From 2002–2005, Panama experienced a malaria epidemic that has been associated with El Niño Southern Oscillation weather patterns, decreased funding for malaria control, and landscape modification. Case numbers quickly decreased afterward, and Panama is now in the pre-elimination stage of malaria eradication. To achieve this new goal, the characterization of epidemiological risk factors, foci of transmission, and important anopheline vectors is needed. Of the 24,681 reported cases in these analyses (2000–2014), ~62% occurred in epidemic years and ~44% in indigenous comarcas (5.9% of Panama’s population). Sub-analyses comparing overall numbers of cases in epidemic and non-epidemic years identified females, comarcas and some 5-year age categories as those disproportionately affected by malaria during epidemic years. Annual parasites indices (APIs; number of cases per 1,000 persons) for Plasmodium vivax were higher in comarcas compared to provinces for all study years, though P. falciparum APIs were only higher in comarcas during epidemic years. Interestingly, two comarcas report increasing numbers of cases annually, despite national annual decreases. Inclusion of these comarcas within identified foci of malaria transmission confirmed their roles in continued transmission. Comparison of species distribution models for two important anophelines with Plasmodium case distribution suggest An. albimanus is the primary malaria vector in Panama, confirmed by identification of nine P. vivax-infected specimen pools. Future malaria eradication strategies in Panama should focus on indigenous comarcas and include both active surveillance for cases and comprehensive anopheline vector surveys.
The study of malaria epidemiology and the spatial distributions of both malaria cases and vectors of malaria are essential for the elimination of this disease. Although Panama experienced a malaria epidemic in the early 2000s, this country reports fewer than 1,000 autochthonous cases each year. By understanding the risk factors for Plasmodium transmission and the locations of hot spots, vector and malaria control interventions can be targeted to both high risk individuals and regions, to maximize impact. In this research article, our results underscore the health disparities experienced by the indigenous people of Panama, as we identify them as those at greatest risk of malaria and their comarcas (indigenous reservations) as transmission hot spots. Additionally, we are able to implicate Anopheles albimanus as the primary vector of malaria in Panama through testing of collected specimens for Plasmodium vivax and P. falciparum infection and by calculating the odds of co-occurrence of both vectors with cases of malaria throughout Panama.
Citation: Lainhart W, Dutari LC, Rovira JR, Sucupira IMC, Póvoa MM, Conn JE, et al. (2016) Epidemic and Non-Epidemic Hot Spots of Malaria Transmission Occur in Indigenous Comarcas of Panama. PLoS Negl Trop Dis 10(5): e0004718. doi:10.1371/journal.pntd.0004718
Editor: Thomas S. Churcher, Imperial College London, UNITED KINGDOM
Received: December 19, 2015; Accepted: April 27, 2016; Published: May 16, 2016
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This research was supported by the Secretariat for Science, Technology and Innovation of Panama (SENACYT) research grants COL11-044 and ITE10-015 and the National Research Investigator Board (SNI) prize awarded to JRL, and the National Institutes of Health grant R01 A154139 to JEC. The National Institutes of Health Biodefense and Emerging Infectious Disease training fellowship grant T32 AI05532901 provided partial support for WL. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Between 2002 and 2005, Panama experienced a malaria epidemic which reached a peak of 5,085 cases in 2004 [1,2], associated with El Niño Southern Oscillation events , decreased malaria control funding [3,4] and extensive landscape modification/deforestation . Intensified control efforts  quickly reduced the number of annually reported cases, with the Panamanian Ministry of Health (MINSA) reporting only 747 cases (100% P. vivax) in 2014, restricted to eastern Panama. Because of its current low level of malaria transmission, Panama is now in the pre-elimination stage of malaria eradication .
The process of malaria elimination will be difficult, as residual malaria cases are likely to occur in hard-to-access communities; thus the reported number of cases is an underestimate of the actual situation . In 2014, ~63% of incident malaria cases in Panama occurred in the comarcas (indigenous reservations) of eastern Panama, near malaria endemic western Colombia . Despite efforts to reduce migration of Colombians into Panama from this region, those successful at crossing the border are mobile, and rarely included in malaria surveillance programs . Elimination-oriented control measures must identify spatial foci (hot spots) of residual infections using spatial tools, such as clustering analyses and risk mapping, to target interventions [9,10,11]. To increase the effectiveness of these control measures, it is useful to identify socio-demographic risk factors, using active surveillance with prompt treatment of identified cases (symptomatic and asymptomatic), evaluating current control methods by measuring their impact, and, finally, conducting vector biology studies to understand local anopheline ecology, biology and behavior .
Disparities exist between indigenous (Ngöbe-Buglé, Kuna Yala, Kuna de Madungandí, Kuna de Wargandí and Emberá-Wounaan) and non-indigenous Panamanians . More than 96% of Panama’s indigenous population lives below the poverty line ($1,126 per person per year in 2008), compared to 17.7% among urban populations [3,12,13]. Additionally, indigenous Panamanians have a 7 to 9 year reduction in relative life expectancy . A recent study has associated increased malaria incidence in Panama’s indigenous populations with living conditions, including homes built with temporary materials . These underserved indigenous populations receive outpatient health services from MINSA, but access is reduced because of the prohibitively long and costly travel required to reach health posts .
Early research into Panama’s malaria vectors identified several Plasmodium-infected Anopheles species, including Anopheles albimanus, An. pseudopunctipennis, An. tarsimaculata (Syn. An. aquasalis), An. bachmanni (Syn. An. triannulatus), An. neomaculipalpus, An. punctimacula (Syn. An. malefactor), An. argyritarsis, An. eiseni, and An. apicimacula s.l. [14,15,16,17,18,19]. To date, 15 species and/or species complexes have been identified in Panama [2,20]. However, until recently, no single Anopheles species was formally incriminated as a vector of human Plasmodium spp. since the 1930s [2,20]. A 2015 study that tested anophelines from Comarca Kuna Yala (CKY) for Plasmodium infection found An. albimanus infected with P. vivax .
