Larval habitats, species composition and distribution of malaria vectors in regions with autochthonous and imported malaria in Roraima state, Brazil

Background Malaria control requires local action. Assessing the vector diversity and abundance provides information on the local malariogenic potential or risk of transmission. This study aimed to determine the Anopheles species composition, habitats, seasonal occurrence, and distribution in areas with autochthonous and imported malaria cases in Roraima State. Methods A longitudinal study was conducted from January 2017 to October 2018, sampling larvae and adult mosquitoes in three municipalities of Roraima State: Boa Vista, Pacaraima and São João da Baliza. These areas have different risks of malaria importation. Four to six mosquito larval habitats were selected for larval sampling at each municipality, along with two additional sites for adult mosquito collection. All larval habitats were surveyed every two months using a standardized larval sampling methodology and MosqTent for adult mosquitoes. Results A total of 544 Anopheles larvae and 1488 adult mosquitoes were collected from the three municipalities studied. Although the species abundance differed between municipalities, the larvae of Anopheles albitarsis s.l., Anopheles nuneztovari s.l. and Anopheles triannulatus s.l. were collected from all larval habitats studied while Anopheles darlingi were collected only from Boa Vista and São João da Baliza. Adults of 11 species of the genus Anopheles were collected, and the predominant species in Boa Vista was An. albitarsis (88.2%) followed by An. darlingi (6.9%), while in São João da Baliza, An. darlingi (85.6%) was the most predominant species followed by An. albitarsis s.l. (9.2%). In contrast, the most abundant species in Pacaraima was Anopheles braziliensis (62%), followed by Anopheles peryassui (18%). Overall, the majority of anophelines exhibited greater extradomicile than peridomicile-biting preference. Anopheles darlingi was the only species found indoors. Variability in biting times was observed among species and municipalities. Conclusion This study revealed the composition of anopheline species and habitats in Boa Vista, Pacaraima and São João da Baliza. The species sampled differed in their behaviour with only An. darlingi being found indoors. Anopheles darlingi appeared to be the most important vector in São João da Baliza, an area of autochthonous malaria, and An. albitarsis s.l. and An. braziliensis in areas of low transmission, although there were increasing reports of imported malaria. Understanding the diversity of vector species and their ecology is essential for designing effective vector control strategies for these municipalities. Supplementary Information The online version contains supplementary material available at 10.1186/s12936-021-04033-1.


Background
Malaria incidence in the Brazilian Amazon reduced from 333,461 cases in 2010 to 156,916 in 2019 [1]. Regardless of such regional progress, control interventions have not been successful in states sharing international borders, where malaria cases continue to increase [2][3][4]. Roraima is a Brazilian state in the extreme north of the country that shares international borders with Guyana and Venezuela. Since 2016, this region has been experiencing an unprecedented flow of migrants from Venezuela due to that country's economic, social and political crises [3,[5][6][7]. This uncontrolled migration translates into a larger number of imported cases [2,8].
Data from the Ministry of Health, Epidemiological Surveillance Information System-malaria (SIVEP-malaria), shows that the number of cases reported in Roraima almost tripled from 8969 in 2016 to 23,369 in 2018. Of those, 34% were imported cases from other states in Brazil (3,630 cases) and other countries (5513 cases) [1,2], particularly from Venezuela (4478 cases) and Guyana (610 cases) [1]. Two municipalities in Roraima, Boa Vista (the capital) and Pacaraima (bordering Venezuela), reported more than half of the imported malaria cases in Brazil across international borders between 2007 and 2018. Although these municipalities are considered areas of low transmission risk, the influx of migrants and cases makes them vulnerable to malaria outbreaks and epidemics [2,8]. The municipality of São João da Baliza (SJB) is located in the south of the state of Roraima, a region of tropical rainforest, endemic for malaria. In 2019, 634 cases were reported, of which 310 were acquired in the urban area of the municipality [1]. Furthermore, in Roraima state, Plasmodium falciparum infections were responsible for 9.8% of the autochthonous cases and 26% of the imported cases from other countries. Cross-border and authochtonous falciparum malaria threatens the Brazilian plan to eliminate this parasite from the country [2,8], fuelling concern about importing parasites that are resistant to the drugs used to treat malaria in Brazil [9][10][11][12]. Therefore, characterizing the vector species involved in local transmission in these border municipalities and their larval habitats is considered a priority.
Considering the ongoing high levels of imported malaria in Roraima [1,2], and the diverse ecological landscape in the state [38], this study aimed to identify anopheline larval habitats and mosquito species distribution and behaviour in areas with imported (Boa Vista and Pacaraima) and autochthonous malaria (São João da Baliza). The identification of potential larval habitats and the ecological conditions that determine the presence of Anopheles species competent to transmit Plasmodium is an important contribution for the control and elimination of malaria in Roraima.
This information will allow for the assessment of potential local malaria vectors in each municipality and may explain differences in transmission patterns. The long-term goal is to direct vector control measures as part of integrated policies to reduce the region's An. albitarsis s.l. and An. braziliensis in areas of low transmission, although there were increasing reports of imported malaria. Understanding the diversity of vector species and their ecology is essential for designing effective vector control strategies for these municipalities. Keywords: Malaria vectors, Larval habitats, Anopheles darlingi, Anopheles albitarsis s.l., Human biting rate, Imported malaria malariogenic potential and mitigate, if possible, the effect of imported malaria cases.

