodopygus squamiventris maripaensis (995), Ny. anduzei (550), Trichophoromyia ubiquitalis (400), Ny. whitmani (291), Ps. paraensis (116), and Bichromomyia flaviscutellata (50). Flagellate infections were detected in 45 flies. Of the 19 parasites isolated in vitro, 15 were Leishmania (Viannia) guyanensis (13 in Ny. umbratilis, 1 in Ny. whitmani, 1 in Ny. anduzei) and three were L. (V.) naiffi (2 in Ps. s. maripaensis, 1 in Ny. anduzei). The results indicate the participation of three phlebotomine species in the transmission of L. (V.) guyanensis and two species in that of L. (V.) naiffi, and show that the same phlebotomine species is involved in the transmission of different Leishmania (Viannia) species in the Guianan/Amazon region. A review of the literature together with the results of the present study, and other published and unpublished results, indicate that eight phlebotomine species potentially participate in the transmission of Leishmania (Viannia) naiffi in Amazonia.


Introduction
The Guiana Shield is a geological formation with various ecological areas within the Amazon biome of Venezuela (Delta Amacuro, Bolívar, and Amazonas States), Brazil (Northern Amapá, Pará, Roraima, and Amazonas States), Guyana, Suriname, and the Overseas Department of French Guiana. Its environmental conditions sustain some specific ecological niches and it is one of the regions with the highest biodiversity in the world [49]. Such characteristics favor an array of vectorreservoir relationships and consequently a mosaic of leishmanian ecosystems [55].
American cutaneous leishmaniasis (ACL) is endemic in the region and so far, five dermotropic coexisting Leishmania species have been found there: Leishmania (Viannia) guyanensis ]. These findings are of concern as they indicate ecological changes that favor the transmission of other parasites associated with debilitating forms of the disease, such as diffuse and mucocutaneous leishmaniasis.
Despite the high incidence of ACL in Amapá State (AP) [64], there is a paucity of ecological studies on its vectors and reservoirs. The state is bordered by northern Pará State to the west and French Guiana to the northwest, where the ecology of the disease has been studied more extensively [16,33,39,43,54,55]. There are only a few studies on the phlebotomines associated with ACL transmission in AP [5,15,17,18,46,58]. The aim of the present survey of the Serra do Navio phlebotomine population is in part to fill this gap by assessing putative transmission cycles in this Brazilian region of the Guiana Shield.

Study area
Serra do Navio (00°53 0 45 00 N; 52°00 0 07 00 W, 148 m a.s.l) is one of the 16 municipalities that constitute AP (Northern Brazil). It is in the central area of AP and is 146 km from the state capital, Macapá. Its area is approximately 7,757 km 2 and it has an estimated population of 4,761 [26] (Fig. 1). The climate is similar to those of the other Amazonian ecoregions of the Guiana Shield, as follows: a short rainy season from mid-November to late January; a short dry season between early February and mid-March; a long rainy season from late March to late July; and a long dry season from late July to mid-November.
Historically, this location is an area of mineral exploitation. In the 1950s the ''Sociedade Brasileira de Indústria e Comércio de Minérios de Ferro e Manganês'' (ICOMI), in association with Bethlehem Steel, began the manganese mining operation that came to a premature end in 1997. A mining complex was built together with a modern residential township. A health department was set up to look after the health of the mine workers. It was exceptionally efficient and well organized and its team of doctors dealt easily with most of the normal health problems. However, the greatest immediate risks to this isolated population were endemic diseases such as malaria and ACL. The careful use of chloroquinized salt successfully protected them from malaria but no similar measures were or are available against ACL. Today, Serra do Navio is a municipality of intense ACL transmission. During a 2002 epidemiological study, the ACL coefficient of infection was 1,476/100,000 inhabitants [8]. Capture sites were located in three areas of primary forest that were less than 10 km from urban areas. These undisturbed environments were considered to be ecologically similar and were therefore treated as a single site. These areas were predominantly covered by submontane (lowland), dense forest.

