Association between the visceral leishmaniasis vector Lutzomyia longipalpis and tree families in a Brazilian tropical urban area

Background Visceral leishmaniasis (VL) is a neglected disease endemic to many tropical and subtropical countries. In Brazil, VL is caused by the protozoan parasite Leishmania infantum and is transmitted by the sand fly Lutzomyia lon-gipalpis. The geographical variation in the incidence of the disease in certain urban areas of Brazil suggests a possible association with features of the urban landscape. The objective of this study was to verify the association of urban tree families with the presence of the vector Lu. longipalpis. Methods The study area of the municipality of Teresina, capital of the State of Piauí. The sand flies were captured using CDC light traps—one placed close to the houses, on a tree in the backyard, less than 5 m from the house, and another in the living room. The traps remained three consecutive nights, or, at the most, alternating days, working 12 h a night (from 6 p.m. to 6 a.m.), during 2016 and 2017. Results Tree species were identified in an area of 100 m radius around CDC-type traps for capturing Lu. longipalpis installed in household backyards. Association between tree families and the occurrence of Lu. longipalpis was evaluated using the Fisher exact test and multiple correspondence analysis. Both Anacardiaceae (represented by mango and cashew trees) and Meliaceae (neem) families were more commonly found around sand fly traps capturing fewer Lu. longipalpis. Conversely, Fabaceae (several species of legumes) and Myrtaceae (guava) trees were associated with traps capturing higher number of VL vectors. In multiple correspondence analysis, the presence of trees from the Myrtaceae, Combretaceae (Almond), Fabaceae and Arecaceae (different species of palm trees) families were grouped with traps capturing high number of insects. Conclusions The findings suggest that trees from the Fabaceae, Myrtaceae, Combretaceae, and Arecaceae families might potentially attract sand flies, and Anacardiaceae and Meliaceae families may have a repellent effect. Finer analyses to distinguish the specific effects from the different species of the Anacardiaceae (mango and cashew), Arecaceae and Fabaceae families are still needed. The manipulation of urban landscaping might be a useful and clean alternative to help in the control of urban VL.


