The current status of the Lutzomyia longipalpis (Diptera: Psychodidae: Phlebotominae) species complex

Lutzomyia longipalpis s.l. is a complex of sibling species and is the principal vector of American visceral leishmaniasis. The present review summarises the diversity of efforts that have been undertaken to elucidate the number of unnamed species in this species complex and the phylogenetic relationships among them. A wide variety of evidence, including chemical, behavioral and molecular traits, suggests very recent speciation events and complex population structure in this group. Although significant advances have been achieved to date, differential vector capacity and the correlation between structure of parasite and vector populations have yet to be elucidated. Furthermore, increased knowledge about recent epidemiological changes, such as urbanisation, is essential for pursuing effective strategies for sandfly control in the New World.

classifying and grouping the sandflies of the New World (Galati 2003). According to Barretto (1962), the American species of the subfamily Phlebotominae included the genera Warileya, Brumptomyia and Lutzomyia, the latter divided into fifteen subgenera, among them Lutzomyia. Young and Duncan (1994), reviewed the genus Lutzomyia, where they maintained the genus, but created the subgenera Coromyia, Psathyromyia and Sciopemyia.
The following year, a classification of the American species with phylogenetic approach was proposed, grouping and regrouping several species, however, the genus status of Lutzomyia is maintained, of which Lutzomyia longipalpis is included (Galati 1995(Galati , 2003. Due to its widespread distribution, early doubts arose about Lu. longipalpis Lutz and Neiva (1912) being a single species.
Lu. longipalpis species complex -The first evidence of morphological differences between populations of Lu. longipalpis s.l. was recorded by Mangabeira Filho (1969) studying Brazilian sandflies. Male sandflies collected in Pará state (PA) (North region of Brazil) had one pair of pale tergal spots on abdominal tergite IV (the one-spot phenotype named '1S'), while the males from Ceará state (CE) (Northeast region of Brazil) had two pairs of spots (the two-spot phenotype named '2S'), one on tergite IV and another on tergite III. Additionally, Mangabeira observed ecological differences between the sandflies of these two collection sites and suggested the existence of different species or varieties. Later, the observation of high-frequencies of intermediate phenotypes (a pair of pale spots with a smaller spot on the tergite III) indicated that this character is actually an intraspecific polymorphism (Ward et al. 1988) (Fig. 1).
Fourteen years after the first recognition of the spot phenotypes, Ward et al. (1983) obtained concrete evidence to support Mangabeira's hypothesis after carrying out crossing experiments with Brazilian populations of Lu. longipalpis s.l. from Marajó Island (PA / phenotype 1S), Sobral (CE / 1S and 2S phenotypes) Morada Nova (CE / phenotype 2S) and Lapinha Cave (MG / phenotype 1S). The failure of insemination between allopatric populations with similar tergal spot patterns and between sympatric populations with two dissimilar phenotypes (1S and 2S) strongly indicated the existence of additional forms in an apparent species complex.
Interest in the taxonomic status of Lu. longipalpis s.l. increased in the subsequent years (reviewed by Uribe 1999, Bauzer et al. 2007). Analyses involving populations from several countries of Latin America strongly supported the species complex hypothesis. Different approaches were used, both alone and integrated, and all pointed to the existence of a Lu. longipalpis species complex. Such analyses included isoenzyme electrophoresis (Morrison et al. 1995, Lanzaro et al. 1998, Mutebi et al. 2002, assessment of the genetic polymorphism of vasodilator peptide maxadilam DNA (Warburg et al. 1994) and mRNA (Yin et al. 2000), cytogenetics (Yin et al. 1999), measurement of nucleotide variation in the NADH dehydrogenase subunit 4 -ND4 (Soto et al. 2001) and cytochrome c oxidase I -COI (Arrivillaga et al. 2002) mitochondrial genes. Variation at microsatellite loci was found to be related to male pheromone type (Watts et al. 2005), and isoenzyme electrophoresis was combined with crossing experiments (Lanzaro et al. 1993), wing morphometry (Dujardin et al. 1997) and single strand conformation polymorphism analysis of COI, 12S and 16S rRNA genes (Arrivillaga et al. 2003), and all supported the species complex hypothesis. Given all of this evidence, there was no more doubting the existence of a Lu. longipalpis species complex that is distributed over a broad area spanning the Neotropic region (Table).