Study areas
The study was conducted in the municipalities of Boa Vista, Pacaraima and São João da Baliza, Roraima State (Fig. 1). These three municipalities were chosen because of their numbers of autochthonous and imported malaria cases, and their status in annual parasite incidence (IPA), a risk indicator for malaria transmission used by the Ministry of Health in Brazil. This indicator expresses the number of confirmed new infections, considering the likely place of infection, per year per 1,000 inhabitants. The urban areas of Boa Vista had a very low risk (IPA 0.4), while Pacaraima had a low risk (IPA 1.0) and São João da Baliza was considered a region with medium risk (IPA 37.8) for the year 2018. Boa Vista had an estimated population of 399,213 in 2019 and is located in the centre of the state (02°49′12" N; 60°40′23" W). It is 231 km from the main Venezuelan town in the country's malaria-endemic region (Santa Elena do Uairén), and it is 133 km from Guyana (Lethem). Its location allows easy access to the international borders and the state of Amazonas by highways. Indeed, Boa Vista predominated as the municipality with the most reported cases imported from neighbouring countries. In 2018, it registered 5,545 cases. Of these, 168 were autochthonous, 2115 were imported from Venezuela, and 533 from Guyana. Plasmodium falciparum infection imported from Venezuela (481 cases) and Guyana (97 cases) represented 80% of the P. falciparum registered in the city (722 cases) (1). Pacaraima is located in the north of the state (04º25′51'' N; 61º08′45'' W), in a mountainous region, above 1000 m altitude with an estimated population of 17,401 inhabitants, including 5838 indigenous peoples This municipality presents an intense movement of people between Pacaraima and Santa Elena do Uairén. The imported malaria in this municipality is almost exclusively from Venezuela. Migration fluctuates according to the treatment offered in Santa Elena do Uairén, a border town with Pacaraima, or according to local political tensions; for example, in 2019, several conflicts in the region caused the closure of the border, with a decrease in the number of imported malaria cases from 2137 cases in 2018 to 268 cases in 2019. However, there has been an increase in autochthonous malaria cases in the indigenous communities (1577 cases) [1]. São João da Baliza is located in the south of the state (0º57′03'' N; 59º54′39'' W), bordering Amazonas state, at a distance of 326 km from Boa Vista. The municipality is sparsely populated, with an estimated population of 8,201. In 2019, São João da Baliza reported 634 malaria cases; 461 were autochthonous, of which 310 were acquired in the urban area of the municipality, equivalent to 66.2%, against 151 in the rural area. Imported malaria cases were low, with only two cases from Venezuela and French Guiana [1].
Study design and larval habitat collection larval and adult mosquitoes were collected every two months from January 2017 to October 2018. In each municipality, four to six mosquito larval habitats were selected and sampled for larvae and pupae, and two residences near the larval habitats were selected for adult mosquito collection (Fig. 2). The larval habitats were selected based on their proximity to houses where autochthonous cases occurred the year prior to the study and were positive for Anopheles spp. larvae in a preliminary survey conducted in October 2016. For each larval habitat, the following physicochemical and environmental parameters were determined: habitat type, size, water temperature, water pH, degree of water turbidity, sun exposure, stream current, and vegetation. GPS estimated the distance from the larval habitat to the nearest house. Millimetric rods were used to measure the depth and size of the water bodies. The pH was determined using coloured dipsticks (MerckTM), and sun exposure and stream currents were visually estimated. Each larval habitat was identified by a number and georeferenced to determine its location (Fig. 2). In Boa Vista, the four larval habitats presented surrounding vegetation, and their geographical landscapes were characterized by the existence of savannah with and without gallery forests. In Pacaraima, 83% of the six larval habitats were located in the mountainous region of an ombrophilous forest. In São João da Baliza, 80% of the six larval habitats were surrounded by lowland tropical rainforest vegetation in a deforested area [38,39].