Field and laboratory procedures on phlebotomines
Captures were performed during five 12-night expeditions in 1996 (May and September), 1997 (July and November), and 1999 (October). Traps were installed as follows along transects in forested areas from the edge inwards: eight ''Center of Diseases Control'' (CDC) light traps set each night at 1.5 m above ground level (6) and at 20 m in the canopy (2), from 06:00 pm to 06:00 am; light-bait Shannon traps, between 06:00 pm and 08:00 pm; and captures on tree bases with a battery-operated aspirator from 07:00 am to 09:00 am. The number of capture hours was calculated by multiplying the number of traps set by the number of hours of exposure. After screening, all phlebotomine females were dissected under sterile conditions as described by Ryan et al. [56]. Males were stored in 70% alcohol. If there was any difficulty in identifying a dissected female, it was mounted in Berlese fluid (GBI Laboratories). Species were identified using Young and Duncan's guide [74] and the nomenclature adopted is in accordance with the taxonomic criteria proposed by Galati [21]. Two letter genus abbreviations are those suggested by Marcondes [42]. Species composition in the four different trapping samples was analyzed using the Shannon-Wiener diversity index (H') and dominance (D), with the aid of the statistical software Past version 2.12 [25].

Characterization of flagellate strains
The dissected female digestive tracts were examined microscopically for flagellates. Infections were classified according to their distribution in the intestine and semi-quantified using the parasitosis scale adopted by Freitas et al. [17]. After this, the intestine was homogenized and inoculated into two culture tubes containing Difco B 45 [72]. Isolates were initially characterized by the use of a panel of 23 monoclonal antibodies by a fluorescein-labeled avidin indirect immunofluorescence method (IIF-McAb reaction), as described by Shaw et al. [59]. Phenotypic characterization was undertaken by isoenzyme electrophoresis for 6 PGDH (E.C 1.1.1. 44), PGM (E.C 5.4.2.2), and MPI (E.C 5.3.1.8) as described by Miles et al. [45], and/or by polymerase chain reaction-restriction fragment length polymorphism analysis (PCR-RFLP) of a 615 bp amplified region of the RNA polymerase II gene digested with endonucleases TpsRI and HgaI, as per Simon et al. [63].
McAb, isoenzymatic, and PCR-RFLP profiles of the isolates were compared with those of the five World Health
The average of phlebotomine putative vector species captured per hour can be found in Table 2. Considering all capture methods, captures of Ny. umbratilis (0.64) were the greatest, followed by Ps. squamiventris maripaensis (0.19). However, when each trap was compared individually, Shannon captures were higher for Ps. s. maripaensis (11.5), Ny umbratilis (5.33), and Ny. anduzei (3.48).  There was no evidence of blood in the intestines of the 45 infected females and the parasitosis ranged from 21-40 (+++) to 41 or above (++++) flagellates per field (40· objective). From their morphology, we were unable to confirm that any of these trypanosomatids were Leishmania.
The highest richness of species was observed for Psathyromyia and Psychodopygus, among the 15 genera captured in Serra do Navio. The former also included four species from the Shannoni series Fairchild, 1955, whose taxonomic status has recently been revised [57]. Voucher specimens originally identified as Pa. shannoni (Dyar, 1929) (MGS Silva det.) were re-examined and after further consultation (AJ Andrade, personal communication), it was concluded that they in fact represent the resurrected Pa. bigeniculata (Floch & Abonnenc, 1941).
The role of different phlebotomine species in ACL epidemiology in the Guiana Shield contrasts with that of the Amazonian lowlands. In the latter region, Psychodopygus species appear to be the dominant vectors, whereas in the Guiana Shield, it is the Nyssomyia species that dominate.
As might be expected, the principal vector of L. (V.) guyanensis, Ny. umbratilis, stands out in the list of vectors, in accordance with several other studies in highly endemic  [33,50,51]. A part of the life cycle of Ny. umbratilis is arboreal, and this is highlighted by its dominant presence in our canopy traps. In an area affected by a hydroelectric dam project on the Jari River (AP), no differences were observed between ground level and canopy catches for this species. This could be because most arboreal animals, which are the principal blood source for this species, were removed during environmental management operations related to the destruction of the forest [18]. Unfortunately, present captures are logistically biased which often weakens conclusions on seasonality. However, we found that Ny. umbratilis, including infected individuals, was only taken from tree trunks at ground level during the dry season. Our data on the treedwelling behavior of Ny umbratilis are compatible with those on the well-studied populations of the Northern Amazon [39,54,73]. The morphologically closely related species found in smaller numbers, Ny. anduzei, behaves similarly and it has been suggested that it is a secondary L. (V.) guyanensis vector in ACL foci where Ny. umbratilis is present [53]. Ps. s. maripaensis has been associated with L. (V.) naiffi in northern Brazil [47] and French Guiana [16], and has been suggested as a possible vector of L. (V.) braziliensis and L. (V.) naiffi in Suriname [30]. Its high numbers in our Shannon captures suggest its anthropophilic behavior and preference for populations living at ground level. Catchers in the Shannon traps constantly found this species trying to bite them despite the use of protective measures. Two infections in Ps. s. maripaensis (Shannon and CDC canopy) were characterized as L. (V.) naiffi. Human infections with this parasite are unknown in AP, but these findings are proof that its enzootic is present in the Serra do Navio forest. Records also link its known distribution to this Leishmania species in the Brazilian states of Amazonas [23,47], Acre [61], Rondônia [2,22], Pará [35] and the belt that includes AP, French Guiana [55], and Suriname [69]. Interestingly, single records for L. (V.) naiffi have also been reported in Ecuador [29], Peru, Martinique (inconclusive?) [52], and Panama [3].