The first evidence of the existence of the Lu. longipalpis species complex was obtained in Brazil, yet initial studies using populations of sandflies collected in this country resulted in conflicting findings. A group of studies, mainly using isoenzyme electrophoresis, supported the single species hypothesis (Mukhopadhyay et al. 1997, 1998a, b, Mutebi et al. 1999, de Azevedo et al. 2000, Arrivillaga et al. 2003, Hodgkinson et al. 2003, Balbino et al. 2006). However, a number of them also identified some degree of genetic structure consistent with intraspecific variation (Mukhopadhyay et al. 1997, 1998a, b, Mutebi et al. 1999, de Azevedo et al. 2000, Hodgkinson et al. 2003. Isoenzyme electrophoresis has become an informational approach for distinguishing species when comparing populations that are quite different. For example, studies with Venezuelan populations showed strong evidence for the species complex hypothesis and suggested greater genetic structuring than the Brazilian studies (Lampo et al. 1999, Arrivillaga et al. 2000. Moreover, additional evidence from morphometric characters has allowed the formal recognition in Venezuela of Lu. pseudolongipalpis as the first species of the Lu. longipalpis species complex (Arrivillaga & Feliciangeli 2001).
There are a large number of studies in Brazil that strongly support the species complex hypothesis. One of the earliest, and most conclusive, studies was the crossing experiments carried out by Ward et al. (1983), mentioned previously. The efforts of Richard Ward and collaborators in studying this species complex continued for several years. They showed the existence of reproductive isolation between Brazilian populations and an association between insemination rate and specific male pheromones (Ward et al. , 1988. In addition, it became apparent that the spot phenotype could not be used to identify cryptic species in all locations. A decade later, Souza et al. (2008) carried out crosses among populations from Natal (Rio Grande do Norte state -RN), Jacobina (Bahia state -BA), Lapinha (MG) and Sobral (CE) and confirmed the association previously described by Ward et al. (1988) (Fig. 2).   Arrivillaga et al. (2000); 24 Arrivillaga and Feliciangeli (2001) Spiegel et al. (2016). #: 18 additional nuclear markers analysed by Araki et al. (2013) (CG9297, CG9769, eno, kinC, mlcc, norpA, obp19a, rpL17A, rpL36, rpS19, sesB, slh, sec22, sod2, tfIIAL, tropC, up, ζcop).
Analysis of the pale abdominal spots by scanning electron microscopy showed the presence of cuticular papules with central pores suggesting that they are sites of pheromone release (Lane & Ward 1984, Spiegel et al. 2002. Later, Morton and Ward (1989) demonstrated the attraction of females to tergal gland extracts, further indicating the aggregation-sex pheromone function of these compounds. From the chemical point of view, pheromones are comprised of a main and several minor components, which are responsible for attracting the female and pheromone enhancement, respectively (Hamilton et al. 1994). The analysis of different chemical compounds obtained from different populations of Lu. longipalpis s.l. was mainly based on the major components identified as homofarnesene (C 16 H 26 ) and diterpenoids (C 20 H 32 ) (Lane et al. 1985, Phyllips et al. 1986). Presently two types of homofarmasene are known, chemotype 1 or (S)-9-methyl-germacrene-B (9MGB) and chemotype 2 or (1S,3S,7R)-3-methyl-α-himachalene (3MαH); and two types of diterpenoids, the chemotype 3 or Cembrene-1 (Cemb-1) and chemotype 4 or Cembrene-2 (Cemb-2). A fifth chemotype, the chemotype 5 or 9-methyl-germacrene-B + (9MGB + ), was also identified as a mixture of compounds with a higher proportion of 9MGB (Brazil & Hamilton 2002, Hamilton et al. 2004. A current and comprehensive review of aggregation-sex pheromones of Lu. longipalpis s.l. shows that 9MGB is the most predominant pheromone-type in Latin America, and is also found in Lu. cruzi from Brazil and Bolivia. The pheromone type 3MαH is more restricted, having only been observed in the eastern region of BA (Northeast region of Brazil) and described in Lu. pseudolongipalpis from La Rinconada and El Paso (Venezuela). Of the diterpenoids, Cemb-1 has been founded only in the Southeast, Midwest, Northeast and North regions of Brazil, and Cemb-2 was only detected in Jaíba, a locality in northern MG (Southeast region of Brazil) (Table) (reviewed by Spiegel et al. 2016). Pheromones are complex multifaceted signals that can have different functions, such as the recognition of individuals of the same species or recognition of a partner for mating or mate assessment (Johansson & Jones 2007, Steiger & Stökl 2014, and represent an interesting trait for studying the evolution of a species complex. Behavior