Malaria Vectors in Ecologically Heterogeneous Localities of the Colombian Pacific Region

The Colombian Pacific region is second nationally in number of malaria cases reported. This zone presents great ecological heterogeneity and Anopheles species diversity. However, little is known about the current spatial and temporal distribution of vector species. This study, conducted in three ecologically different localities of the Pacific region, aimed to evaluate the composition and distribution of Anopheles species and characterize transmission intensity. A total of 4,016 Anopheles mosquitoes were collected representing seven species. The composition and dominant species differed in each locality. Three species were infected with malaria parasites: Anopheles darlingi and An. calderoni were infected with Plasmodium falciparum and An. nuneztovari with Plasmodium vivax VK210 and VK247. Annual EIRs varied from 3.5–7.2 infective bites per year. These results confirm the importance of the primary vector An. nuneztovari in areas disturbed by human interventions, of An. darlingi in deforested margins of humid tropical rainforest and An. albimanus and the suspected vector An. calderoni in areas impacted by urbanization and large-scale palm oil agriculture close to the coast. This constitutes the first report in the Colombia Pacific region of naturally infected An. darlingi, and in Colombia of naturally infected An. calderoni. Further studies should evaluate the epidemiological importance of An. calderoni in the Pacific region.


Introduction
The Colombian Pacific (PAC) region is second in the number of national malaria cases reported [1], registering an average of 30% of the total cases [2][3][4] with the majority of cases due to infection with Plasmodium falciparum Welch (.60%) [1]. PAC is located in the biogeographic Chocó zone [5] and presents diverse ecological and climate conditions with five natural subregions [6] that are predicted to affect Anopheles vector species distribution and behavior [7].
Although Anopheles (Nys.) darlingi is considered the main vector in the Chocó rainforest [11,12], there have been no published records of naturally infected An. darlingi in this area. Anopheles (Kerteszia) neivai Howard, Dyar & Knab, a species closely associated with forest bromeliads and mangroves [13], is a locally important vector in PAC and has been reported infected with P. falciparum [13] and P. vivax [8,14]. Existing records also suggest an epidemiological association of Anopheles (Anopheles) punctimacula Dyar & Knab with malaria outbreaks in this area [15,16]; this species was detected with salivary gland sporozoites of P. vivax in NW Colombia [17]. Various members of the Arribalzagia Series such as An. punctimacula, Anopheles malefactor Dyar & Knab, Anopheles calderoni Wilkerson and Anopheles guarao Anduze & Capdevielle, are characterized by a high degree of isomorphism that complicates their differentiation and accurate identification by morphological characters [18][19][20]. For example, in mosquito studies in PAC, An. punctimacula was the species recorded; however, An. calderoni was recently reported in various PAC localities, as supported by adult and larval diagnostic characters and mosquito barcoding [21]. In Piura Department in neighboring Peru, An. calderoni was infected with P. vivax and considered a locally important vector [22,23]. Collectively, these observations suggest the need for further evaluation of the identity of these species and their involvement in malaria transmission.
Most vector biology studies in PAC have focused on distributions and parasite infection of the Anopheles species present, but little is known about variation in vector densities and transmission intensities at spatial and temporal scales. Therefore, a longitudinal survey was conducted from April 2009 to June 2010 in three ecologically diverse localities of the epidemiologically important PAC region to evaluate species composition and distributions at spatial and temporal levels and to determine important entomological parameters.

Study sites
Adult Anopheles were collected from three ecologically diverse malaria endemic localities of the PAC region (Table 1, Figure 1); they were San Antonio de Padua (SAP), Zacarias (ZAC), and Carbonera and Pindales (CAR-PIN). SAP is in the PAC natural subregion Alluvial Valleys of the Atrato and San Juan rivers which is characterized by tropical forest [6]. The main economic activities in SAP are fishing, small-scale agriculture, especially rice and logging that is displacing the forest borders (Figure 2A-B). ZAC is in the PAC Coastal Plains subregion characterized by flooded plains, lakes and swamps [6] and presents one of the highest levels of precipitation in the world (.5,000 mm annually), with rainfalls occurring most of the year [6]. In ZAC, deforestation for human settlements, river extraction of building materials and pisciculture are the most important activities ( Figures 2C-D). CAR and PIN are separated by 9 km, and, like ZAC, are located in the Coastal Plains subregion. Both localities are heavily impacted by anthropic changes. CAR is located approximately 3 km from the coast and is a periurban neighborhood of the municipality of San Andres de Tumaco, where urbanization has replaced wetlands and mangroves ( Figure 2E). The first collection for these localities was performed in CAR. However, because of problems of public order, the following three collections were completed in PIN, located approximately 11 km from the coast and characterized by large-scale palm oil agriculture ( Figure 2F).