In 2008, Loaiza et al.  summarized selected vector biology metrics of anophelines collected over 35 years throughout the western Atlantic coast and eastern provinces/comarcas of Panama. These data provided important information on the complex nature of local Plasmodium transmission, and An. albimanus and An. punctimacula s.l. were identified as the most widespread and abundant anophelines . Anopheles albimanus is a major regional malaria vector, with a distribution from southern Mexico to northern South America [22,23]. It is generally considered an exophagic, zoophilic vector that bites in the evening and throughout the night, though its behavior varies across its distribution [22,24]. Anopheles punctimacula s.l., a zoophilic vector , shares much of its distribution with An. albimanus, and has been observed from Mexico to Argentina, and in the Caribbean Islands [26,27].
The present study aims to address some of the knowledge gaps, advocated by , which might impede the implementation of effective malaria elimination strategies in Panama. The goal of elimination can be achieved by identifying risk factors associated with epidemic and non-epidemic malaria years in Panama, from 2000–2014, and determining the locations of malaria hot spots, both of which are critical during the pre-elimination stage of malaria eradication . Additionally, this study aims to determine which Anopheles species are likely involved in malaria transmission, using spatial statistics, species distribution modeling, and testing of specimens for Plasmodium infection.
Materials and Methods
Malaria incidence, census, and geographic data
De-identified, national malaria incidence data (2000–2014) were obtained from the Department of Statistics and Vector Control of MINSA. These data included case location (i.e., province, district, and corregimiento–administrative subdivision of a district), age (years), and Plasmodium species (determined by blood smear microscopy). Cases identified as “imported” in the database were removed prior to analyses. Demographic information was obtained from the 2010 Panamanian census (Institute of Census and Statistics of the Comptroller office of the Republic of Panama) [28,29], and included both the total number of persons per corregimiento and the number of people in each province/comarca per 5-year age category. An ArcMap-compatible (ESRI, Redlands, California) shapefile depicting the geographic boundaries of Panama’s provinces, districts and corregimientos was obtained from the Smithsonian Tropical Research Institute (STRI) geographic information systems (GIS) information portal . Case data that could not be matched to a corregimiento present in the STRI GIS shapefile were excluded prior to analyses.
Chi-squared statistics were used to determine statistically significant differences in distributions of cases between non-epidemic (2000–2001, 2006–2014) and epidemic (2002–2005) years, by relevant demographic variables, including sex, age category (5-year intervals), location (i.e., province or comarca), and Plasmodium species. Logistic regression was used to determine statistically significant differences between non-epidemic and epidemic years (binary variable; 0 = non-epidemic, 1 = epidemic), while controlling for Plasmodium species (binary; 0 = P. vivax, 1 = P. falciparum) and multiple demographic variables simultaneously [province/comarca (binary; 0 = province, 1 = comarca), age in years (continuous), sex (binary; 0 = female, 1 = male), and an interaction variable between sex and province/comarca). Annual Parasite Indices (APIs; annual number of cases per 1000 persons) per Plasmodium species were plotted by year and location to visualize differences in temporal transmission intensities between provinces and comarcas. Finally, the average API in epidemic versus non-epidemic years was plotted versus age category, by sex and location, for each Plasmodium species separately, to better characterize risk factors for epidemic and non-epidemic malaria. Non-parametric Kolmogorov-Smirnov tests for equality in continuous distribution functions were employed to assess differences in age category patterns of cases by sex and Plasmodium species between epidemic and non-epidemic years. R v.3.1.3 software  and RStudio v.0.98.1091 (Boston, MA) were used for all statistical testing.
Identification of spatial foci of increased transmission
Two cluster detection methods were used to identify malaria incidence hot spots for each year of the study: Kulldorff’s spatial scan statistic [32,33] and Getis-Ord Gi* [34,35]. The Kulldorff method was employed using Clusterseer software (BioMedware, Ann Arbor, Michigan) and requires spatial, census and case data. Getis-Ord statistics were undertaken using ArcMap software v.10.2.2 and the Spatial Statistics extension (Mapping Clusters >> Optimized Hot Spot Analysis), using spatial and incidence rate (API) information. All hot spot analyses were conducted using yearly case data or API information and the 2010 census. Additionally, the spatial location of cases was considered the centroid of the corregimiento from which the cases were reported. The number of centroids (corregimientos) per province and comarca can be found in Table 1. The results of the two spatial analyses were combined by summing the frequency of a corregimiento being identified (per year) by either or both methods. These frequency data were then summarized by summing the frequencies of each corregimiento in epidemic years and non-epidemic years separately, as in Xia et al. . Maximum corregimiento-specific frequency in epidemic years is 8 (2 detection methods x 4 epidemic years), and the maximum in non-epidemic years is 22 (2 detection methods x 11 non-epidemic years). The summed frequencies were then projected onto the STRI GIS Panama shapefile to visualize their geographic locations and to provide a summarized distribution of the foci of increased malaria transmission during each period.
Species distribution modeling
Species distribution models for An. albimanus and An. punctimacula s.l. were generated using maximum entropy modeling, implemented in MaxEnt v.3.3.3k [37,38]. Species occurrence data were obtained from VectorMap  and through literature review [20,25,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57], representing collection locations across the entire distribution of both species. MaxEnt models were built using climate and landscape variables, including 19 Bioclim variables (spatial resolution of 30 arc-seconds) , world soil suborder , altitude , hydrological variables (flow accumulation and flow direction) , and tree cover . The MaxEnt program was run with default parameters, with the following differences: create response curves, jackknife to measure variable importance, random seed, do not write clamp grid when projecting, 25 random test percentage, 2 regularization multiplier, 15 replicates, subsample replicated run type, do not write output grids, and 5000 maximum iterations. A bias file was created for each species to inform the program that the area of interest was not sampled uniformly .
After running the full model for each species, predictor variables were assessed for their percent contribution to the model (≥ 3%) and pair-wise correlations (cut-off ≥|0.80|), as in Young et al. . The percent contribution of each variable is given in the MaxEnt output. However, pairwise correlations were determined using the SDM Toolbox extension for ArcMap . The predictor variables that met the above criteria were then used to create the final, parsimonious model, representing the average distributions of 15 MaxEnt iterations. These distributions, which represent the probability of species occurrence at each pixel of the map, were reclassified into a binary image (0 = absent, 1 = present) using the MaxEnt calculated maximum training sensitivity plus specificity logistic threshold in ArcMap. This threshold has been shown to be reliable when using presence-only data .