Larval surveys
Larval surveys were performed on the same days as adult collections, from 07:00 to 13:00 h and in the afternoon, when necessary. Larval collection was performed using a 350-mL manual dipper (BioQuip, Ranch Dominguez, CA, USA) following a standardized larval sampling methodology recommendation of the Health Surveillance Secretariat of the Ministry of Health. The larvae and pupae collected were transferred to plastic tubes labelled with the larval habitat code, location, date, and collector's name. Larvae and pupae were either placed in 80% ethanol or reared in the laboratory to fourth instars or to adults, respectively, for identification using in Consoli and Lourenço-de-Oliveira dichotomous key for anopheline larvae [40].

Adult survey
Adult mosquitoes were collected in January, April, July and October of 2017 and 2018, comprising dry (November to April) and rainy (May to October) seasons. In each municipality, the collections were performed simultaneously at two sites, for three consecutive days at sunset from 18:00 to 21:00 and compared in the intradomicile, peridomicile and extradomicile areas. Sampling efforts throughout the studied period consisted of 480 h in each site: 144 h in the intradomicile (Nasci aspirator), 144 h in the peridomicile (MosqTent), and 192 h in the extradomicile (96 h with MosqTent and 96 h with Shannon).
In the intradomicile, mosquitoes were collected at rest on the internal walls and furniture, inside the residence, using a Nasci aspirator [41]. Peridomicile collections were performed approximately 15 m from the house using a protected human-baited mosquito trap (MosqTent) [42]. Extradomicile collections were carried > 15 m from the residence and the main larval habitat using MosqTent. When a horse was available, a baited Shannon trap was also used in the extradomicile concomitantly with the MosqTent collections. In order to assess the peak of haematophagy, collections were also performed in the intradomicile and extradomicile, for a 12-h period, from 18:00 to 06:00, in the months of April (transition from dry to rainy season) and October (transition from rainy to dry season) of 2017 and 2018. All data in relation to each collection (temperature, humidity, wind, rain) were recorded. Mosquitoes were collected using a manual Castro aspirator, placed in paper cages and transported to the Laboratory of the State Entomology Centre in an appropriate Styrofoam box. The specimens were then euthanized with ethyl acetate. Individuals were transferred to plastic tubes (Eppendorf type) and kept in boxes containing hermetically sealed silica gels [40].
Species were morphologically identified using dichotomous keys for the Neotropical Anopheles species. Members of the An. albitarsis, An. nuneztovari and An. triannulatus complexes were not identified in this study; these species are referred to as sensu lato. Daily and monthly rainfall data were obtained from the National Water Agency (Agência Nacional de Águas, ANA).