Mosquito collection
Each locality was visited four times, on average once every three months, from April 2009 to June 2010. Mosquitoes were collected indoors and outdoors (within ,10 m of the house) by human landing catch under a protocol and informed consent agreement approved by a University of Antioquia Institutional Review Board (Comité de Bioética, Sede de Investigación Universitaria, CBEIH-SIU, approval number 07-41-082). Collections were conducted by two people in each setting, for five days, from 18:00 to 24:00 h and one additional night from 18:00 to 06:00 h. Adult mosquitoes were identified using morphology-based keys [24][25][26]. Species presenting difficulties during morphological identification were confirmed by PCR-RFLP-ITS2 [27][28][29] and specimens An. calderoni by COI barcoding [30] using primers by Folmer et al. [31] and standardized conditions [32]. In addition, potential larval habitats within approximately 1 km of adult collection sites were inspected to identify those positive for immature stages and third and fourth instar larvae were identified. Mosquito vouchers were deposited in the collection of ''Grupo de Microbiología Molecular, Universidad de Antioquia'', and An. calderoni COI sequences were deposited in GenBank (Accesion numbers KF698816-KF698832).

Detection of natural Plasmodium infection
Natural infections with P. falciparum and P. vivax (VK210 and VK247) were determined by ELISA [8,33,34], using pools of heads and thoraces of up to five specimens of the same species. The ELISA was used as a screening test to evaluate a large number of mosquitoes. Positive pools in the ELISA were confirmed by a second ELISA. Nested genus specific PCR using DNA extracted from individual abdomens was performed to confirm ELISA results and to incriminate the individual specimen infected [35].

Data analysis
Infection and entomological risk parameter analyses were described previously [36]. Briefly, the infection rate (IR) was calculated as the percentage of Plasmodium positive mosquitoes out of the total number of mosquitoes analyzed by species and locality. The human biting rate (HBR) was expressed as the average of female Anopheles bites per person per night. The annual EIR for each site corresponds to the number of infective bites that a person may receive in one year. Spearman's correlation coefficient in the SPSS program, version 18 (SPSS Inc., Chicago, IL) was used to estimate the relationships between rainfall data of the previous month to a collection and mosquito abundance; the monthly rainfall records for each locality were obtained from the Instituto de Hidrología, Meteorología y Estudios Ambientales (IDEAM).

Mosquito abundance
A total of 4,015 Anopheles mosquitoes corresponding to seven species were collected during 467 h of sampling (Table 1). There was not a significant (p.0.05) correlation between mosquito abundance and rainfall for any species. Species composition and dominance varied in each locality. In SAP, An. darlingi dominated (98.6%), and its highest abundance was observed during a period of moderate rainfall ( Figure 3A). The main vector, An. nuneztovari was collected in low abundance in two samplings coinciding with the increase of rainfall in this locality ( Figure 3B). Anopheles punctimacula was detected in low abundance in two consecutive collections (Table 1), with peak abundance at the beginning of the rainy season (Table 1, Figure 3B). Anopheles costai/forattinii was also collected by human biting but only in the last sampling period. Anopheles-positive larval habitats were rice irrigation channels and small ponds near forest borders, both sustained by water from the Atrato's river ( Figure 2A and 2B, respectively). The species detected in these larval habitats was An. darlingi. In one pond near the forest border, An. malefactor, a species not detected as an adult, co-occurred with An. darlingi.
In ZAC, An. nuneztovari was the dominant species (90%), detected in all collections; peak abundance was at the onset of a rainy period, when it was the only species collected (Table 1, Figure 3C). Anopheles neivai was only collected in this PAC locality, during the three initial samplings, always in low abundance. It was not detected in the fourth collection when there was a decrease in rainfall (average 900 to 300 mm) ( Table 1, Figure 3D). The Anopheles-positive larval habitats in ZAC were fishponds ( Figure 2D), swamps and temporal flooded ponds and the only larvae collected corresponded to An. nuneztovari; bromeliads were not sampled.
The dominant species in CAR was An. albimanus (97.3%), and in PIN, An. calderoni (87.21%). In the three collections in PIN, An. calderoni always outnumbered An. albimanus and its peak abundance corresponded to the highest rainy period in this locality (Table 1, Figure 3E). In the analyses of species distribution and peak abundance performed independently or together for specimens from CAR and PIN there were no significant differences, therefore these localities were presented as one (CAR-PIN). In CAR, An. albimanus-positive larval habitats were wetland drainage channels near houses ( Figure 2E