Leishmania DNA has been found in some Trichophoromyia species and their possible importance in ACL epidemiology has been discussed [30,44,67]. Th. ubiquitalis is the proven vector of L. (V.) lainsoni [53,62] and its potential importance in ACL epidemiology in AP is supported by the fact that L. (V.) lainsoni has been diagnosed in patients from this state (FT Silveira, personal observations).
In Pará State, Ny. whitmani is associated with transmission of L. (V.) shawi shawi [32] and latterly Lu. gomezi has been indicated as a putative vector of this same parasite [65]. However, Ny. whitmani has also been associated with the L. (V.) guyanensis enzootic in the Monte Dourado region of the Guiana Shield [33]. In the north-eastern and Atlantic forest region of Brazil, Ny. whitmani is an important vector of L. (V.) braziliensis [4,6,70]. In these environments, Ny. whitmani occurs predominantly in peri-domestic situations. Its sylvatic occurrence in the present study is in accordance with early studies [35,60], and may represent a typical Amazonian population also present in French Guiana [41] and Suriname [30]. One of the three infections found in Ny. whitmani was characterized as L. (V.) guyanensis. Lainson et al. [33] found five infections of L. (V.) guyanensis in Ny. whitmani from the Monte Dourado in northern Pará State. These and our results reignite the intriguing hypothesis of Lainson et al. [33] who suggested that Ny. whitmani may play a vector role together with  Ny. umbratilis in the transmission of L. (V.) guyanensis. Moreover, in AP, a single infection from Ny. umbratilis captured near Porto Grande was also compatible with L. (V.) guyanensis by morphology and behavior in hamsters [17]. Rangel and Lainson [53] suggested that flagellates found by Lainson et al. [33] in Ny. whitmani could have been L. (V.) shawi. Our characterizations by three different methods (IIF-McAb, isoenzyme electrophoresis, and PCR-RFLP) of the isolates from the three available Ny. whitmani collected in 1981, that Rangel and Lainson [53] referred to, were all confirmed to be L. (V.) guyanensis. From this, we conclude that it is very unlikely that L. (V.) shawi occurs in the Guiana Shield, as indicated in [55] and that Ny. whitmani in fact contributes to L. (V.) guyanensis transmission.
There is strong evidence that Ny. whitmani sensu lato is genetically complex [27], with regional differences in the behavior of distinct geographical populations [7]. Its status as a species complex is presently being discussed and it is quite possible that the Guiana Shield population represents a distinct genetic group. Recent results demonstrating the lower Amazon Absence of infections in this species in the present study could be misleading and explained by the fact that only 35 females were dissected. The small numbers in catches are perhaps related to this fly being less attracted to light traps [37]. However, in other situations, CDC sets installed at ground level have collected significant numbers of Bi. flaviscutellata [14,20].
The present rate of 1.43% of Leishmania-like infections in Ny. umbratilis, with 13 infections proven to be L. (V.) guyanensis, is in accordance with other surveys conducted in French Guiana with 1.3% [40] and Amazons State with 1.04% [51], but higher frequencies were reported in the former territory (15%) [39], as well as in a nearby area on the outskirts of Porto Grande (26%) [17]. These facts do not merely consolidate the role of this species as the main ACL vector in AP, but make us aware of a factual risk of Serra do Navio being a hotspot for ACL transmission. Reported ACL cases have been identified early as being caused by this parasite in the region [23]. In addition, two strains isolated from ACL patients (MHOM/BR/1996/M15781; MHOM/BR/1996/ M15937), likely infected in Serra do Navio, with diagnosis performed by our team at the time of field expeditions, were typed by the IIF-McAb technique as L. (V.) guyanensis and are identical to our Ny. umbratilis isolates (unpublished observations). The feasible classical mechanism of diurnal infection documented involves females resting on tree bases, and when humans come close to these ecotopes, they may disturb the insects that start aggressive biting behaviors [50].