Background
Visceral leishmaniasis (VL) is a worldwide distributed neglected tropical diseases, receiving little attention and resources, despite its serious impacts on the population's quality of life.VL affects mostly the poorest populations, mainly in developing countries (Boelaert et al., 2010).More than 600 million people are at risk of contracting the disease, while some 50 000-90 000 new cases of VL occur worldwide annually (Dndi, 2020).
VL is the most severe form of the leishmaniasis.In 2018, over 90% of the new cases occurred in ten countries: Brazil, China, Ethiopia, India, Iraq, Kenya, Nepal, Somalia, South Sudan, and Sudan (Dndi, 2020).In the Americas, VL is caused by the protozoan parasite Leishmania (L.) infantum and has the phlebotomine sand fly Lutzomyia longipalpis as its main vector (WHO, 2020).The disease occurs in 12 countries, where 63,331 new cases have been registered from 2001 to 2018, with 97% of the cases reported in Brazil (WHO, 2010).The Northeast Region of Brazil is responsible for 56.7% of cases reported in 2019, while the State of Piauí accounted for 9.6% of case notifications in the country.In 2018, the capital Teresina was responsible for approximately 5.7% of reported national cases and 83% of cases in the state (Brazilian Ministry of Health, 2021).
The increase in visceral leishmaniasis incidence in Teresina is associated with the city's disordered urbanization in the late twentieth century.This growth was driven by rural exodus and severe droughts in the state's interior, which led to significant population migration to the city.This migration resulted in the occupation of tropical forest areas, the natural habitat of the VL vector (Lutzomyia longipalpis), allowing human contact with the diseasecausing agent.Moreover, the mosquito's adaptation to the peri-domestic environment and the poor living conditions in peripheral areas lacking adequate sanitation infrastructure contribute to vector proliferation, exacerbating the problem (Costa et al., 2005;Werneck et al., 2008).
The urbanization of VL is a relatively recent and poorly understood phenomenon in Brazil, bringing new challenges to control programs (Costa, 2008;PAHO, 2019).Deprivation, migration, haphazard urban settlement, environmental damage, lack of provision of urban services, such as sanitation and garbage collection, inadequate housing, and malnutrition are some of the numerous putative determinants of VL urbanization (Costa, 2008;Werneck, 2010;Harhay, 2011).
Like other diseases, VL distribution is related to the landscape, the primarily responsible for maintaining the vector's life cycle in nature.The first major urban outbreak took place in Teresina, Piauí State, in the early 1980s after a massive planting of acacias.It has thus been hypothesized that the two events were connected.
The vegetation has a strong connection with the disease, given that Lu. longipalpis needs sugars obtained from plants as its energy source (Costa et al., 2005).This relationship influences the parasitism of the insects because sugar is important to the development of the parasite in the sand fly's digestive tract (Belo et al., 2013;Lima et al., 2016).
When looking for sources of carbohydrates, phlebotomines may be attracted by typical characteristics of certain plants (Cameron et al., 1995;Costa, 2005;Dostálová e Volf, 2012;Hassaballa et al., 2021).These characteristics may be high CO 2 emission levels and volatile substances emitted by the plants (Muller et al., 2011;Lima et al., 2016).Besides that, an environment shadowed by vegetation, humid soil and the fall of leaves and fruits, making the soil rich in organic matter, conform a favorable habitat to the presence of these insects (Beyaert & Hilker, 2014;Abbasi et al., 2018).
One of the most curious phenomena concerning the urbanization of VL in Brazil is the relative absence of the disease in some large cities from the North and Northeast regions, such as Belem (Pará State), Recife (Pernambuco State) and Salvador (Bahia State), although they are in the same regions of other densely populated cities with high levels of transmission, such as Teresina (Piauí State), Fortaleza (Ceará State), Araguaína (Tocantins State), and São Luís (Maranhão State).The first three cities accumulated 177 reported cases of VL from 2007 to 2019 (0.29 cases per 10 thousand people), and the last four reported a total of 4893 cases in the same period (10.1 cases per 10 thousand) (Casaril et al., 2014).A possible explanation for such aspect might be related to different features of the urban environment among the cities.It has been suggested that certain vegetal species may be repellent or attractive to Lu. longipalpis and in some way could be beneficial to the dipterous insect, resulting in greater proliferation of the vector in areas where these plants are abundant (Cameron et al., 1995;Costa et al., 2005;Dostálová e Volf, 2012;PAHO, 2019;Hassaballa et al., 2021).
This work has the objective of verifying the association between the abundance of Lu. longipalpis and tree families in an urban endemic area in Brazil.

Location and general characteristics of the study area
The city of Teresina is located in the Center-North mesoregion of Piauí State, Northeastern Brazil.It lays over an area of 1392 km 2 area and has 861,442 inhabitants.It is situated between Parnaíba and Poti rivers at an altitude of 72 m above sea level and is fairly close to the Equator, at 05° 05′ 21″ latitude South and 42° 48′ 07″ longitude West (IBGE, 2010).
The climate in Teresina is tropical megathermic (AW) (Köppen), with average annual temperature of 27.7 °C, with two well-defined seasons: a rainy season in the first semester (from January to May) and a dry season in the second semester (from July to November) with an average annual rainfall of 1332 mm.Its soil is mainly latosol yellow and podzolic red-yellow.The city's vegetation is influenced by the sub-humid Amazonian and semiarid northeastern climates, characterizing a transitional environment, and along with invading species, has a diversified urban flora (Machado et al., 2010).
We have chosen a total of ten urban areas to perform the study, encompassing twelve neighborhoods throughout the city.The neighborhoods where the phlebotomines were captured and where the vegetation was later analyzed were Santa Isabel, Novo Uruguai, Alegre, Parque Ideal, Santa Maria da Codipi, Angelim, Todos os Santos/São Sebastião, Socopo/Morros, Catarina and Parque Piauí (Fig. 1).The areas studied, despite having native plants, also have exotic plants that were planted over time for urban landscaping purposes.