Geographical location of studied sites
The geographic locations of larval habitats and adult collection sites were georeferenced in the Universal Transverse Mercator (UTM) coordinate system using the World Geodetic System 84 (WGS84). All geographic data were processed in the Landscape Metrics Laboratory of the Department of Geography of the Federal University of Roraima (UFRR), using satellite images and a cartographic basis of the municipalities. The data were processed in the QGIS 2.14.17 software to generate the maps such as those previously described [8].

Statistical analysis
Statistical analyses were performed using GraphPad InStat, version 3 (GraphPad Software, San Diego, CA, USA) and R version 4.0.5. Comparisons between mean temperature and pH were performed using the Wilcox test. To evaluate the association of environmental factors with the occurrence of species in larval habitats, univariate logistic models was used with species as the outcome. In particular, the environmental factors (habitat type, area, water turbidity, water current, water pH, water temperature, house distance, sunlight exposure, and debris) were used as independent variables. The associations between species and environmental variables were presented as odds ratios (ORs) and their 95% confidence intervals. Differences between the two non-reference categories were related to reference categories.

Characteristics of larval habitats
Anopheles spp. larvae were found in all 15 permanent larval habitats in the studied municipalities: four sites in Boa Vista, six in Pacaraima, and five in São João da Baliza. Table 1 shows the main characteristics of each larval habitat and larvae of Anopheles spp. were found in all four types of water collection. Streams were widely distributed (66.6%) in the study area, followed by dams (13.3%), ponds (13.3%) and lakes (6.6%). In Boa Vista,

Identification of immature forms of anophelines
A total of 544 Anopheles spp. specimens were obtained from 15 larval habitats, and the relative abundance of each larval species encountered is presented in Table 2.
A total of 12 species were identified from 209 late-instar larvae (third and fourth stages). However, 335 Anopheles spp larvae were not identified at the species level and considered for statistical analysis, mainly because they were early instar larvae (first and second stages) or were damaged. An. albitarsis s.l. (8.8%) were also prevalent. Pacaraima showed a total of 148 larvae of seven species collected, and An. triannulatus s.l. (33.3%) was the most abundant species, followed by An. albitarsis s.l. (24.2%) and An. argyritarsis (18.1%). All An. triannulatus s.l. were collected from a single larval habitat (5) in a fishpond. Immature forms of An. darlingi were present in streams, lakes, and dams with transparent or turbid water, in partially shaded or sunny habitats. However, in these larval habitats, the water had a lower pH and higher water temperature.

Association between anopheline larvae and environmental factors
The univariate logistic models for the four most frequent Anopheles species are summarized in Table 3. Anopheles darlingi was more likely to occur in the stream larval habitat (OR = 6.85, p < 0.01) compared with An. nuneztovari s.l., which was found in a dam (OR = 0.16, p < 0.01). Anopheles triannulatus s.l. had a preference for fishponds (OR = 40.7, p < 0.01). Larvae of An. albitarsis s.l. and An. darlingi had a higher chance of being captured in larval habitats with 150-300 sq m of surface area than those larger than 300 sq m (OR = 3.83, p < 0.05, and OR = 2.27, p < 0.05, respectively. Anopheles darlingi was also less abundant in stagnant water (OR = 0.08, p < 0.01) and slow currents (OR = 0.12, p < 0.01) than in moderate water currents. The logistic models also showed a higher chance of An. triannulatus s.l. to be found in larval habitats > 300 sq m (OR = 0.13, p < 0.01) than those < 150-300 sq m (OR = 0.29, p < 0.05). Anopheles albitarsis s.l. had a greater preference for aquatic habitats with transparent waters over turbid water (OR = 0.24, p < 0.01). In contrast, An. darlingi and An. nuneztovari s.l. occurred in aquatic habitats with both transparent and turbid water. Habitats exposed to the sun were related to the occurrence of An. albitarsis s.l. (OR = 2.93, p < 0.05), and An. triannulatus s.l. (OR = 15.67, p < 0.01) in comparison to those partially shaded, while An. darlingi showed a preference for partially shaded over full sun habitats (OR = 0.1, p < 0.01). Anopheles darlingi was more frequently observed in the presence of roots, tree trunks and leaves (OR = 12.58, p < 0.01) than in the absence of debris, while An. triannulatus s.l. was less likely to occur in habitats with debris (OR = 0.08, p < 0.01).

