Geographic Distribution of the Genus Panstrongylus Berg, 1879 in the Neotropic with Emphasis on Trypanosoma cruzi Vectors

Panstrongylus is a Neotropical taxa of 16 species, some more widespread than others, that act as vectors of Trypanosoma cruzi, the etiologic agent of Chagas disease (CD). This group is associated with mammalian reservoir niches. There are few studies of the biogeography and niche suitability of these triatomines. Using zoo-epidemiological occurrence databases, the distribution of Panstrongylus was determined based on bioclimatic modelling (DIVA GIS), parsimonious niche distribution (MAXENT), and parsimony analysis of endemic species (PAE). Through 517 records, a wide presence of P. geniculatus, P. rufotuberculatus, P. lignarius, and P. megistus was determined and recorded as frequent vectors of T. cruzi in rainforest habitats of 24–30 °C. These distributions were modeled with AUC >0.80 and <0.90, as well as with the seasonality of temperature, isothermality, and precipitation as relevant bioclimatic variables. Individual traces for each taxon in Panstrongylus—1036 records—showed widely dispersed lines for frequent vectors P. geniculatus, P. lignarius, P. rufotuberculatus, and P. megistus. Other occasional vectors showed more restricted dispersal, such as P. howardi, P. humeralis, P. lenti, P. lutzi, P. tupynambai, P. noireaiui, and P. chinai. Areas of defined environmental variation, geological change, and trans domain fluid fauna, such as the American Transition Zone and the Pacific Domain of Morrone, had the highest Panstrongylus diversity. Pan-biogeographic nodes appear to be areas of the greatest species diversity that act as corridors connecting biotopes and allowing fauna migration. Vicariance events in the geologic history of the continent need to be investigated. The geographical distribution of Panstrongylus overlapped with CD cases and Didelphis marsupialis/Dasypus novemcinctus presence, two important reservoirs in Central and South America. The information derived from the distribution of Panstrongylus provides knowledge for surveillance and vector control programs. It would increase information on the most and less relevant vector species of this zoonotic agent, for monitoring their population behavior.


Introduction
The genus Panstrongylus Berg, 1879, belongs to the tribe Triatomini, subfamily Triatominae (Hemiptera, Reduviidae), and comprises 16 living species and one fossil species described from the Dominican Republic [1][2][3]. This taxa is restricted to the American continent and the other different species are found from Mexico to Argentina, some of which are very widespread [4][5][6].

Study Area
The study area corresponded to the geographic distribution of the genus Panstrongylus, which is restricted to the Neotropical taxa, specifically from Mexico to Argentina [4,6,17].

Species Database and Georeferencing
Records of triatomine occurrences from technical reports, museum records, biological laboratory collections and specialized bibliography were used to compile a database for each species. All were geo-referenced using DICES platform (http://www.dices.net, accessed on 21 September 2020) and GADM version 2.8 (http://www.gadm.org, accessed on 12 February 2021). The database was cleansed of duplicate occurrences to eliminate sampling effort bias, enable species distribution modelling and pan-biogeographic analysis, and only species separated by more than 1 km were selected to reduce autocorrelation.
The georeferenced database of each species was transformed into a comma-separated values file (CSV) with .csv extension by converting the original grd format into a MAXENT compatible ASCII format (*.asc). DIVA-GIS 7.0 was used as the transformation platform.
In total, 19 layers of bioclimatic variables from global climate database Worldclim were used (http://www.wordclim.org, accessed on 13 March 2021). They were generated from the interpolation of mean temperature, its maximum and minimum, and monthly precipitation for 50 years, with a spatial resolution of 1 pixel = 0.86 km 2 in Ecuador [20]. An elevation layer available at the same resolution was also considered.
A first run of 10 models was performed using the MAXENT parameters: 1000 iterations, convergence threshold of 1.0 × 10 −5 , 75% of data sets for model calibration and 25% for evaluation. Spearman correlation tests were performed with PAST software version 4.10 on the climatic data obtained with DIVA GIS to reduce collinearity of the variables. Variables with values of <0.7 were included in the model.
The Jackknife test for AUC (area under the curve) was used to determine the percentage contribution and best fit of the model. After pilot runs of MAXENT, the most predictive variables were obtained and compared with the most biologically relevant variables according to the distribution/habitat where Panstrongylus and mammalian species are likely to reproduce and survive [6,[21][22][23][24][25][26][27].
Using the selected variables, MAXENT was run again in 10 replicates to obtain an average model for each species [28]. The sensitivity and specificity of the predictions were obtained from the receiver operating characteristic curve (ROC). The area under the curve (AUC) provided a measure of model performance across different limits, with AUCs close to 1 indicating good fit and those near 0.5 indicating poor fit.
The action script file (.asc) output of the MAXENT model was converted with DIVA GIS and visualized on a QGIS map (QGIS version 3.16.10; threshold cut-off at the 10th percentile) for possible areas of suitability for the presence of the different species. In addition, from the database presence of Panstrongylus species, spatial analyses of species diversity/richness (alpha diversity) were performed using DIVA GIS.