and courtship song -Towards the end of the 1980's, Ward et al. (1988) observed that male and female Lu. longipalpis s.l. produce sounds by wing movement. This wing-flapping could be observed during aggression between males, and during courtship and mating between males and females. Moreover, auditory signaling was described for the first time in two samples, Sobral 1S and Sobral 2S, which differed in burst repetition rates and intraburst frequencies of pre-copulatory songs, and thus raised all kinds of questions about the relationships between these signals and reproductive isolation in Lu. longipalpis s.l. . More recently, the full sequence of pre-mating behaviors has been described (Bray & Hamilton 2007). Regarding courtship behaviors, the approach-flapping and semi-circling performed by males and the station-ary-flapping of females were found to be predictors of eventual copulation. Interestingly, during copulation, females remained stationary whereas males vibrated their wings producing a species-specific song.
At the beginning of the 2000's, Alexandre Peixoto and collaborators initiated studies of song patterns emitted during the copulations and demonstrated that this trait can identify incipient species within the Lu. longipalpis species complex (Souza et al. 2002. The effective insemination of females seems to depend on the patterns of these songs, and can explain the reproductive isolation observed previously by Ward et al. (1983Ward et al. ( , 1988. Males of Lu. longipalpis s.l. produce two different copulatory courtship songs called primary and secondary songs (Souza et al. 2002. The primary song varies and, at present, three main types have been found in the Lu. longipalpis species complex: Burst-type, Pulsetype and Mix-type , Araki et al. 2009). The Burst-type song is composed of trains with highly polycyclic pulses modulated in frequency and amplitude. The Pulse-type song is more variable and five different patterns (subtypes P1 to P5) have been identified from among Brazilian populations. Finally, the Mix-type song has a pattern that is a mixture between Burst-and Pulsetype songs, and to date has only been detected in Mesquita (Rio de Janeiro state -RJ). More recently, Vigoder et al. (2015) carried out a more geographically comprehensive analysis and corroborated the five distinct patterns of Pulse-type songs with geographical separation and no overlap among their distributions. The group of Bursttype populations had a more widespread distribution spanning the five eco-regions of Brazil. Interestingly, sympatric coexistence of the Pulse-type and Burst-type populations occur in at least four localities: Sobral, Estrela de Alagoas (Alagoas state -AL), Jaíba and Palmas (Tocantins state -TO). The recognition of male aggregationsex pheromones by conspecific females, as mentioned previously, and cryptic female auditory choice during copulation seem to be critical for pre-zygotic reproductive isolation among sibling species of Lu. longipalpis s.l. (Maingon et al. 2008a, Vigoder et al. 2013. Molecular evidence -The absence of diagnostic morphological characters combined with evidence obtained from other sources of data have stimulated the implementation of approaches (Table). Beginning in the early 2000's, Alexandre Peixoto and collaborators started studying population genetics with nuclear markers in order to clarify the taxonomic status of Brazilian Lu. longipapis s.l.. Independently, polymorphisms of the loci period (per), cacophony (cac) and paralytic (para) were examined and found to strongly support the existence of the Brazilian species complex (Bauzer et al. 2002a, b, Bottecchia et al. 2004, Lins et al. 2008). In Drosophila, these genes have roles in generating courtship songs and represent interesting options for studying species complexes. In addition, the correlated evidence obtained from different approaches has been adequate in addressing the species complex question (Costa & Stanewsky 2013). Male copulation song data along with per gene polymorphisms , Vigoder et al. 2010, or with para gene variation (Lins et al. 2012), have resulted in even more robust evidence. Moreover, correlations between the distribution of allele frequencies of microsatellite loci and male aggregation-sex pheromones-types (Maingon et al. 2003, Watts et al. 2005, and per gene variation data combined with copulation song patterns (Vigoder et al. 2010), allowed the recognition of Lu. cruzi Mangabeira, 1938, as another sibling species within the Lu. longipalpis complex (Watts et al. 2005, Vigoder et al. 2010). In the same way, sandflies from Posadas (Misiones state, Argentina) might represent yet another sibling species, different from those found in the Northeast and Southeast regions of Brazil (Salomón et al. 2010).