Sand fly capture
The sand flies were captured using CDC light traps-one placed close to the houses, on a tree in the backyard, less than 5 m from the house, and another in the living room.The geographic coordinates of the traps are shown in Fig. 1.The traps remained three consecutive nights, or, at the most, alternating days, working 12 h a night (from 6 p.m. to 6 a.m.), during 2016 and 2017.The captured sand flies were taken to the laboratory for identification and counting.The morphological identification of the sandfly Lutzomyia longipalpis (Lutz & Neiva, 1912) was carried out by specialists.

Plant sampling and identification
We collected botanic material (leaves and flowers) in an area of 100-m radius around a CDC-type trap for capturing Lu. longipalpis, installed in household backyards at a height equal to or above 1 m.Therefore, each area of the study has a total area of 314 m 2 , a small part of the city's 220 km 2 .A total of 418 plant specimens were collected following Rota (2008) and identified in the field by the botanist Marcelo Mesquita and later identified through material examination under a stereoscopic microscope, consulting the available specialized literature and by comparison with specimens from collections already included in the Herbarium.The Cronquist (1988) classification system was used and, for the abbreviation of authors' names, the system described in Brummit & Powell (1992).After herborization, they were included in the Herbarium collection (Table 1).The distance from each plant to the CDC traps was registered using a GPS location device (Garmin Dakota 10, Garmin, EUA).

Statistical analysis
To evaluate the association between captured sand flies and the local tree families, the total aggregate number of captured phlebotomines from the three night capturing efforts was considered.The traps were split in two groups, the first were those with less than 10 captured sand flies (varying from zero to seven sand flies in total) and the second were those with more than 10 captured sand flies (varying from 25 to 117).
We compared the proportion of tree families at a certain distance from the trap (< 30 vs > 30 m and < 50 vs > 50 m) stratified by the number of captured insects (< 10 and > 10 insects).Only those tree families presenting a total frequency of individual plants above 20 were considered (Anacardiaceae, Arecaceae, Combretaceae, Fabaceae, Meliaceae, and Myrtaceae).We employed the Fisher exact test to assess the association between the presence of tree families and the abundance of captured Lu. longipalpis.A level of 5% (p < 0.05) was used for considering an association as statistically significant.
We used multiple correspondence analysis (MCA) to evaluate the relationships between the presence of tree families within a radius of 50 m of sand fly traps and the abundance of Lu. longipalpis (Sourial et al., 2010).MCA provides a graphical representation of the results depicting each category of the analyzed variables as a point on the Cartesian axes; closer points indicate that the corresponding variable categories are related.Associations were assessed by inspecting the MCA plot and evaluating the proximity of the categories of the tree family variables (presence and absence of each family) to the supplementary variable indicating the abundance of sand flies (traps with < 10 and > 10 captured insects).