Correlation of anopheline predicted distributions with incidence of malaria
Predicted An. albimanus and An. punctimacula s.l. distributions were compared, statistically, with those of malaria hot spots and API per corregimiento using odds ratios (99% confidence intervals), as in . Eight comparisons were made per vector species (P. vivax and P. falciparum for API and hot spot analyses, separately, in epidemic versus non-epidemic years). Odds ratios (ORs) were calculated with the raster  and abd R packages , using binary presence/absence SDM maps compared to binary hot spot maps (0: never identified in a hot spot; 1: identified in a hot spot) or corregimiento-level API maps (0: unstable transmission, average API < 0.1 cases per 1,000; 1: stable transmission, average API ≥ 0.1 cases per 1,000).
Collection and Plasmodium testing of anopheline specimens
Anopheles mosquitoes were captured during overnight collections using human-landing catch, CDC light traps, Shannon traps, or resting. Collections occurred in thirty-one localities throughout Panama (n = 22 in 2006–2007, n = 14 in 2008–2015; Fig 1, S1 Table). Specimens were morphologically identified using a dichotomous key . The heads and thoraces of specimens were pooled (n ≈ 5 per pool) by species, locality and collection date, and extracted using a Qiagen BioSprint 96 robot DNA extractor and Qiagen BioSprint 96 DNA Blood kits (Venlo, Netherlands). Specimens collected between 2006 and 2007 were tested for Plasmodium infection using ELISA , and those collected afterward (2008–2015) were tested using nested, real-time TaqMan PCR . Inconclusive PCR results (i.e., Plasmodium genus-specific product amplified, but P. vivax and P. falciparum-specific products not amplified), were amplified again and the genus-specific product was sequenced and submitted to the National Center for Biotechnology Information (NCBI) Blastn database to assess species homology within the Plasmodium genus. Additionally, any Plasmodium-positive mosquito specimens not able to be morphologically identified (e.g., a member of a species complex) were molecularly identified using the ITS2 region of the mosquito 5.8S ribosomal RNA .
2006–2007 collection sites = black circles; 2008–2015 collection sites = grey squares; provinces = white; comarcas = grey. Yellow star indicates location of Panama City. Each province and comarca is labeled. BOC = Bocas del Toro; CHI = Chirquí, CNB = Comarca Ngöbe-Buglé, VER = Veraguas; HER = Herrera; LST = Los Santos; COC = Coclé, COL = Colón; PAN = Panamá, CKY = Comarca Kuna Yala; CKM = Comarca Kuna de Madungandí; CKW = Comarca Kuna de Wargandí, CEM = Comarca Embera-Wounaan; DAR = Darién. CKM is a territory within PAN province; CKW is a territory within DAR province. Insets depict details in northern BOC and in southwestern DAR provinces. Panama GIS shapefile obtained from STRI .
Epidemiological data and analyses
Between 2000 and 2014, 24,937 malaria cases were reported from Panama. Of these, 256 (236 P. vivax, 16 P. falciparum, and 4 unknown) were excluded from analysis because the case corregimiento was not present in the Panama shapefile (n = 153), the age of the case was not available (n = 99), or the Plasmodium species was unknown (n = 4). The final data set included 24,681 cases, and no differences were noted in proportions of Plasmodium species between the original and final data sets. Plasmodium vivax cases were reported in every year of the study. However, no P. falciparum cases have been reported in Panama since 2010 (Fig 2). The epidemic peaks for both P. vivax and P. falciparum occurred in comarcas in 2003, and a year later in the provinces. During the malaria epidemic years (2002–2005), the average number of cases per year far exceeded that of non-epidemic years (2000–2001, 2006–2014) for both parasites (average 3.8-fold and 20.9-fold increases for P. vivax and P. falciparum, respectively). For all P. falciparum analyses, 2000–2001 and 2006–2010 were used as the non-epidemic years, since no P. falciparum cases were reported after 2010.
Grey lines = comarcas; black lines = provinces; solid lines = P. vivax; dashed lines = P. falciparum.
Chi-squared analyses detected statistically significant differences in the distribution of cases in epidemic and non-epidemic years by sex ( = 4.968, p = 0.026), location ( = 61.856, p < 0.001), age category ( = 50.484, p < 0.001), and Plasmodium species ( = 1228.476, p < 0.001; Table 2). During epidemic years, the indigenous people of the comarcas, females, and some age groups experienced more cases than expected. In addition, more cases of P. falciparum were reported in epidemic years than expected.
Logistic regression analyses uncovered statistically significant differences between non-epidemic and epidemic years with respect to Plasmodium species and some demographic variables. The final, adjusted logistic model included Plasmodium species, province/comarca, age in years, sex, and an interaction term between sex and province/comarca. In epidemic years, there was a statistically significant 41% increase (odds ratio = 1.41, 95% confidence interval = 1.39–1.44) in the odds of a case being due to P. falciparum compared to non-epidemic years, controlling for all other variables. Additionally, in epidemic years, there was a statistically significant increase in the odds of a case occurring in a comarca (OR = 1.08, 95% CI = 1.06–1.10), compared to non-epidemic years, and controlling for all other variables. Finally, in epidemic years, there was a statistically significant interaction between sex and province/comarca resulting in a decrease in odds of a male case occurring in a comarca (OR = 0.97, 95% CI = 0.95–0.99), compared to non-epidemic years, and controlling for all other variables. Age (in years) and sex did not contribute to a statistically significant difference in odds of a case between non-epidemic and epidemic years in these analyses, after controlling for all other variables.
Visualization of average API per Plasmodium species showed striking differences in malaria transmission and epidemiology experienced by provinces and comarcas. The API for P. vivax in comarcas was higher than that of the provinces for all years in this study (peak ~12.3 cases per 1,000 persons in 2003; Fig 3A). However, the P. falciparum API was higher only in the epidemic years (peak of ~2.6 cases per 1,000 persons in 2003) and nearly equal in non-epidemic years (Fig 3C). Further analysis by individual provinces and comarcas allowed for a more direct determination of the regions which had a disproportionate number of cases. CNB contributed the most P. vivax cases to the epidemic early on (~15 cases per 1,000 in 2003); in contrast, other comarcas and provinces, such as Comarca Emberá-Wounaan (CEM) and BOC reached their peaks in 2005 (Fig 3B). Despite their recognition as indigenous territories since January 1996 and July 2000 respectively , no cases of malaria were reported from Comarca Kuna de Madungandí (CKM) and Comarca Kuna de Wargandí (CKW) before 2008. However, since reporting began, the P. vivax API in both has increased, on average–a pattern unique to these comarcas (Fig 3B). Increases in P. falciparum API occurred primarily during the 2002–2005 malaria epidemic, beginning with a drastic increase in CKY (16.2 cases per 1,000 in 2003; Fig 3D). This increase was followed by subsequent increases in CEM (~3.5 cases per 1,000 in 2004), and in DAR (~3.7 and 4.6 cases per 1,000 in 2004 and 2005, respectively) and CKY (~3.1 cases per 1,000 in 2005) (Fig 3D).