An integrative analysis using a combination of biochemical, behavioral and molecular traits (Araki et al. 2009) strongly supports the hypothesis of two main groups within the Lu. longipalpis complex in Brazil. One group is a genetically homogeneous species whose males produce the Burst-type copulation song and the Cemb-1 pheromone (Cemb-1/Burst). The other group is genetically heterogeneous and probably represents a number of sibling species with different levels of divergence. Males of this latter group produce different subtypes of the Pulsetype copulation song (P1 to P5) in combination with different sex pheromones (9MGB, 9MGB + , 3MαH, Cemb-1 and Cemb-2). More recently, para gene variation was found in agreement with the two-group hypothesis (Lins et al. 2012). Moreover, this molecular marker showed diagnostic fixed polymorphisms, which can be used as a reliable indicator of two species. In addition, comparisons of life cycles between siblings species showed that populations from the second more heterogeneous group, such as from Jacobina (3MαH/P1), Lapinha (9MGB/P2) and Sobral 1S (9MGB + /P3), more easily adapt to the conditions of laboratory than do populations from Natal and Sobral 2S, which belong to the Cemb-1/Burst group. These phenological differences are a further indication of the differentiation between two main groups of the Lu. longipalpis species complex ).
When studying a species complex, the existence of two putative species in sympatry is one of the strongest pieces of evidence that they are indeed distinct. In Brazil, this scenario has been observed in at least four localities, as mentioned previously (reviewed by Vigoder et al. 2015, Spiegel et al. 2016. At these localities, males can be distinguished by the number of abdominal pale spots, which is supported by molecular analysis, and so these two phenotypes are considered to be two sympatric species at Sobral (Bauzer et al. 2002b, Bottecchia et al. 2004, Watts et al. 2005, Lins et al. 2008, Araki et al. 2013, Estrela de Alagoas and Jaíba (Araki et al. 2009, Lins et al. 2012. It is expected that future molecular analysis with samples from Palmas and Porto Nacional will also show differentiation at the molecular level. Incongruent evidence shown by some molecular markers (e.g., variable levels of divergence and phylogenetic relationships) could be due to different rates of evolution, introgression between counterparts, or the relative brief time of divergence among members of this species complex, and could explain the conflicting interpretations among early studies of Brazilian popula-tions. For example, the per gene was considered a useful molecular marker in studies of population genetics, and even more so considering the additional evidence from pheromones and copulation song analysis (Bauzer et al. 2002a, b, Araki et al. 2009) and the fixed polymorphisms detected in nearby populations in Northeast Brazil (Lima-Costa Jr et al. 2015). The published per data were reanalysed along with sequences deposited in Genbank in 2004 by Meneses and collaborators (unpublished observations) using different phylogenetic methods and found low bootstrap support and numerous polytomies (Golczer & Arrivillaga 2010). These findings are compatible with rapidly evolving markers, and indicates multiple speciation events and, further, recombination and introgression (Araki et al. 2013). On the other hand, mitochondrial markers are very commonly used for systematics because of their slow evolutionary rate and low recombination, but they also present some restrictions. Some studies questioned the use of mtDNA alone to explore phylogenetic relationships between closely related taxa, especially in cases with introgression (Hurst & Jiggins 2005, Galtier et al. 2009). More recently developed barcode analysis does not seem to be suitable for species recognition in Lu. longipalpis species complex due to introgression, but is more promising for higher taxonomic levels (Pinto et al. 2015).