Human biting rate (HBR)
During the collection of April 2009 in SAP, An. darlingi exhibited the highest HBR among all species collected in any locality, 89.9 bites/person/night (b/p/n; Table 1). HBRs for An. nuneztovari were less than 10% of the highest HBR for An. darlingi; for example in ZAC this ranged between 2.7 and 8.6 b/ p/n; in SAP, HBRs were between 0.2 and 1.3 b/p/n (Table 1). Particularly, An. nuneztovari was not detected in CAR-PIN where An. calderoni and An. albimanus were the only species collected in the four visits; furthermore, these species were exclusive to CAR-PIN. Anopheles calderoni exhibited its highest HBR in April 2010  Infectivity and entomological inoculation rates Four infected specimens were detected in the three PAC localities (Table 1). Anopheles darlingi in SAP and An. calderoni in CAR-PIN were infected with P. falciparum (IR = 0.036% and 0.194%, respectively). In ZAC, two An. nuneztovari were infected, one with P. vivax VK247 and the other with P. vivax VK210 (IR = 0.203% each). The annual EIR ranged from 3.5 infective bites per year (CAR-PIN) to 7.2 infective bites per year (ZAC), the highest recorded for this study (Table 1). Transmission intensity peaks or periods where Anopheles mosquitoes were detected  infected with Plasmodium usually corresponded to low rainfall periods, except when An. calderoni was detected infected in PIN, during a rainy period (Table 1, Figure 3), that was also the period of its peak abundance.