Species distribution of adult anopheline
A total of 1,488 female Anopheles from 11 species were collected from all three municipalities (Table 4). Nine species were identified in Boa Vista and An. albitarsis s.l. (88.4%) was the predominant species, followed by An. darlingi (6.9%) and An. peryassui (3%). Among the seven species collected from São João da Baliza, An. darlingi (85.6%) was the most predominant species, followed by An. albitarsis s.l. (9.3%), and An. nuneztovari s.l. (3.9%). Pacaraima showed lower vector density than the municipalities of Boa Vista, and of the six species identified, the most abundant was An. braziliensis (64.6%), followed by An. peryassui (18.7%) and An. albitarsis s.l. (8.3%). In all three municipalities An. albitarsis s.l., An. darlingi and An. nuneztovari s.l. were present, whereas An. matogrossensis and An. evansae were observed only in Boa Vista and An. triannulatus s.l. only in São João da Baliza. Overall, the majority of anophelines exhibited greater extradomicile than peridomicile biting preferences. Anopheles darlingi was the only species found in the intradomicile. In Boa Vista and São João da Baliza An. darlingi and An. albitarsis s.l. were more abundant in the extradomicile than in the peridomicile. However, in Pacaraima and Boa Vista An. darlingi and An. albitarsis s.l. were also collected in the Shannon trap. In Pacaraima, An. braziliensis was captured in the peridomicile (33.8%), and in the extradomicile by MosqTent (29%) and by the Shannon trap collection (37.1%).

Seasonal variation of Anopheles in each municipality
The longitudinal study showed variation in the abundance of Anopheles species according to the period studied (Fig. 3). The collection each month represents the mean number of Anopheles captured in that month per collector per hour (4 h of capture for three consecutive days) for every two months in each locality.
The rainy season in the state of Roraima occurs between April and September. Even so, in Boa Vista this was the period with the greatest abundance of Anopheles. In Pacaraima and São João da Baliza these species were more frequent in the driest months. In Boa Vista (Fig. 3A) An. albitarsis s.l. had the highest density during the studied months, with the greatest frequency observed in April in the dry season. The month of July for both years (2017 and 2018) had the greatest rainfall, and a large number of An. albitarsis s.l. were captured in 2018. Anopheles darlingi was not abundant during the study months but showed a population increase in October at the end of the rainy season.
In Pacaraima (Fig. 3B), An. braziliensis was the most abundant species collected in higher numbers in October, two months after the high record for rain precipitation, while An. albitarsis s.l. were collected in fewer numbers during most of the months of collection. In São João da Baliza (Fig. 3C), An. darlingi was predominant throughout the year, and this species was more abundant in October, whereas An. albitarsis s.l. was also collected in every month but in lower numbers.
The dispersion of Anopheles by average rainfall intensity and temperature showed that An. darlingi was predominant when the temperature was above 24ºC and in the periods with less rainfall, while An. albitarsis s.l. prevailed during periods of rain, with temperatures above 23ºC. Anopheles braziliensis was collected at lower temperatures (18ºC and 25ºC) during both the dry and rainy periods. Anopheles peryassui was the vector that was intermediate to all these periods and was not captured in the extremely dry period (Additional file 1: Fig. S1).