Pan-Biogeographical Analysis
The construction of individual traces, based on geodetic distance as a starting point, for 12 out of 15 species of Panstrongylus, with three or more occurrence records, was carried out using the pattern analysis, spatial statistics and geographic exegesis program (PASSaGE version 2) from a shapefile constructed with the geographic distribution of each species in DIVA GIS [15]. Individual traces were superimposed on the hierarchical biogeographic regionalization of Morrone [29,30], based on biogeographic homologies for the Neotropical region ( Figure S1).
Generalized traces were designed from the coincidence of individual unoriented trajectories in minimal spanning trees (MST) based on Jaccard index, compatible with continental scale analysis by PAST sofware.
Zones of convergence between two or more biotic/geological regions (pan-biogeographic nodes) were determined by constructing a binary matrix (1 = presence; 0 = absence) per locality x taxon and a connectivity matrix from MST, driven by the sum of the row values corresponding to the node value/locality and the average node value. Nodes were defined where individual nodal values exceeded the calculated average [29,30].
Parsimony of endemicity (PAE) and maximal parsimony cladistic analyses were used to determine areas of endemism and the most parsimonious cladograms and their corresponding consistency indices (PAST software), with terminal dichotomies indicating more recent biotic exchange with terminal dichotomies indicative of more recent biotic exchanges [13,30,31].

Geographical Distribution Analysis of Panstrongylus Species as Frequent or Sporadic Vectors of T. cruzi
A total of 517 documents containing geo-referenced records of Panstrongylus were selected from Scielo, PubMed, BioOne, Redalyc, e-journals, and laboratory records. The geographical distribution of Panstrongylus according to the biogeographical regions of Morrone [31] showed that the majority of its species are found in rainforests, between 24 • C and 30 • C ( Table 1). Panstrongylus diasi (Pinto y Lent, 1946) Distributed in "Caatinga" (Brazil), characterized as a semi-arid, warm, and xerophytic area; Santa Cruz (Bolivia), with a warm tropical climate. [4,17,34] Panstrongylus geniculatus (Latreille, 1811) * From Mexico and Central America to South America (excluding Chile). Found in caves of armadillos, rodents, reptiles (geckos), and bats; also associated with opossums and bird nests, hollow trunks, under the bark of trees, in bromeliads and among palm leaves (Attalaea, Acromia, and Copernitia). Eurythermic species adapted to dry and humid ecotypes such as the Amazon rainforest and xenomorphic grassland areas. [4][5][6]8,17,35] Panstrongylus guentheri (Berg, 1879) Areas of thorny scrub and dry forest in the Chaco region. Found in armadillo and rodent caves; also associated with opossums. Can be found in the peridomicile associated with woodpiles and environments of goats and dogs. [4,17,36] Panstrongylus hispaniolae ( †) (Pionar, 2013) Species are described from a fossil contained in a piece of amber found in the La Toca mine, located in Puerto Plata, Dominican Republic. [3] Panstrongylus howardi (Neiva, 1911) Equatorial rainforest environment (Manabí province). It has been associated with marsupials and rodent nests inserted in the Aiphanes eggersi palm. Found in dwellings and the peridomicile in microhabitats such as piles of bricks or wood coexisting with rodents. [4,26] Panstrongylus humeralis (Usinger, 1939) Distributed in humid forest from Panama and Colombia. [4,5] Panstrongylus lenti (Galvão y Palma, 1968) Habitat typical of "Cerrado"; shrubby, herbaceous, and dry vegetation in Goiás and Bahia (Brazil). [17,36] Panstrongylus lignarius (Walker, 1837) * Described from tropical dry forest habitats in Brazil, Colombia, Ecuador, French Guiana, Perú, Suriname, and Venezuela. Found in trees, hollow trunks, and palm trees; it has been associated with toucan nests and mammals such as porcupines and spiny rats. In Perú, it is considered a frequent species in the home and peridomicile. [17,37] Panstrongylus lutzi (Neiva y Pinto, 1923) * Endemic to northeastern Brazil ("Caatinga" area). It colonizes armadillo caves and has been found in dwellings associated with rodents and opossums. [4,17,[38][39][40]