Provinces = black dashed lines; comarcas = grey solid lines. A and C) Plasmodium vivax and P. falciparum, respectively, with provinces and comarcas grouped; B and D) P. vivax and P. falciparum, respectively, with provinces and comarcas separated. Note different y-axis scales on each panel. CNB = Comarca Ngöbe-Buglé, CKW = Comarca Kuna de Wargandí, CKM = Comarca Kuna de Madungandí, CKY = Comarca Kuna Yala, CEM = Comarca Emberá-Wounaan, DAR = Darién province.
In general, average API per age category was greatest in comarcas in epidemic years for both sexes and both parasites (Fig 4). Plasmodium vivax API age category patterns (Fig 4A and 4C) do not differ greatly by province/comarca in epidemic and non-epidemic years, though male APIs were greater than those of females in general. Additionally, there is an overall decreasing trend in P. vivax API with increasing age (Fig 4A and 4C). However, P. falciparum APIs were much more variable across age categories (Fig 4B and 4D). Kolmogorov-Smirnov tests comparing age category-related patterns in cases between epidemic and non-epidemic years found no differences by sex and province/comarca for P. vivax (comarca males, p = 0.819; province males, p = 0.560; comarca females, p = 0.819; province females, p = 0.819). However, statistically significant differences in the age category-related patterns were identified for P. falciparum cases in all four categories (p < 0.001, p = 0.001, p < 0.001, and p = 0.348 respectively).
Epidemic years = solid lines; non-epidemic years = dashed lines; provinces = black; comarcas = grey. Left y-axis represents comarca APIs and right y-axis represents province APIs. A and C) Plasmodium vivax API among males and females, respectively; B and D) P. falciparum API among males and females, respectively.
Identification of spatial foci of increased transmission
Hot spot analyses show two major foci of increased P. vivax transmission in epidemic years: CNB, and southern DAR and the southwestern part of CEM (Figs 5A and S1). The locations of these foci differ from those of non-epidemic years, where only one major focus was observed in eastern Panama and included CKM, CKY, northern and southern DAR, CKW, and CEM (Figs 5C and S1). A smaller focus, identified less frequently in non-epidemic years, was observed in BOC, Veraguas (VER), and CNB (Fig 5C). Foci of P. falciparum transmission were only found in eastern Panama and largely overlapped between epidemic and non-epidemic years (Figs 5B, 5D and S2). The focus was centered near the Caribbean coast of eastern Panama in epidemic years and included CKM, CKY, northern DAR, CKW and northern CEM. However, it was centered near the Pacific coast in non-epidemic years, primarily in DAR and CEM.
Corregimientos are colored by the frequency at which they were identified by both hot spot detection methods over the designated period. A) P. vivax cluster frequency by corregimiento in epidemic years (2002–2005; maximum frequency = 8). B) P. falciparum cluster frequency by corregimiento in epidemic years (2002–2005; maximum frequency = 8). C) P. vivax cluster frequency by corregimiento in non-epidemic years (2000–2001, 2006–2014; maximum frequency = 22). D) P. falciparum cluster frequency by corregimiento in non-epidemic years (2000–2001, 2006–2010; maximum frequency = 14). Frequencies were calculated using data shown in S1 and S2 Figs. Panama GIS shapefile obtained from STRI .
Species distribution models and correlation with Plasmodium
The An. albimanus full model produced a mean area under the curve (AUC) of 0.938. AUC ranges from 0.5 (random ranking of presence versus background sites) to 1.0 (perfect ranking), and is a value used to assess model performance . Altitude, soil substrate, tree cover and Bioclim variables bio2 (mean diurnal range), bio4 (temperature seasonality), bio6 (minimum temperature of the coldest month), and bio9 (mean temperature of the driest quarter) all contributed ≥ 3% to the model. Pairwise variable correlations ≥ |0.8| were observed among the Bioclim variables bio2 and bio6 (bio6 was removed). The final/parsimonious model (mean AUC = 0.931) included altitude, soil substrate, tree cover, bio2, bio4, and bio9. The maximum training sensitivity plus specificity logistic threshold (0.1969) was used to create an An. albimanus species presence/absence species distribution map (Fig 6A).
A and C) Anopheles albimanus and B and D) Anopheles punctimacula s.l. in Panama. Panels A and B represent the full extent of the species distribution models. Panels C and D represent the distributions of each species within Panama. Color shading indicates areas of predicted suitable habitat/presence of the species; white indicates areas of predicted absence of the species. Central and South American GIS shapefiles freely available from DIVA-GIS . Panama GIS shapefile obtained from STRI .
The An. punctimacula s.l. full model produced a mean AUC of 0.885. Altitude, soil substrate, tree cover, bio2 (mean diurnal range), bio6 (minimum temperature of the coldest month), bio7 (annual temperature range), and bio14 (precipitation of the driest month) all contributed ≥ 3% to the model. Pairwise variable correlations ≥ |0.8| were observed among the Bioclim variables bio2, bio6 and bio7 (bio6 and bio7 were removed). The final/parsimonious model (mean AUC = 0.888) included altitude, soil substrate, tree cover, bio2, and bio14. The maximum training sensitivity plus specificity logistic threshold (0.2588) was used to create an An. punctimacula s.l. presence/absence species distribution map (Fig 6B).