Biting activity and human bite rate
Four 12-h collections were conducted in each municipality simultaneously in the intradomicile and peridomicile to analyse biting activity. Although all collections were performed in both environments, no samples were collected in the intradomicile. Figure 4 shows the frequency (%) of the two more frequent Anopheles species in each locality occurring in the peridomicile at hourly intervals. Variability in biting times was observed among species and municipalities. In Boa Vista, An. albitarsis s.l., the most abundant species, presented biting activity at dusk and throughout the night, with most of the activity between 02:00 and 04:00. Anopheles darlingi presented biting activity at 20:00-21:00 and throughout the night, but in lower numbers with the majority between 20:00 and 21:00 and 02:00-3:00, with only one specimen of An. peryassui captured at 20:00-21:00. In Pacaraima, the most frequently occurring species was An. braziliensis, with a significant peak at 19:00-20:00. Anopheles argyritarsis was caught between 18:00 and 19:00 and 19:00-20:00, and An. peryassui at 18:00-19:00. In the municipality of São João da Baliza, An. darlingi exhibited a pronounced crepuscular biting activity with an early peak at 21:00-22:00. However, they were also collected throughout the night until 06:00. Anopheles nuneztovari was the second-most common species, occurring from 20:00 to 21:00, 21:00 to 22:00 and from 23:00 to 00:00, while An. albitarsis s.l. occurred in higher numbers at 05:00 to 06:00.
The human biting rate (HBR) was estimated for the most common species collected in each municipality using the following calculation: the total number of anophelines captured during the 4-h collection (16:00 to 21:00) divided by the total number of collection days and the number of collectors. The data are presented in Table 5. Among all species collected in Boa Vista, An. albitarsis s.l. showed the highest rate in the extradomicile (10.8 bites per night) followed by 3.8 in the peridomicile.
The presence of An. darlingi predominates with a low HBR in the peridomicile (0.3 bites) and in the extradomicile (0.8 bites). In Pacaraima, HBR was low in both the peridomicile (0.26) and extradomicile (0.22) for An. braziliensis. In the municipality of São João da Baliza, An. darlingi presented higher HBR in the peridomicile (1.07) and in the extradomicile (1.64) than in Boa Vista while An. albitarsis s.l. had a very low HBR. Moreover, São João da Baliza was the only municipality where An. darlingi Table 2 The relative abundance of each larvae species collected at larval habitats in the municipalities of Boa Vista, Pacaraima and São João da Baliza -Not done *Anopheles spp.: Anopheles that were not identified at the species level, mainly because they were early instar larvae (first and second stages) or were damaged   21:13 was collected in the intradomicile (three specimens at 19:00-20:00 and one at 20:00-21:00).