Species Distribution and Habitat References
Panstrongylus martinezorum (Ayala, 2009) Described from a male, collected in Puerto Ayacucho (Amazonas, Venezuela) and other males and females located in collections whose records indicate urban and peri-urban areas of capture.

[2]
Panstrongylus megistus (Burmeister, 1835) * Wide distribution in Brazil, with great potential for colonization of artificial environments. In the wild, it is associated with palm trees and animal burrows. The Atlantic forest seems to represent the center of the distribution, although the species is also distributed in humid areas of the "Caatinga" and "Cerrado". [17,34,37,41] Panstrongylus mitarakensis (Bérenger y Blanchet, 2007) Distribution in the Mitarakara Mountains, French Guiana. [42] Panstrongylus rufotuberculatus (Champion, 1899).
Most widespread after P. geniculatus in Central and South America. Forest habitat, found in palms, hollows and mammal shelters such as bats, monkeys, armadillos. Occasionally in the peridomicile in association with domestic animals. [4,10,17,37,43] Panstrongylus tupynambai (Lent, 1942) Atlantic forest region of southern Brazil and Uruguay. Collected in humid rocky habitats, palms, trees, shelters of mammals; human dwellings, and in the peridomicile. [4,17] Panstrongylus noireaui sp. nov The specimens were found in the locality of Ayata (Camata), in the department of La Paz, province of Ildefonso de las Muñecas, Bolivia. It is located in the Yunga ecoregion. [44] * reported as infected by T. cruzi and some vectorial role. † fossil species.
The MAXENT model for widespread of insects vectors, such as P. geniculatus, P. rufotuberculatus, P. lignarius, and P. megistus, showed an AUC of >0.80 (robust AUC values based on test and training data are >0.80 and <0.90) [18], which is similar to restricted species such as P. chinai.
The species recorded as frequently infected with T. cruzi did not escape the bias of wider coverage (more samples) because of their epidemiological interest. Below is a more detailed description of the distribution of these vector species.

Panstrongylus geniculatus
This species was found to be the most widespread and abundant in the Neotropics (293 records). It occurs in different climatic regions, from 360 to 1750 m above sea level (m.a.s.l), such as tropical dry forests, tropical forests, Amazon forests, anthropogenic savannas (in the central and northern regions of Colombia and Venezuela), xerophytic areas of the "Caatinga"/dry grasslands of the "Cerrado" (Brazil), and the areas of influence of the Chaco Domain (ChD) (corridors in Brazil such as Bahia, Minas Gerais, and Goias).
The potential distribution model ( Figure 1A) was generated with a robust predictive value (AUC = 0.859).
Isothermality (Bio3; 24.8%) was the most weighted variable for the model, with the remaining weight equally distributed among the other variables (4.7% on average). This would explain the eclecticism of this species towards different ecoregions, which facilitates the anthropogenic influence on its distribution and reinforces its role as a vector of the trypanosomas [8,10,45,46].