A multi-locus approach was undertaken to estimate and compare levels of divergence and gene flow for 21 nuclear loci (including cac, para and per) between the sympatric siblings from Sobral (1S: 9MGB + /P3 and 2S: Cemb-1/B) and two allopatric species from the localities of Lapinha (9MGB/P2) and Pancas (Cemb-1/B) in Southeast Brazil (Araki et al. 2013). The nuclear data fit the isolation with migration model of speciation and reveals that introgressive hybridisation has played a crucial role in speciation of the lineages Cemb-1/Burst and 9MGB/Pulse (P2 and P3), which occurred in allopatry at around 0.5 MYA (Fig. 3). Following secondary contact and another period of hybridisation, reinforcement of reproductive isolation might have promoted the evolution of more efficient mate discrimination, such as the recognition of conspecific male aggregation-sex pheromones and copulation songs, and/or other isolation mechanisms (Machado et al. 2007, Servedio 2004. Perhaps differences in life cycle traits ) and patterns of locomotor activity (Rivas et al. 2008) are the results of divergence process of the two sympatric siblings.
Epidemiology -The sandfly Lu. longipalpis s.l. is the most important Neotropical vector of Leishmania (Leishmania) infantum Nicolle 1908, the causative agent of American visceral leishmaniasis (AVL). Formerly AVL was associated with rural and peri-urban areas, but more recently dispersion and urbanisation has been the most relevant epidemiological change observed in Brazil, Paraguay and Argentina (Salomón et al. 2015). In Brazil, AVL used to occur mainly in the Northeast region (Romero & Boelaert 2010), but has since spread to urban centers in the Central-West and Southeast regions. In the last three decades, the disease has begun to move into urban areas and the pattern observed suggests minor active dispersion activity by sandflies but a more significant passive component of dispersal, such as the transportation of soil from rural regions to cities (Brazil 2013a). In contrast, in Venezuela and Colombia, AVL still occurs mainly in rural areas, and no increases in the frequency of urban cases has been observed (Salomón et al. 2015).
Besides AVL, L. infantum causes atypical American cutaneous leishmaniasis (ACL) in Central and South America. This clinical pleomorphism might be due to sandfly genetic variability, as well as the genetic variability of Leishmania species, host susceptibility and immune status, and/or environmental factors. It is likely that Leishmania transmission, virulence and clinical outcome are influenced by coevolutionary interactions between specific Leishmania and specific sandfly genotypes (Maingon et al. 2008a). A comprehensive study of the population structure of L. infantum in the New World was carried out using several microsatellite loci and at least three main populations were identified (Kuhls et al. 2011, Ferreira et al. 2012. The existence of a link between these recently identified Leishmania groups and the species of the Lu. longipalpis species complex remains to be elucidated. Lu. longipalpis s.l. is a highly anthropophilic species, and sick dogs and foxes, reservoirs of L. infantum, have often been found naturally infected (Deane 1956, Lainson & Shaw 1979, Ryan et al. 1984. These reports stimulated a great need to demonstrate transmission by the bite of Lu. longipalpis under experimental conditions. Although this sandfly has been shown to be more likely to establish colonies in the laboratory, the parasitehost relationship is still not fully elucidated, however,  there is some evidence. In the early 1960's, Sherlock and Sherlock (1961) were conducting studies on experimental infection of Lu. longipalpis from Fortaleza (CE) and reported variation in the ability of this sandfly to infect and transmit L. infantum in different areas of Brazil. The first successful experimental transmission was that of Lainson et al. (1977), who demonstrated the transmission of the parasite to a hamster through the bite of Lu. longipalpis from Morada Nova reared in laboratory, although nothing was mentioned about differential capabilities. In a more recent study, Warburg et al. (1994) suggested that components in the saliva of the vector may play a role in inducing the impairment of liver and spleen and not the parasite. Since L. infantum transmitted by sandflies usually causes AVL in Brazil and Colombia, while infections in Central America usually result in skin lesions, the authors claim that maxadilan is more potent in insects found in Brazil and Colombia than in Costa Rica. They were able to demonstrate that sandflies in Costa Rica are vectors of ACL because the parasites remain in the skin due to very low vasodilator activity with little effect from the maxadilan in their saliva, thus leading to the cutaneous form of the disease. The sandflies in Brazil and Colombia have a great amount of maxadilan, which exarcerbates even a minor skin infection, allowing the parasites to invade even the liver and spleen, leading to visceral leishmaniasis. These findings led the authors to suggest that Lu. longipalpis is a complex species that may modulate the pathology of the disease they transmit depending on the amount of maxadilan. On the other hand, a study by Maingon et al. (2008b) has led to speculation about the association between environmental factors and host response to vector-transmitted parasitic disease. In Honduras it has been reported that ACL and AVL are caused by apparently genetically identical L. infantum (Noyes et al. 1997), and that inorganic parti-cles of volcanic origin accumulated in the salivary gland might have an immunomodulatory effect and alter the virulence of Leishmania (Maignon et al. 2008b). More recently, Casanova et al. (2006) reported that Lu. longipalpis from Araçatuba and Espírito Santo do Pinhal (SP, Brazil) produced different aggregation-sex pheromones, 9MGB and Cemb-1, respectively. This observation, coupled with the remarkable difference between the epidemiological frameworks, suggests an indirect and different vectorial capacity. It is worth emphasizing that experimental comparisons of infections by Lu. longipalpis of the two main pheromone/song types with L. infantum remains still a matter in need of special attention. In particular, such comparisons would be important in areas of sympatry such as Sobral, Estrela de Alagoas, Palmas and Porto Nacional.