Overall, An. punctimacula s.l. has a wider distribution than An. albimanus in Panama. Although areas of predicted An. punctimacula s.l. presence are found throughout western Panama and in nearly all areas of central and eastern Panama, areas of predicted An. albimanus presence tended to be found nearer to the coasts. In general, correlative analyses showed significantly increased odds of both P. vivax and P. falciparum (corregimiento-associated API and areas identified in hot spot analyses) in areas of predicted An. albimanus presence in both epidemic and non-epidemic years (Table 3). However, there was a significant reduction in odds of epidemic and non-epidemic Plasmodium transmission in areas of predicted An. punctimacula s.l. presence (Table 3).
Plasmodium testing of anophelines
Collecting efforts in thirty-one localities across Panama (Fig 1) resulted in 19,163 anopheline specimens (13,803 in 2006–2007 and 5,360 in 2008–2015; Table 4, S1 Table). Among these specimens, members of eight species (An. albimanus, An. aquasalis, An. malefactor, An. neivai, An. neomaculipalpus, An. nuneztovari s.s., An. pseudopunctipennis, and An. vestitipennis) and four species complexes (An. apicimacula s.l., An. punctimacula s.l., An. strodei s.l., and An. triannulatus s.l.; Table 4, S1 Table) were identified using morphological identification. Because the  morphological key is not reliable for all anopheline species in eastern Panama, specimens originally considered to be An. oswaldoi were molecularly identified using the COI barcode, as in ; all these specimens were An. nuneztovari s.s. Four specimens could not be identified morphologically (2 –Anopheles (Nyssorhynchus) spp. and 2 –Anopheles (Arribalzagia series) spp.).
ELISA testing of specimens collected in 2006–2007 identified nine pools of An. albimanus infected with P. vivax (3 with VK210 variant and 6 with VK247 variant). These specimens were collected in BOC (Fig 1, Table 4, S1 Table). Real-time PCR of specimen pools from 2008–2015 resulted in the identification of one Plasmodium spp. positive An. punctimacula s.l. pool (resting collection in BOC, Table 4, S1 Table). This PCR result was inconclusive, and the Plasmodium genus-specific PCR product sequence analysis showed the presence of Plasmodium juxtanucleare (163 bp fragment of 18S ribosomal RNA gene, 99% identity, 0 gaps, GenBank accession AF463507.1). Further analysis of the Plasmodium positive An. punctimacula s.l., using ITS2, determined that it belongs to clade B (100% identity, 0 gaps, GenBank accession JX212812.1). Furthermore, the collection site of this mosquito pool is congruent with the published distribution of An. punctimacula clade B in Panama .
As Panama makes progress toward malaria elimination, a greater push for identification of cases in the remote corregimientos and comarcas of the country will be essential. Despite increases in the number of cases among some demographic groups, such as females, during epidemic years, the indigenous people of Panama were identified as those most disproportionately affected by malaria in all years of this study (Table 2, Figs 3 and 4). Many indigenous people live in poverty and have limited access to health services [3,12]. In 2014, with only 747 P. vivax and 0 P. falciparum cases nationally, the people living within the indigenous territories of CKM and CKW experienced APIs of ~58 and 27 cases per 1,000 people, respectively (Fig 3). The province or comarca with the third highest API in 2014 in Panama (CKY), had an API of ~3.6 (17-fold and 7-fold lower than CKM and CKW). Even though these territories have relatively small populations (CKM: 4,271; CKW: 1,914) , they serve as an important focus of residual malaria transmission.
During the 2002–2005 malaria epidemic, there was a marked increase in the number of P. falciparum cases (16.9% of cases in epidemic years, 2.3% of cases in non-epidemic years; Table 2, Fig 2). This P. falciparum epidemic was restricted to eastern Panama, and primarily affected CKY, CEM and DAR (Figs 3 and 5). Restriction of cases near the Colombian border is consistent with the results of a recent study characterizing P. falciparum haplotypes in Panama, confirming Colombia as their origin . However, the P. vivax epidemic was more widespread, affecting eastern Panama, along with CNB and BOC in the west (Figs 3 and 5). Interestingly, only one major non-epidemic focus of P. vivax was identified, and it was centered in eastern Panama (Figs 3 and 5). These results suggest that current P. vivax transmission in Panama could be related to the influx and movement of migrants from the malarious regions of western Colombia .
For the most part, Kulldorff’s spatial scan statistic and Getis-Ord Gi* provided congruent results in analyses identifying spatial foci of increased Plasmodium transmission (S1 and S2 Figs). However, there are some instances where small numbers of cases in a given region were identified by only one method as a hot spot. For example, hot spots were identified by Kulldorff’s scan statistic for P. vivax (S1 Fig, black circles; 2003, 2005, 2006, 2008–2014) and P. falciparum (S2 Fig, black circles; 2010), despite not being identified by Getis-Ord Gi*. The opposite is true, too. Getis-Ord Gi* identified areas of statistically significantly increased incidence of P. vivax (S1 Fig, pink/red shading; 2005) and in P. falciparum (S2 Fig, pink/red shading; 2009). These findings support the use of two separate statistical methods for the identification of hot spots of Plasmodium incidence, and the combination of the results from the two methods allows for a better, more complete picture of the true spatial heterogeneity of malaria cases in Panama over time.
With the exception of P. falciparum in comarcas during epidemic years, Plasmodium APIs tended to decrease with increasing human age, suggesting that some individuals in Panama, especially in comarcas, could be asymptomatic carriers/reservoirs of the parasite [78,79]. Similar patterns have been reported elsewhere [78,80,81,82,83], and, in some cases, asymptomatic patients have been shown to outnumber symptomatic ones by 4–5 fold . Determining whether there are asymptomatic carriers of Plasmodium among those living in areas with continued malaria transmission in Panama is essential for current and future malaria elimination efforts. These individuals cannot be identified through passive disease surveillance, but must be detected through active case detection strategies with PCR testing of samples, as suggested by , and must be accompanied by comprehensive treatment.
In this study, we identified nine specimen pools of An. albimanus infected with P. vivax. However, because collection efforts were spread throughout Panama and, in general, of short duration, it is possible that other important local vectors may have been missed. However, the absence of Plasmodium infected anopheline pools from 2008–2015 may truly reflect the low level of malaria endemicity in Panama, though a recent study in CKY found the presence of P. vivax-infected An. albimanus in CKY . Anopheles mosquito surveys and Plasmodium testing should be repeated using the epidemiological and spatial statistics results presented here as a framework for locality identification. For example, these surveys should be completed in the indigenous comarcas (e.g., CKM and CKW) since indigenous people reported 63.2% of the malaria cases in Panama during this study, but make up only 6.2% of the population, according to the 2010 census . Also, mosquito-sampling effort should increase in future studies to account for the potential regional transmission role of An. darlingi, especially in southern DAR.