Discussion
Malariogenic potential measures the risk of transmission in a given area. It is critical for determining strategies to eliminate and prevent the re-establishment of malaria transmission [43], particularly in areas where the potential of imported malaria is high. As part of an integrated strategy for control and elimination, interventions should be deployed regionally based on detailed knowledge of local vector habitats and distribution. Here, during two years, the larval habitat and adults were sampled to determine the Anopheles species composition, seasonal occurrence and distribution of vectors in areas of autochthonous (São João da Baliza) and with high numbers of imported (Boa Vista and Pacaraima) malaria in Roraima. Differences in mosquito distribution and richness among sites may be the most important factors for identifying local variation in malaria transmission. Roraima presents a diverse ecological landscape and malaria transmission in Boa Vista, which is mainly savannah, which differs from that of the Yanomami Indian reserves in the northwestern mountains, which is a region with dense rainforest [24-26, 38, 44].
In all municipalities, the most-encountered larval habitats were streams. Anopheles darlingi was the most abundant larval species in São Joao da Baliza and Boa Vista. However, in São João da Baliza, where vegetation consists of a lowland tropical rainforest in a deforested area, An. darlingi occurred at all five larval habitats, while in Boa Vista, a savannah vegetation, this species was present in only two larval habitats and in lower numbers. The differences in the ecological environment between Boa Vista and São João da Baliza may account for this [38]. Studies have shown that An. darlingi is often associated with low-lying vegetation along forested river margins and deforestation provides favourable conditions for this vector, thereby increasing malaria cases and decreasing scores of the Human Development Index [14,[45][46][47][48].
Anopheles darlingi has been found in a variety of larval habitats, both natural and man-made, and from large and permanent to small and temporary water bodies [13,[46][47][48]. In Boa Vista and São João da Baliza, An. darlingi were collected mainly in aquatic habitats with transparent water and moderate current with temperatures  [14], the immature stages of An. darlingi develop in aquatic habitats partial shaded, with temperature and pH varying, respectively, between 20 and 28 ℃, and 6.5 and 7 [14]. In Pacaraima, a mountainous region of an ombrophilous forest, An. darlingi larvae were not observed but one adult mosquito was collected. This species is not normally found in areas with high altitudes and in low temperatures [40]. However, recently it was collected at altitudes above 800 m in Venezuela, close to the border of Roraima [13]. Instead, An. triannulatus s.l. was the most abundant species collected in a single fishpond exposed to the sun that was located in a horticultural area and also in dams and streams in savannah and tropical forest larval habitats. Anopheles triannulatus s.l. seems to be a generalist as it demonstrated both widespread distribution and little if any environmental constraints [20,49,50]. This species appears to play an important role in malaria transmission in the states of Amapá, Amazonas, Pará and Peru [21,23,51,52].
Anopheles albitarsis s.l., An. nuneztovari s.l., An. triannulatus s.l., An. oswaldoi s.l., and An. peryassui were found in all municipalities but in low numbers. In addition to the composition of anopheline larval species, a total of 1488 adult female Anopheles of 11 species were collected from all three municipalities. In São João da Baliza, where autochthonous malaria occurs, the most abundant species was An. darlingi (85.6%), while in Boa Vista, the municipality with a high number of imported cases, An. albitarsis s.l. (88.4%) was predominant, while An. braziliensis (64.6%) was abundant in Pacaraima. Except for An. peryassui, all species collected have been  implicated as malaria vectors in the Brazilian Amazon [21,53,54]. Previous studies in Boa Vista corroborate this report regarding the composition of species found, with a greater abundance of An. albitarsis s.l. and a lower density of An. darlingi [24][25][26]. Anopheles albitarsis s.l. also had the highest HBR with 10.8 bites in the extradomicile and 3.8 in the peridomicile. However, in the 12-h collection period it presented biting activity throughout the night, with the highest biting peak between 02:00 and 04:00, which differs from that previously reported [24][25][26]. Anopheles albitarsis s.l. has been assigned as an important vector in Para, Roraima, Amapá and Rondonia [23,27,54]. Because molecular analysis between members of An. albitarsis, An. nuneztovari, An. triannulatus, and An. oswaldoi complexes were not defined in this study, the presence of An. janconnae or other species from this complex could not be confirmed in the studied municipalities. Presently, the only species of the An. albitarsis complex identified in the state of Roraima, has been An. janconnae, considered an important malaria vector in savannah habitat around Boa Vista [26]. Entomological surveillance to identify the component members of these complexes are mandatory to update the geographic distribution, ecological and biological aspects, and its possible role as a regional vector of malaria parasites in Roraima.
In Boa Vista, An. darlingi were collected in smaller numbers from 20:00 to 21:00 and 02:00 to 03:00, and its presence emerges with low HBR of 0.3 bites in the peridomicile, and 0.8 bites in the extradomicile. The presence of An. darlingi, even at low density in Boa Vista, can significantly contribute to transmission because it is highly anthropophilic and susceptible to Plasmodium parasites. Minimal biting indices could be sufficient to maintain the transmission cycle in Boa Vista [24,51].