Panstrongylus rufotuberculatus
This species (430 records) also had a wide geographical distribution ( Figure 1B), covering Central and South America and some Caribbean islands. It was the most widespread after P. geniculatus and had a good AUC value (0.873).
The model showed areas of bioclimatic suitability in the province of Pantepui in the Brazilian Boreal Domain (BBD) up to 1700 m.a.s.l., including Rondônia, Acre (Amazon corridors); Mato Grosso (Pantanal) and "Caatinga" (Brazil). Other areas were the Argentine Chaco (349 m.a.s.l) and the northeastern region, closer to the Caribbean coast (between 164 and 1175 m.a.s.l., Venezuela).
The variable that contributed most to the model was the seasonality of temperature (28.1%). Other variables were included in the model, each with a weight of no more than 4%. This would justify the fact that it is an eclectic species in terms of the ecotopes in which it is found and where it could be a potential vector for T. cruzi [4,43,[47][48][49][50] Figure 1C).
The bioclimatic variable with the highest contribution to the model was seasonal temperature (Bio 4) (56.8%), another variable with less predictive value was isothermality (Bio 3). Although it is a moderately widespread species, its role as a vector for T. cruzi should be studied, given its natural infection and its occasional presence in human dwellings from Brazil and Colombia [51,52].
3.1.4. Panstrongylus megistus P. megistus (232 records) showed a wide distribution in humid areas of "Caatinga" and "Cerrado" (230-530 m.a.s.l) Some authors report that the Atlantic Forest is the center of its distribution, with a confluence of populations of this vector in São Paulo linked to ruralurban human migrations, and a great capacity to colonize artificial ecotopes [34,[52][53][54].
A suitable bioclimatic area for the occurrence of this species was the MTZ (Sierra Madre West, Sierra Madre East, Sierra Madre del Sur, and Chiapas) and PD (provinces of Western and Guajira), all of which were modeled with a robust AUC (0.925) ( Figure 1D).
Seasonality of rainfall (18.8%) was the variable with the highest contribution in the model (low weighting); the other variables would influence the distribution of the species with 4.5% each. This shows that there is no specific bioclimatic condition for its presence, which makes it eclectic and therefore susceptible to anthropogenic effects, some of which are determinants in the epidemiology of CD.  The variables with the highest contribution to delimiting these suitable areas were isothermality (31.7%) and annual rainfall (30%), indicating precise bioclimatic conditions for its occurrence. This would be the first possible distribution model for P. chinai, which is a potential successor to the primary vector of T. cruzi. This was modeled with a good value (AUC = 0.859) and was consistent with bibliographic registers [55,56].

Individual and General Traces of the Distribution of Panstrongylus in the Neotropic
Individual traces for each taxon within the genus Panstrongylus showed that some species, frequent vectors of T. cruzi, such as P. geniculatus, P. lignarius, P. rufotuberculatus, and P. megistus are widely distributed. Other species that are occasionally infected by T. cruzi have a restricted distribution, such as P. chinai, P. diasi, P. lenti, P. lutzi, P. guentheri, P. howardi, P. humeralis, P. noireaui, and P. tupynambai.
A generalized trace analysis based on the congruent topology of individual traces for Panstrongylus (Table S1; Figures S2 and S3) revealed four traces across the entire Neotropical taxa [57,58] (Figure 2A Triatomini appeared 32 million years ago in the Oligocene, when South American fauna began to migrate towards North America. This coincided with the radiation of mammals (including marsupials and Xenarthra), neotropical birds, and the diversification and distribution of palm species of Attalea, Acrocomia, and Butia throughout South America, which served as micro niches for the triatomines, including specific T. cruzi vectors [12,[59][60][61][62][63][64].
The distribution of southern cone biota began in the Neogene when the Andes were uplifted and cold ocean currents developed, causing a drier climate towards the Chaco. During the interglacial periods, the fragmentation and mixing of the Yungas and Parana forests with the xerophilous forests of the Chaco may have been a factor in the diversification (vicariance) of the Chaco group (P. guentheri, P. diasi, P. lutzi, P. lenti, P. tupynambai) [65,66].
The other group was made up of P. howardi, P. chinai, P. humeralis, P. rufotuberculatus, P. lignarius, P. geniculatus, P. noireaui, and P. megistus, and was associated with the Andean uplift and marine invasions during the Miocene (Pebas Lake), with river channels, swamps, and a characteristic savanna climate [67].
P. geniculatus, P. rufotuberculatus, P. lignarius, P. howardi, P. noireaui and P. chinai converged between the SATZ zone and the PD. In addition, P. diasi, P. lutzi, P. lenti, P. megistus, P. lenti, P. rufotuberculatus and P. geniculatus were distributed between the ChD and the Parana domain. Common species in SATZ and PD could be consequence of a trans domain fluid fauna [66].