Anopheles albimanus was identified in this study as the most important malaria vector in Panama. However, the lack of an association between the distributions of An. punctimacula s.l. and Plasmodium does not mean that this species has no role in local transmission. Within this species complex, there are multiple molecular lineages , and further research is needed to model their individual distributions and to characterize the possible importance of each in malaria transmission. One of the assumptions of the spatial hot spot analyses in this study is that incident malaria occurred randomly throughout each corregimiento since it was not possible to georeference every locality with reported malaria cases. Anopheles albimanus has increased odds of co-occurrence and is likely the major vector in this country, but this does not mean that An. punctimacula and other anophelines did not play important roles in local transmission of Plasmodium in both epidemic and non-epidemic years. Additionally, because An. punctimacula appears to be much more of an ecological generalist, compared to An. albimanus in these analyses, it is predicted to be present in many areas where no malaria transmission was reported during the study period, greatly diminishing its odds of co-occurrence with incident malaria and potentially underestimating its importance as a vector in Panama. Among the anophelines tested for Plasmodium, one pool of An. punctimacula s.l. (collected resting in BOC) was positive for Plasmodium juxtanucleare, an avian parasite [85,86,87,88]. Despite being found in an anopheline (previous work suggests that Anopheles mosquitoes are refractory to this parasite ), the primary vectors of P. juxtanucleare are normally Culex spp. [89,90,91,92].
This study has a number of limitations. Firstly, the epidemiological analyses used in this study relied on malaria cases reported to the Panamanian MINSA either passively, through patients visiting health posts, or actively by MINSA workers visiting villages with current malaria cases. Despite the hard work of the Panamanian MINSA, the number of reported malaria cases in Panama, particularly those within comarcas, is likely to be an underestimate, due to the difficulty accessing remote villages and/or traveling from these villages to local health posts. Secondly, some administrative areas in Panama underwent restructuring between the 2000 and 2010 censuses. Because of these changes, and because the GIS shapefiles used in this study reflected the current administrative boundaries in Panama, it was not possible to use the 2000 census data for API calculations. As a result, we used 2010 census numbers for all years of the study, knowing that these numbers may not reflect the true populations in each province, district or municipality during all years. Thirdly, because it was not possible to georeference each reported malaria case, data were aggregated to summarize each corregimiento per year, reducing the resolution of the spatial hot spot analyses. Finally, despite the mosquito collections summarized in this study occurring during times of known malaria transmission, these collections occurred after the 2002–2005 epidemic, making it impossible to determine the vectors playing roles in transmission during this period. However, Loaiza et al.  summarizes 35 years of anopheline collections in Panama, and throughout that time period, An. albimanus was found to be the most abundant vector, suggesting that it was likely to play an important role in 2002–2005 malaria epidemic in Panama.
Recent research on malaria hotspots in areas of low or unstable transmission has shown the importance of PCR and serology based data for identification of regions with increased levels of asymptomatic carriage of Plasmodium parasites [93,94,95], rather than the use of microscopy [79,93]. In Panama, we recommend the use of PCR for the identification of people with low levels of parasitemia, rather than serology, because high antibody titers may not represent a current infection, but rather a past exposure . After prospective surveys of the hot spots identified in this study, using PCR identification of Plasmodium infection, these data can be analyzed using Bayesian geostatistics  to predict the spatiotemporal patterns of asymptomatic Plasmodium infections in Panama, giving further guidance to malaria elimination efforts. A recent cluster-randomized, controlled trial tested the effectiveness of interventions targeting malaria hot spots and identified important factors that need to be recognized and addressed in the future . Bousema et al.  suggest the failure of their interventions to decrease local malaria transmission is due to unrecognized insecticide resistance in the local vector populations, transmission of Plasmodium or spread of infected mosquitoes from other, local transmission hot spots, and/or introduction of Plasmodium through human movement/migration.
Overall, this study underscores the disparities between the indigenous and non-indigenous people of Panama, with respect to health care access. Current and future malaria elimination efforts must be focused on the comarcas to maximize their effect, and must include active surveillance systems to identify asymptomatic reservoirs of Plasmodium, and thorough anopheline surveys, to identify the species that are involved in transmission. Vectors identified as important in the identified hot spots should be studied to determine their feeding and resting behaviors, biting time patterns, population genetics and insecticide resistance, in order to create tailored, effective vector control interventions. Additionally, human movement and migration must be better studied and understood in Panama, since it is likely that Plasmodium transmission in eastern Panama is associated with the immigration of people from Colombia, and because the importation of Plasmodium is known to hinder malaria elimination programs [99,100].
S1 Table. Anopheles specimen collection and Plasmodium infection information.
S1 Fig. Identification of annual foci of increased Plasmodium vivax malaria incidence in Panama (2000–2014) using two detection methods.
Colored municipalities (Getis-Ord Gi* statistic; red and pink = 99% and 95% confidence, respectively) and circled areas (Kulldorff’s spatial scan statistic) signify statistically significant malaria incidence hot spots. Dark (99% confidence) and light blue (95% confidence) colored municipalities in the 2005 panel represent cold spots of malaria incidence, as determined by Getis-Ord Gi* statistics.
S2 Fig. Identification of annual foci of increased Plasmodium falciparum malaria incidence in Panama (2000–2014) using two detection methods.
No foci were identified for P. falciparum for years 2011–2014. Colored municipalities (Getis-Ord Gi* statistic; red and pink = 99% and 95% confidence, respectively) and circled areas (Kulldorff’s spatial scan statistic) signify statistically significant malaria incidence hot spots. No statistically significant hot spot was detected by Kulldorff’s spatial scan statistics in 2009, whereas none were detected by Getis-Ord Gi* statistics in 2010.