In São João da Baliza, An. darlingi was the most abundant species throughout the year, with a pronounced crepuscular biting activity in the peridomicile and extradomicile with early peak at 21:00-22:00 and throughout the night until early morning. Tadei et al. also found similar biting activity, whereas in Amapá this species showed a wide range of blood-feeding pattern [53][54][55]. In Venezuela, this mosquito bites throughout the night, with minor peaks at 23:00-00:00 and 03:00-04:00 [56]. The time of biting activity of anophelines depends on several factors, such as species density, seasonality and host availability [53,57]. Regarding seasonality, there was no significant correlation between abundance (of the species) and rainfall or the abundance of each species with locality or rainfall (p > 0.05). The lack of correlation between anopheline species and rainfall could be due to the variability in mosquito abundance among localities. In Boa Vista, the highest density of An. albitarsis s.l. occurred during the dry season but it was also collected in the period of high rainfall in 2018. Anopheles darlingi showed a peak after the rainy season (transition from rainy to dry season), in both years of collection, and a slightly lower peak in the middle of the rainy season, only in 2018. In Rondonia, genetic differences between the populations of An. darlingi were characterized by microsatellites with distinct seasonal patterns [58], one of which, population A, presented the highest density at the end of the rainy season, and was characterized as a less abundant population in interior regions than in riparian areas. Another sub-population, population B, peaked at the beginning of the rainy season and was more abundant in the interior.
The HBR per night for An. darlingi was also considered high in this municipality with 1.64 bites in the extradomicile and 1.07 in the peridomicile. This likely contributes to transmission in this municipality. Heterogeneous biting behaviour of An. darlingi in terms of blood feeding inside and outside houses and variations in the peak time   21:13 of biting have been shown by numerous studies carried out across the Amazon and in Venezuela and French Guiana [15,40,47,51,56,59,60]. This behaviour is considered the main cause of the ineffectiveness of insecticide sprayed inside houses [61,62]. One of the control measures used in the malaria control programme in Roraima is indoor residual spraying, which may explain the exophilic behaviour of this species in the studied area.
In addition to the presence of An. darlingi, An. nuneztovari s.l. and An. albitarsis s.l. were observed in the peridomicile and the extradomicile, but in low numbers compared to that of An. darlingi; An. nuneztovari s.l. was the second-most abundant species in São João da Baliza. This species is considered a primary malaria vector in Venezuela, Peru and Colombia [13,[63][64][65]. In Brazil, An. nuneztovari s.l. has been found naturally infected with P. vivax and P. falciparum in the Amazonian region and it has been incriminated as local vector in Amapá [21,35,53,54]. However, most of Brazilian populations of An. nuneztovari s.l. are predominantly zoophlilic and seem not to sustain malaria transmission in the absence of the primary vector An. darlingi [15,66]. In the urban areas of Roraima, studies have shown a low population density and a low number of infectious An. nuneztovari s.l. [25].
The abundance of An. darlingi throughout the year in São João da Baliza, along with their propensity to seek hosts throughout the night, and their ability to adapt host-seeking behaviour to local environments contributes to their impact as the most important vector in this municipality. In contrast in Pacaraima, since An. braziliensis was the predominant species, most adults were collected in the Shannon trap, where a horse was used for attraction. Anopheles braziliensis is present in most Amazonian states and is considered a secondary vector because it is primarily zoophilic and exophilic and are rarely involved in malaria transmission [25,40]. Consoli and Lourenço-de-Oliveira [40], claimed that An. braziliensis can be found biting during the day, especially when the host is relatively close to its larval habitat [40]. Importantly, it has been found to be infected with human malaria parasites in the states of Amazonas, Amapá, Rondônia and Roraima [15,19,54,67]. Only one specimen of An. darlingi was collected in Pacaraima, but there is a report predicting the presence of this vector in indigenous areas in this municipality, where most of the autochthonous cases occur [1,68]. Taken together, An. braziliensis may play some role in malaria transmission in Pacaraima when at high densities.

Conclusion
This study showed a diversity of anopheline larvae species and habitats in Boa Vista, Pacaraima and São João da Baliza. Anopheles darlingi appeared to be the most important vector in São João da Baliza, the area of autochthonous malaria, and An. albitarsis s.l. and An. braziliensis in areas of low transmission, although there are increasing reports of imported malaria. Considering the behaviour of the vectors in Boa Vista, Pacaraima and São João da Baliza, interventions such as intradomiciliary spraying will likely be insufficient to reduce malaria transmission. It can be speculated that the local vector assemblage in Boa Vista is sufficient to sustain the disease, and also provides a time buffer to mitigate the effects that imported cases would otherwise have if An. darlingi were more prevalent. This circumstance offers an opportunity to reduce the effects of cross-border malaria via early diagnosis and timely treatment. In addition, environmental management of vector larval habitats and health education actions addressing individual and collective forms of prevention are indicated.