Species Richness of Panstrongylus, the Associated Mastofauna and the CD Cases in the Neotropical Region
The modeled distribution of D. novemcinctus (302 records; robust AUC = 0.81) is so wide that it stretches from the southern United States to the middle of Argentina (including some Caribbean islands). It was comparable with Panstrongylus distribution in America. Mean annual temperature (Bio1) was the highest contributing variable (28%) in its distribution; annual precipitation (Bio12), mean coldest quarter temperature (Bio11), and temperature seasonality (Bio4) showed 14% (mean) of contribution ( Figure 3A). geological changes in the Neotropics (generalized traces b, c and part of d, and pan-biogeographic nodes 1 and 2; Figure 2A,B). One of them includes the PD (western Ecuador, Ecuadorian zone and Cauca), SATZ ("Paramo"), SBD (Yungas and Ucayali), and BBD (Napo). The second area was delimited in the CH (Chaco and Parana provinces) with 4-5 species [4,29].
Modelling showed that Panstrongylus species richness overlapped with high Dasypus/Didelphis (potential blood source for this vector) areas, all coinciding with CD cases (latest 2014). This scenario of overlap would be a potential risk factor for an increase in the incidence of this anthroponotic disease ( Figure 3C,D) [11,[72][73][74]  Dasypus novemcinctus would be a versatile reservoir species in terms of the bioclimatic and geomorphological conditions for its presence. It has been reported to be infected with several T. cruzi genotypes [11,21,[68][69][70].
The distribution of D. marsupialis (203 records; robust AUC = 0.86), a primary reservoir in Central and South America (with systemic and anal gland colonization by T. cruzi, the latter with infective forms similar to those of the vector) [11,71], was consistent with Panstrongylus distribution, restricted to the Neotropics (from Central America to southern Brazil; Figure 3B). Seasonality of temperature (Bio 4) accounted for 56.6% of the distribution, while iso-thermality (Bio 3), mean daily temperature range (Bio 2), and mean temperature of the coldest quarter (Bio 11) accounted for 9% (mean) ( Figure 3B). Spatial analysis of diversity (alpha-diversity) using DIVA GIS showed that the most abundant sites for Panstrongylus were located at the junction of provincial boundaries of different areas (15 records of georeferenced Panstrongylus species).
Richness analysis revealed two geographical areas with the highest number of Panstrongylus species, both coinciding with previously described areas of environmental and geological changes in the Neotropics (generalized traces b, c and part of d, and pan-biogeographic nodes 1 and 2; Figure 2A,B). One of them includes the PD (western Ecuador, Ecuadorian zone and Cauca), SATZ ("Paramo"), SBD (Yungas and Ucayali), and BBD (Napo). The second area was delimited in the CH (Chaco and Parana provinces) with 4-5 species [4,29].
Modelling showed that Panstrongylus species richness overlapped with high Dasypus/Didelphis (potential blood source for this vector) areas, all coinciding with CD cases (latest 2014). This scenario of overlap would be a potential risk factor for an increase in the incidence of this anthroponotic disease ( Figure 3C,D) [11,[72][73][74] The evolutionary history of Xenarthra in the Neotropic is linked to the uplift of the Andes, changes in sea level, temperature variations (transition between the Eocene and Oligocene), and the transition from warm/humid/temperate tropical forest environments to more arid/dry environments dominated by savannas. This explains the distribution of Euphractine in the Chaco and Dasypus in the Pacific [60]. These burrowing mammals are considered groups with important zoo geomorphic environmental impacts [69], which should be investigated for eco pathogenic complexes of some zoonoses.
This evolutionary event coincided with the major adaptive radiation of the Didelphimorphia (Ameridelphia, marsupials) in the Late Cretaceous/Early Cenozoic. These taxa diversified when they crossed Australia/America and disappeared from Europe [67,75]. This would explain why some of its species, such as D. marsupialis, the primary reservoir of T. cruzi in the Americas, are found in a wide range of biomes in the Neotropic [11].
It is also important to note that D. marsupialis is the only T. cruzi reservoir that has circulating forms of the parasite in its peripheral blood and similar forms of the triatomine infective forms in its anal glands. It therefore acts as both a reservoir and a disperser, ensuring presence and parasite load in the transmission cycle in the absence of a vector [11,71,76].
The highest diversity of Panstrongylus was found in rainforest habitats. Seasonality of temperature, isothermality, and precipitation were the most important bioclimatic variables, which is in agreement with the literature.
Species from the Chaco subregion showed wider distributions, with rainfall as a more important variable, whereas temperature was more important for more widespread species from PD.
The geographical distribution of Panstrongylus overlapped with CD cases in Central and South America, indicates the need for more monitoring of the most widespread species within these taxa. Individual traces suggest that evolutionary histories would show different trajectories, and that some species may have expanded into atypical regions. One example is P. chinai, which is typical of Perú and Ecuador, but may have expanded into the Andean-Venezuelan mountains.
The generalized traces and nodes show that the Andean and Chaco "Cerrado" regions contain the highest number of species of the genus Panstrongylus.
The diversification and separation of Panstrongylus may have been influenced by the Andean uplift, with the formation of "Caatinga", "Cerrados", and marine transgressions, such as the formation of Pebas lakes.
Pan-biogeographic nodes appear to be areas of greatest species diversity that act as corridors connecting biotopes and allowing faunal migration [13]. Vicariance events in the geologic history of the continent need to be investigated in function to explain the origin of the geographic distribution of Panstrongylus and the potential impact of its subsequent dispersal.
The same scenario that determined the distribution of Panstrongylus led to the diversification of Xenarthra and Didelphimorpha, two mammal taxa associated as blood sources for triatomines.