We thank Gabriel Z. Laporta for his assistance with the calculation of odds ratios and for supplying his R programming code, and Catherine Prussing for her data analysis insights. We thank Sara A. Bickersmith for her assistance with Plasmodium testing. We thank the staff of the department of statistics and vector control of MINSA for providing malaria epidemiological data, and Panama’s Environmental Authority (ANAM) for supporting scientific collecting and exporting of mosquitoes. We are thankful to the Panamanian Border Service (SENAFRONT) for field assistance while working near the Colombian border. Finally, we acknowledge the Applied Genomic Technologies Core at the Wadsworth Center, New York State Department of Health for their sequencing services.
Conceived and designed the experiments: WL JEC JRL MMP. Performed the experiments: WL LCD JRR JRL IMCS. Analyzed the data: WL. Contributed reagents/materials/analysis tools: MMP JEC. Wrote the paper: WL JEC JRL.
- 1. PAHO (2008) Malaria in the Americas: time series epidemiological data from 2000 to 2007. Pan American Health Organization.
- 2. Loaiza J, Scott M, Bermingham E, Rovira J, Sanjur O, et al. (2009) Anopheles darlingi (Diptera: Culicidae) in Panama. Am J Trop Med Hyg 81: 23–26. pmid:19556561
- 3. Hurtado LA, Caceres L, Chaves LF, Calzada JE (2014) When climate change couples social neglect: malaria dynamics in Panama. Emerg Microbes Infect 3: e28.
- 4. WHO (2014) World Malaria Report 2014. Geneva, Switzerland: World Health Organization.
- 5. Global Health Group (2013) Eliminating Malaria in Panama. San Francisco, California: University of California, San Francisco.
- 6. Arevalo-Herrera M, Quinones ML, Guerra C, Cespedes N, Giron S, et al. (2012) Malaria in selected non-Amazonian countries of Latin America. Acta Trop 121: 303–314. doi: 10.1016/j.actatropica.2011.06.008. pmid:21741349
- 7. Cotter C, Sturrock HJ, Hsiang MS, Liu J, Phillips AA, et al. (2013) The changing epidemiology of malaria elimination: new strategies for new challenges. Lancet 382: 900–911. doi: 10.1016/S0140-6736(13)60310-4. pmid:23594387
- 8. Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI (2005) The global distribution of clinical episodes of Plasmodium falciparum malaria. Nature 434: 214–217. pmid:15759000
- 9. Nájera JA, González-Silva M, Alonso PL (2011) Some lessons for the future from the Global Malaria Eradication Programme (1955–1969). PLoS Med 8: e1000412. doi: 10.1371/journal.pmed.1000412. pmid:21311585
- 10. de Castro MC, Monte-Mor RL, Sawyer DO, Singer BH (2006) Malaria risk on the Amazon frontier. Proc Natl Acad Sci U S A 103: 2452–2457. pmid:16461902
- 11. Obaldia N (2015) Determinants of low socio-economic status and risk of Plasmodium vivax malaria infection in Panama (2009–2012): a case-control study. Malar J 14: 14. doi: 10.1186/s12936-014-0529-7. pmid:25603818
- 12. PAHO (2012) Panama. Health in the Americas: Country Volume. 2012 ed: Pan American Health Organization.
- 13. Instituto Nacional de Estadística y Censo (2008) Encuesta de Niveles de Vida ENV 2008. Panama, Panama: Contraloría General de la República de Panamá.
- 14. Simmons JS (1936) Anopheles (Anopheles) neomaculipalpus, Curry, experimentally infected with malaria plasmodia. Science 84: 202–203.
- 15. Simmons JS (1936) Anopheles (Anopheles) punctimacula (Dyar & Knab) experimentally infected with malaria plasmodia. Science 83: 268–269.
- 16. Simmons JS (1936) Anopheles (Anopheles) punctimacula naturally infected with malaria plasmodia. Am J Trop Med Hyg s1-16: 105–108.
- 17. Rozeboom LE (1935) Infection of Anopheles bachmanni Petrocchi, with Plasmodium vivax, Grassi and Feletti, and observation on the bionomics of the mosquito. Am J Trop Med Hyg s1-15: 521–528.
- 18. Darling ST (1910) Studies in relation to malaria. Washington, DC: Laboratory of the Board of Health.
- 19. Simmons JS (1937) Observations on the importance of Anopheles punctimacula as a malaria vector in Panama, and report of experimental infections in A. neomaculipalpus, A. apicimacula, and A. eiseni. Am J Trop Med Hyg s1-17: 191–212.
- 20. Loaiza JR, Bermingham E, Scott ME, Rovira JR, Conn JE (2008) Species composition and distribution of adult Anopheles (Diptera: Culicidae) in Panama. J Med Entomol 45: 841–851. pmid:18826025
- 21. Calzada JE, Marquez R, Rigg C, Victoria C, de la Cruz M, et al. (2015) Characterization of a recent malaria outbreak in the autonomous indigenous region of Guna Yala, Panama. Malar J 14: 459. doi: 10.1186/s12936-015-0987-6. pmid:26578076
- 22. Sinka ME, Rubio-Palis Y, Manguin S, Patil AP, Temperley WH, et al. (2010) The dominant Anopheles vectors of human malaria in the Americas: occurrence data, distribution maps and bionomic precis. Parasit Vectors 3: 72. doi: 10.1186/1756-3305-3-72. pmid:20712879
- 23. Zimmerman RH (1992) Ecology of malaria vectors in the Americas and future direction. Mem Inst Oswaldo Cruz 87 Suppl 3: 371–383. pmid:1343717
- 24. Conn JE, Quiñones ML, Póvoa MM (2013) Phylogeography, vectors, and transmission in Latin America. In: Manguin S, editor. Anopheles mosquitoes–New insights into malaria vectors. Intech Open.
- 25. Ulloa A, González-Cerón L, Rodríguez MH (2006) Host selection and gonotrophic cycle length of Anopheles punctimacula in southern Mexico. J Am Mosq Control Assoc 22: 648–653. pmid:17304932
- 26. Forattini OP (1962) Entomologia Medica, Vol. 1. São Paulo, Brazil: Faculdade de Higiene e Sáude Publica. 622 p.
- 27. Knight KL, Stone A (1977) A catalog of the mosquitoes of the world (Diptera: Culicidae). College Park, Maryland: Thomas Say Foundation and the Entomological Society of America.