Results
A total of 327 sand flies were captured during the study.Among those, 201 were captured in the second semester of 2016 and 126 in the first semester of 2017.The greatest numbers of captured sand flies were registered in the Angelim (92 phlebotomines) and Alegre (117 phlebotomines) neighborhoods (Table 2).
There were quite different landscape compositions in the ten assessed areas, although they displayed common species.Concerning their physiognomy, they can be distinguished into three types: neighborhoods with few vegetal specimens and great built-up areas (Parque Piauí, Angelim); neighborhoods with vegetation areas larger than built-up areas (Santa Maria da Codipi, Socopo/ Morros, Alegre, Novo Uruguai, Todos os Santos/São Sebastião) and neighborhoods where the distribution ratio was homogenous (Santa Isabel, Catarina e Parque Ideal).The neighborhood with greatest vegetal diversity was Santa Isabel (n = 20 species), followed by Todos os Santos (n = 19) and Santa Maria da Codipi (n = 18).On the other hand, the neighborhoods with least diversity were Lourival Parente and Parque Piauí, with 9 and 11 species, respectively.Mangifera indica (Mango) and Azadirachta indica (Neem) were the only species common to all areas.
We verified that, among the traps that captured the least number of sand flies (< 10), there were proportionally more Anacardiaceae close to the traps (< 30 m) than at distances greater than 30 m (p = 0.022).In contrast, there were more trees of the Fabaceae family in the proximities of traps capturing more than 10 insects, than that found at greater distances (p = 0.036).It is noteworthy that there were no Fabaceae nor Myrtaceae close to the traps that have captured fewer insects (Table 3).
Table 4 shows that, among the traps that captured more than 10 sand flies, there were proportionally more Fabaceae close to the traps (< 50 m) than at distances greater than 50 m (p = 0.006).In contrast, there were more trees of the Meliaceae family at greater distances from traps capturing more than 10 insects, than that found at smaller distances (p = 0.025).
Figure 2 shows the results of MCA, in which the presence of trees from the Arecaceae, Myrtaceae, Combretaceae, and Fabaceae families were closely related to traps capturing more than 10 sand flies.The presence of trees   from the Anacardiaceae and Meliaceae families were somewhat related to traps capturing fewer insects.
Table 5 shows correlation between the distance to the tree and the abundance of Lu. longipalpis.

Discussion
We verified a significant variation in the abundance of captured sand flies in relation to different types of vegetation.For instance, trees from the Anacardiaceae and Meliaceae families were more frequently found close to traps that captured few sand flies, suggesting a possible repellent effect.In contrast, there were more trees of the Arecaceae, Myrtaceae, Combretaceae, and Fabaceae families in the proximities of traps that captured more than 10 insects, indicating a direct relationship between this family and the abundance of the vector and a possible attractiveness of this family for sand flies.From our knowledge, this is the first study to verify the associating of landscape urban flora with the vector of VL.
In the phlebotomine capture spots, plant species from 26 different families were collected and identified.The families with the most abundant specimens in the study area were Anacardiaceae, Meliaceae, Myrtaceae, Arecaceae, Fabaceae, and Combretaceae.The special abundance of these families was already expected, given that they are highly distributed in Brazil and in the world (Monteiro & Fisch, 2005;Muellner et al., 2006;Nadia & Machado, 2007;Nascimento & Antunes de Paiva, 2007;Andrade et al., 2009;Couvreur et al., 2011).
The Fabaceae family, whose most prevalent species in the study were Anadenanthera macrocarpa (blackangico) and Tamarindus indica L. (tamarind), was consistently associated with trap that collected larger amounts of sand flies, suggesting a possible attractive effect to Lu. longipalpis.Indeed, it has already been observed in the city of Teresina, that its relatively high presence could be associated to greater phlebotomine densities (Lima et al., 2016).In a DNA barcode study, the Fabaceae DNA was detected in 94.7% of the phlebotomines, while only 2.9% of the plants in the area belonged to this family (Lima et al., 2016).A possible explanation, is that the Fabaceae family has deep roots, which have secondary radical nodules containing bacteria of the Rhizobium gender which are capable of fixing enough nitrogen, which would be available for the soil biota where the phlebotomine larvae proliferate (Vivero et al., 2015).The occurrence of the sand fly vector Phlebotomus orientalis, the main vector of the protozoan parasite Leishmania donovani in East Africa, in habitats with the Fabaceae tree Acacia seyal has also been reported in the literature (Elnaiem, 2011).The high abundance of the sand fly Phlebotomus orientalis in Acacia seyal vegetation was attributed to the tree density and soil type, but biological characteristics of the Fabaceae family should also be considered, such as its large biomass containing multiple nutrients, and specific volatile organic compounds (Elnaiem, 2011;Hassaballa et al., 2021).They are usually large and have deep roots containing bacteria that can fix nitrogen making the soil a rich microhabitat to sand fly larvae (Vivero et al., 2015).
The Myrtaceae, Combretaceae and Arecaceae families were identified only in MCA as associated with traps capturing greater quantity of insects.The most common Myrtaceae species found in this study, Psidium guajava (guava), has been previously detected as attractive to the sand fly Phlebotomus papatasi, the main vector of Leishmania major, a protozoan parasite that causes cutaneous leishmaniasis in the Old World (Junnila et al., 2011).
Members of the Arecaceae and Meliaceae family were associated with traps capturing fewer quantities of sand flies.The most common species from the Arecaceae family was Cocos nucifera (coconut palm) and from the Meliaceae family was Azadirachta indica (Neem).Plants from the Meliaceae family such as Azadirachta indica and Melia azedarach have been associated with a significant increase mortality of Lutzomyia longipalpis larvae (Andrade-Coelho et al., 2009).Cocos nucifera may act negatively in relation to Leishmania, since the polyphenol contents found in the shell of the coconut completely inhibit the cellular growth of promastigote forms of Leishmania (Lima et al., 2015).Studies show that the phenolic lipids found in the Anacardiaceae family have toxic or allergic properties, perhaps also with toxic or repellent properties to phlebotomines (Ferreira de Carvalho et al., 2019).In an ethnobotany study it has been verified that the Anacardiaceae and Meliaceae families were the most used as repellent species to insects (Mavundza et al., 2011). Indeed, (Pinheiro et al., 2013), when studying six areas in the Northeast of Brazil, registered smaller number and diversity of phlebotomine species in areas of mango monoculture and in cashew areas.In opposition, Vivero et al., (2015), in a study performed in Colombia, verified that the soil under A. excelsum (wild cashew) constitutes a breeding site for larvae appropriated for phlebotomines, as it is a source of organic material and protection for the higher layers of the soil in which the phlebotomines develop.An advantage of this study was the spatially dispersed distribution of the study areas across the city of Teresina, allowing some representativeness for the whole city.However, as the study was conducted over only 1 year, during which the study areas were represented differently in time and as the weather influences the abundance of phlebotomines (Salomon, 2021), the quantities of phlebotomines obtained may not be representative of the abundance of the neighborhood throughout the entire year.The use of luminous CDC-type traps was the chosen method to measure the density of phlebotomines, considered a good option of this type of insect (Alten et al., 2015).Another limitation of the study is the small sample size constraining the statistical power of the analysis.Additionally, the evaluation of the influence of trees on the abundance of sand flies might have been impaired by the action radius of the CDC-type trap for phlebotomines is closer than 6 m, depending on the species (Alexander, 2000), but that of the trees may depend on the dimension of the odor plume carried by the wind and perceived by the insect may go up to 10 m (Voskamp, 1998).