Conclusions
This paper presents the first analysis of the geographical distribution of the genus Panstrongylus in the Neotropics, focusing on potential vector species of T. cruzi, using DIVA GIS, MAXENT and PAE in conjunction. It has also been possible to predict the likely areas of occurrence, which extend the known distribution for almost all of the species.
The modelling of the distribution of the species, with emphasis on the vectors of the CD pathogen, has made it possible to identify areas of greater suitability for the species of the genus Panstrongylus.
The vicariance and/or dispersal of such taxonomic groups could be linked to the paleo-ecological processes that took place on the continent.
The models of the species of the Chaco subregion were influenced in a greater importance for the precipitation, while the species of the Pacific domain, with wide distribution, were determined by the variables of temperature.
The geological and paleoclimatic events that led to the diversification of Xenarthra and Didelphidomorpha, mammal groups closely linked to T. cruzi transmission cycles, and the diversification of palms, a niche of excellence for triatomines, may be related to the distribution of Panstongylus. Further studies will be necessary in the future.
The information derived from Panstrongylus models would provide knowledge for the surveillance and vector control programs, as it would increase information on the most relevant species as vectors of zoonotic agents and less studied species for monitoring of their population behavior.
Supplementary Materials: The following supporting information can be downloaded at https:// www.mdpi.com/article/10.3390/tropicalmed8050272/s1, Figure S1. Biogeographic regionalization scheme of Morrone of the Neotropical region; Figure S2. Identification of the nodal areas by means of minimum spanning tree with the Euclidean distance index; Figure S3. Identifying nodal areas using Minimum Spanning Tree with Jaccard distance index. Table S1: Area/taxa matrix, with an external area added for reference, used for building minimal spanning trees and parsimony analysis of endemism (PAE).

Institutional Review Board Statement:
The study did not require ethical approval, because the research not involving humans or animals.