Single L. (V.) guyanensis infection in Ny. anduzei is consistent with its apparently secondary participation in the transmission cycle of Guianan ACL. However, the first finding of this fly harboring L. (V.) naiffi raises the interesting hypothesis of an increasing demand for candidates for transmitting  (Table 4). ''Molecular-based'' vector speculations were not included in the former list due to their questionable ability to determine true infection, as recently argued by Brazil et al. [6]. The lack of Leishmania typing methods in AP means that our conclusions are biased toward the hypothesis that L. (V.) guyanensis is the principal etiological agent of ACL. However, it is reasonable to assume that ACL cases due to L. (V.) naiffi are underreported in AP as it is normally self-healing and there is no routine identification of ACL parasites. Different ecological situations may affect the incidence of this form of ACL, and in Amazonas State, two of eight L. (V.) naiffi cases did not respond to initial treatment [12]. Two individuals of Ny. pajoti from CDC ground were found infected with Leishmania-like flagellates, which were not successfully cultured. This species harbors an unidentified Leishmania species [10,31,74]. Its arboreal behavior coincides with those of Ny. umbratilis, Ny. anduzei, and Ny. whitmani, all of which have been shown to be infected with L. (V.) guyanensis. This leads us to speculate that the two unknown flagellates in Ny. pajoti were also L. (V.) guyanensis. These four fly species were all captured in ground level CDCs, increasing their potential role as vectors to humans during the night.
Despite medical evidence highlighting Mg. migonei as a vector of L. (V.) braziliensis in north-eastern and south-eastern Brazil [53], there are no signs of its implication in the context of leishmaniasis in the Amazon region. The isolate from this species was compatible with an unknown trypanosomatid which should be better characterized in the future. Moreover, another similar infection in Mg. migonei was recently found during an entomological study in Oiapoque, far north of AP, and although parasite isolation failed, DNA was obtained from the slide used for dissection. In both situations, the PCR-RFLP technique was negative, confirming that these parasites are distinct from Leishmania (TV Santos, M Ginouves, G Prévot, personal observation).
Failure of flagellate isolation occurred with Sc. sordellii and Lu. gomezi. Regarding Sc. sordellii, this is not of concern for the ecology of ACL because infection of this species is historically recognized as being caused by other non-Leishmania trypanosomatids [56], and even further PCR-based findings of Leishmania DNA within this fly [24,38] did not suggest a determinant role in the transmission of leishmaniasis. In the case of Lu. gomezi, however, its epidemiological relevance as a vector of L. (V.) panamensis outside the Brazilian Amazon [13,66,71] and PCR-based suspicion of it carrying L. (V.) naiffi in Panama [3] raise the hypothesis that it may be competent in harboring Leishmania species from the Guiana Shield. This speculation is supported by its early infection by unidentified flagellate in French Guiana [9], as well as by our findings of a single specimen that proved to be infected with L. (V.) shawi shawi in the lower Amazon region [65].
Trapping methods favor different species, resulting in a bias that portrays the enzootic and/or the zoonotic cycle. In the present study, CDC light trap catches were the highest, but they were set for a greater number of hours. When the captures were corrected for the number per/hour (Table 2), the numbers of Ny. umbratilis, Ps. s. maripaensis, and Ny. anduzei captured in the Shannon trap exceeded those of the CDCs. Even though we were unable to identify all the flagellates found in these three species, we consider that our present and past results add weight to their importance as ACL vectors.
The present results provide an update on the phlebotomine fauna inventory of AP and indicate putative ACL vectors for the region. L. (V.) guyanensis infections in Ny. umbratilis and Ny. anduzei confirmed them, respectively, as primary and secondary vectors. Our analysis of passed infections of Ny. whitmani suggests that this species may also participate in the transmission scenario. On the other hand, the circulation of L. (V.) naiffi in Ps. s. maripaensis, which is highly anthropophilic, raises the possibility of the occurrence of underreported ACL cases related to this parasite. The finding of L. (V.) naiffi in Ny. anduzei adds yet another vector to the long list (Table 4)