Conclusions
This study, despite its limitations, has shown important relationships between urban trees and Lu.longipalpis abundance and might inform landscape planning of cities where the control of diseases transmitted by flying arthropods is a priority.A balance between attractive and repellent common trees-already largely in use in tropical cities-might turn to be a clean alternative to control urban vector borne diseases such as leishmaniasis and arboviruses.

Fig. 1
Fig. 1 Location of the study area.Red points indicate the traps location in the studied neighborhoods in the city of Teresina

Fig. 2
Fig. 2 Multiple correspondence analysis of tree families and sand flies captures.All categories of tree families are expressed as present (Yes) or absent (No) followed by the name of the family (Melia, Meliaceae; Anac, Anacardiaceae; Combret-Combretaceae; Areca, Arecaceae; Fabac, Fabaceae; Myrta, Myrtaceae)

Table 1
List of angiosperm plants found in the city of Teresina and accession number in the Herbarium Graziela Barroso-UFPI

Table 3
Percentage of tree families over the total identified trees, according to the areas delimited by the distance to the CDC trap, stratified according to the number of captured Lu. longipalpis specimens The bold was just to emphasize the p-value below 0.05, significantNumber of Lu.

Table 4
Percentage of tree families among the total identified trees according to the number of Lu. longipalpis captured, stratified according to a distance up to 50 m from the CDC trapThe bold was just to emphasize the p-value below 0.05, significant

Table 5
Correlation between the distance to the tree and the abundance of Lu. longipalpis