Concluding remarks -Since its first description as Ph. longipalpis by Lutz and Neiva in 1912, the systematics of Lu. longipalpis s.l., has undergone revisions with the continual acquisition of new knowledge. Presently, the existence of a Lu. longipalpis species complex is accepted and has raised the prospect of assigning valid taxonomic names to its included species (Brandão-Filho et al. 2009). Although a few morphologic studies have shown differences among some populations (de la Riva et al. 2001, Santos et al. 2015, no discrete anatomical attribute has proven to be reliably diagnostic and extensively employed. The exact number of sibling species in the Lu. longipalpis species complex remains unclear, but at least seven different species have been suggested in Brazil alone (Araki et al. 2009), and additional species certainly exist according to more recent data (Table, Fig .  4). To better understand the interesting radiation of this group, a research strategy of combining approaches will probably prove productive in demonstrating how many species are in the Lu. longipalpis complex and the relationships and divergences among them. The advancement of next generation sequencing technologies provides an opportunity to explore molecular variation on a larger scale, which may lead to better understand of the molecular evolution of this interesting group. Further analysis throughout the genome is needed to better understand whether loci related to vectorial capacity can influence the transmission dynamics of Leishmania parasites by the different Lu. longipalpis sibling species. Furthermore, it would be interesting to investigate whether a particular population of Leishmania can be correlated with different species of this complex, as well as possible relationships with clinical pleomorphism.
The knowledge of chemical communication in Lu. longipalpis s.l. has advanced remarkably (reviewed by Spiegel et al. 2016), and is contributing to an alternative strategy for the control of this sandfly (Brazil 2013b). The use of synthetic (S)-9-methylgermacrene-B and the analogue (+/-)-9-methylgermacrene have shown to be useful in disrupting mating because females are highly attracted to these compounds (Hamilton 2008, Bray et al. 2010. Moreover, the attractiveness of the synthetic sex pheromones to males avoids the formation of lek aggregations, which would be helpful for sandfly population management (Vanessa Barbosa, personal communication). The use of this approach represents an interesting alternative strategy for vector control programs. Insecticide resistance of Lu. longipalpis s.l. has not yet been fully studied, however, there are some indications of its occurrence (Coutinho-Abreu et al. 2007, Alexander et al. 2009). The differential and reduced susceptibilities assessed among sandflies from the localities of Lapinha and Morada Nova (Alexander et al. 2009) indicate the need to take into consideration the pattern of insecticide resistance among siblings species of Lu. longipalpis s.l. in control strategies in Brazil and in other countries endemic for AVL.
The wide variety of evidence, including chemical, behavioral and molecular traits, suggests very recent speciation and complex population structure in the Lu. longipalpis species complex. Extending studies to other populations will give us a better sense of the geographical distribution of the sibling species of Lu. longipalpis and clarify their particularities, especially relative to their potential implication in incidence of AVL. Although significant advances have been achieved to date, differential vectorial capacity and the correlation between genetic structure of parasite and vectors populations remain to be elucidated. Furthermore, increased knowledge regarding recent epidemiological changes, such as urbanisation, is essential for pursuing effective strategies for sandfly control in the New World.