Discussion
Anopheles species dominance and abundance are influenced by environmental factors such as rainfall, land cover and topography [37][38][39]. In this longitudinal study in ecologically different localities of the malaria epidemiologically important PAC region, the influence of rainfall on Anopheles species abundance and diversity was analyzed and the main ecological characteristics of the localities described. In addition, the occurrence of the three main Colombia malaria vectors, An. darlingi, An. nuneztovari and An. albimanus was confirmed in PAC. However, the distribution and importance of these species as malaria vectors varied among localities.
For San Antonio de Padua locality, in a subregion characterized by tropical forest [6], An. darlingi was the most abundant and dominant species. In this locality An. darlingi-positive breeding sites were flooded, deforested forest margins and terrains used by the villagers for personal rice cultivation. Such ecological conditions and type of larval habitats favor the presence of this important vector [40][41][42]; i.e., in the Peruvian Amazon, An. darlingi adults dominate in deforested environments [43]. In SAP, the highest HBR for An. darlingi was during a moderate rainy period, but it is notable that this locality is characterized by persistent rain throughout the year. The HBR indicates that a person in SAP might receive 2,697 An. darlingi bites in a month, suggesting a higher likelihood of malaria transmission by this important vector. In a NW Colombian endemic area where malaria incidence is twice that of PAC, An. darlingi entomological parameters varied according to the locality and higher HBRs were observed at the beginning or end of rainy periods [36]. Data from this and other studies suggest that rainfall is a key factor affecting An. darlingi presence and dominance. For example, high HBRs have been correlated with low rainfall in Brazil [44,45] and Venezuela [46], but also with rainy periods in the Amazonian forest of French Guiana [47] and Brazil [48]. These different responses to rainfall regimes combined with An. darlingi-positive larval habitats that had specific ecological conditions, suggest the importance of land cover [43,49] and probable interaction effects among rainfall, ecological and land cover parameters on abundance and distribution of this main malaria vector in endemic regions of Colombia.
Annual EIR indicated that one person may receive one An. darlingi infective bite every 2.3 months. This EIR was higher than previously detected for this vector in the most important Colombian malaria endemic region, the Urabá Bajo Cauca and Alto Sinú (UCS) [36], indicating that An. darlingi is also important in malaria transmission in impacted habitats near rainforest. The infected An. darlingi specimen was detected when the rains were increasing after very low rainfall, in agreement with reports of high malaria transmission in Colombia, usually related to low or moderate rainfall periods [50]. In SAP, An. darlingi showed a biting peak between 19:00-22:00 h with a slight tendency for biting outdoors after 20 h. This behavior would be expected to increase infection risk since residents are active outdoors after work (N. Naranjo, personal communication), increasing human-vector contact [46][47][48].
In contrast, in SAP, the main Colombian vector An. nuneztovari was detected in low abundance, with HBRs #1. The highest density for this species was observed at the beginning of a rainy period when larval habitats are more stable compared to during heavy rainfall [51]. In Colombia, An. nuneztovari is characterized as being more tolerant of human impacted zones than An. darlingi, and it was previously detected in higher abundances in localities of the epidemiologically important UCS, NW Colombia [36,52]. Also in SAP, An. punctimacula, a species of local importance, was present when the rain began to increase and showed even lower HBR values than An. nuneztovari. The presence of An. punctimacula during periods of low rains was also observed in NW Colombian localities [36], and its low abundance may be related to the collection methodology used in relation to its reported zoophilic tendency [53,54].
In ZAC, in the Coastal Plains subregion of central PAC, An. nuneztovari and An. neivai were the only anthropophilic species detected. This locality is characterized by flooded plains and strongly influenced by deforestation and pisciculture. These human activities provide optimal larval habitats for An. nuneztovari [48], the dominant species during the entire sampling in ZAC. However, An. nuneztovari abundances were lower during the most intense rainfall period, coinciding with a decrease of Anopheles-positive larval habitats. Rainfall in this region is very high, with 6,980 mm annual average [6], which may cause flooding of the breeding sites affecting this species' abundance. HBRs for An. nuneztovari ranged from 2.7 to 8.6 bites per person per night. Even lower HBRs have been previously reported for this species in Buenaventura municipality in PAC [55] and in other malaria endemic Colombian regions [36,52,56]; regardless of these observations, An. nuneztovari has previously been incriminated as a vector [36,52].
In this study, the highest EIR was registered in ZAC for An. nuneztovari (7.5 infective bites per year), indicating that a person may receive one infective bite every one and a half months. This information, the fact that the two infected An. nuneztovari were detected in the period of its maximum abundance, together with the significant preference for biting indoors, suggest an increase in malaria risk for humans when they are resting at home (21:00 to 24:00 h). This information is of importance to reduce exposure and risk, especially inside the house, for example by the use of insecticide-treated mosquito nets, one of the main tools for vector control [57]. Various characteristics in ZAC favor the presence of An. neivai, a local vector in PAC [13,58]. The presence of this species has been correlated to high humidity, lowland tropical forest and a variety of epiphyte plants that serve as its breeding sites [13]. However, in ZAC An. neivai was detected in low abundance (HBRs ,1). It is possible that in-progress deforestation in this locality is reducing or eliminating preferred breeding sites of An. neivai. The low HBRs detected for An. neivai are similar to those previously reported in PAC for disturbed areas, compared with sylvatic conditions [7].
In CAR-PIN, localities impacted by anthropic changes, An. albimanus and An. calderoni were the only anthropophilic species detected. In particular, the dominant species in CAR was An. albimanus whereas in PIN it was An. calderoni. These differences may be related to their preferred larval habitats [7,59,60]. In CAR, mangroves and wetland drainage channels provide appropriate breeding sites for An. albimanus. The highest HBR for An. albimanus in CAR may be influenced by the proximity of this locality to the coast. Previous reports also show HBRs for An. albimanus increasing with proximity to the coast because of availability of its preferred larval habitats [7,55]. Variation in An. albimanus abundance was not related to rainfall; however, HBRs decreased with increasing rainfall, and probably floods affected the stability of larval habitats. In PIN, small irrigation channels of oilpalm agriculture and wells favored the dominance of An. calderoni, previously associated with extensive palm agriculture near the coast [20,61]. The highest HBR for An. calderoni was observed during the period of heaviest rainfall. Comparable HBRs have been reported for An. calderoni in other PAC localities, also during rainfall periods [62]. In PIN, high human-vector contact occurs from 18:00 to 24:00 h, with An. albimanus more active around sunset followed by An. calderoni from 21:00 h. This specific information should help direct local vector control measures. In PIN, the annual EIR indicated that a person might receive one An. calderoni infective bite every three months. This species was recently confirmed in Buga, a PAC municipality, where it was initially identified as An. punctimacula [21], a species of local importance in some PAC localities [58]. However, An. calderoni has not yet been incriminated as a vector in Colombia, although it is of some importance in Peru [22,23]. In this study, An. calderoni was detected infected with P. falciparum by the first ELISA and nested PCR but could not be confirmed by the second ELISA [63]. Given the difficulties still encountered in the tests to determine parasite infected mosquitoes, this result is reported here according to recommendation of defining a positive infected mosquito with the result of at least two positive tests [63]. Nevertheless, the status of An. calderoni as a local vector in PAC should be further evaluated. Even though An. albimanus is considered the main malaria vector in coastal PAC [8,9,59], it was not detected infected by Plasmodium spp. The predominant exophilic behavior of these two anthropophilic species suggest that application of residual insecticides on the outside walls of houses may be appropriate vector control.

Conclusions
The ecological heterogeneity of the PAC localities was reflected in the variation of Anopheles species dominating in each one. Of concern is the fact that these species are among the main Colombian malaria vectors. Their importance in malaria transmission in different human impacted settings was confirmed. Infected An. darlingi was detected in deforested margins of humid tropical forest and An. nuneztovari in an area affected by various human interventions. The main PAC vector An. albimanus and the suspected vector An. calderoni were in areas impacted by urbanization and large-scale palm oil agriculture. Knowledge of their contribution to local malaria transmission and their spatial/ temporal distributions helps to understand the dynamics of malaria transmission in this region and will provide the basis for the design and evaluation of more focused control measures.