- 28. Instituto Nacional de Estadística y Censo (2015) Superficie, población y densidad de población en la República, según Provincia, Comarca indígena, Distrito y Corregimiento: Censos de 1990, 2000 y 2010. In: P3601Cuadro11.xls, editor. Panama, Panama: Contraloría General de la República de Panamá.
- 29. Instituto Nacional de Estadística y Censo Población en la República, por sexo, según Provincia, Comarca indígena y grupos de edad: Censo 2010. In: P3601Cuadro13.xls, editor. Panama, Panama: Contraloría General de la República de Panamá.
- 30. Smithsonian Tropical Research Institute (2013) Division Administrativa de la Rep. de Panamá—Mapa y Capas. In: Panama_DivisionAdministrativa.lyr, editor. Panama, Panama: Smithsonian Institute.
- 31. R Core Team (2015) R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing.
- 32. Kulldorff M, Nagarwalla N (1995) Spatial disease clusters: Detection and inference. Stat Med 14: 799–810. pmid:7644860
- 33. Kulldorff M (1997) A spatial scan statistic. Communication in Statistics: Theory and Methods 26.
- 34. Getis A, Ord JK (1992) The Analysis of Spatial Association by Use of Distance Statistics. Geogr Anal 24: 189–206.
- 35. Ord JK, Getis A (1995) Local Spatial Autocorrelation Statistics: Distributional Issues and an Application. Geogr Anal 27: 286–306.
- 36. Xia J, Cai S, Zhang H, Lin W, Fan Y, et al. (2015) Spatial, temporal, and spatiotemporal analysis of malaria in Hubei Province, China from 2004–2011. Malar J 14: 145. doi: 10.1186/s12936-015-0650-2. pmid:25879447
- 37. Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modeling of species geographic distributions. Ecol Modell 190: 231–259.
- 38. Phillips SJ, Dudik M (2008) Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31: 161–175.
- 39. The Walter Reed Biosystematics Unit (2015) VectorMap. Silver Spring, Maryland: Smithsonian Institution.
- 40. Gonzalez-Ceron L, Santillan F, Rodriguez MH, Mendez D, Hernandez-Avila JE (2003) Bacteria in midguts of field-collected Anopheles albimanus block Plasmodium vivax sporogonic development. J Med Entomol 40: 371–374. pmid:12943119
- 41. Achee NL, Grieco JP, Andre RG, Rejmankova E, Roberts DR (2007) A mark-release-recapture study to define the flight behaviors of Anopheles vestitipennis and Anopheles albimanus in Belize, Central America. J Am Mosq Control Assoc 23: 276–282.
- 42. Vazquez-Martinez MG, Rodríguez MH, Arredondo-Jiménez JI, Méndez-Sánchez JD, Bond-Compeán JG, et al. (2002) Cyanobacteria associated with Anopheles albimanus (Diptera: Culicidae) larval habitats in southern Mexico. J Med Entomol 39: 825–832. pmid:12495179
- 43. Grieco JP, Rejmankova E, Achee NL, Klein CN, Andre R, et al. (2007) Habitat suitability for three species of Anopheles mosquitoes: larval growth and survival in reciprocal placement experiments. J Vector Ecol 32: 176–187. pmid:18260505
- 44. Wagman J, Grieco JP, Bautista K, Polanco J, Briceño I, et al. (2014) A Comparison Of Two Commercial Mosquito Traps for the Capture Of Malaria Vectors In Northern Belize, Central America. J Am Mosq Control Assoc 30: 175–183. doi: 10.2987/14-6411R.1. pmid:25843092
- 45. Burkett-Cadena N, Graham SP, Giovanetto LA (2013) Resting environments of some Costa Rican mosquitoes. J Vector Ecol 38: 12–19. doi: 10.1111/j.1948-7134.2013.12004.x. pmid:23701603
- 46. Cienfuegos AV, Rosero DA, Naranjo N, Luckhart S, Conn JE, et al. (2011) Evaluation of a PCR–RFLP–ITS2 assay for discrimination of Anopheles species in northern and western Colombia. Acta Trop 118: 128–135. doi: 10.1016/j.actatropica.2011.02.004. pmid:21345325
- 47. Gutierrez LA, Gonzalez JJ, Gomez GF, Castro MI, Rosero DA, et al. (2009) Species composition and natural infectivity of anthropophilic Anopheles (Diptera: Culicidae) in the states of Cordoba and Antioquia, Northwestern Colombia. Mem Inst Oswaldo Cruz 104: 1117–1124. pmid:20140372
- 48. Gómez GF, Márquez EJ, Gutiérrez LA, Conn JE, Correa MM (2014) Geometric morphometric analysis of Colombian Anopheles albimanus (Diptera: Culicidae) reveals significant effect of environmental factors on wing traits and presence of a metapopulation. Acta Trop 135: 75–85. doi: 10.1016/j.actatropica.2014.03.020. pmid:24704285
- 49. Zerpa N, Moreno J, GONZALEZ R J, Noya O (1998) Colonization and Laboratory Maintenance of Anopheles albimanus Wiedemann in Venezuela. Rev Inst Med Trop Sao Paulo 40: 173–176. pmid:9830731
- 50. Arredondo-Jiménez JI, Rodriguez MH, Loyola EG, Bown DN (1997) Behaviour of Anopheles albimanus in relation to pyrethroid‐treated bednets. Medical and Veterinary Entomology 11: 87–94. pmid:9061682
- 51. Orjuela LI, Ahumada ML, Avila I, Herrera S, Beier JC, et al. (2015) Human biting activity, spatial–temporal distribution and malaria vector role of Anopheles calderoni in the southwest of Colombia. Malar J 14: 256. doi: 10.1186/s12936-015-0764-6. pmid:26104785
- 52. Grieco JP, Achee NL, Roberts DR, Andre RG (2005) Comparative susceptibility of three species of Anopheles from Belize, Central America, to Plasmodium falciparum (NF-54). J Am Mosq Control Assoc 21: 279–290. pmid:16252518
- 53. Achee NL, Grieco JP, Rejmankova E, Andre RG, Vanzie E, et al. (2006) Biting patterns and seasonal densities of Anopheles mosquitoes in the Cayo District, Belize, Central America with emphasis on Anopheles darlingi. J Vector Ecol 31: 45–57. pmid:16859089
- 54. Schiemann DJ, Pinzón MLQ, Hankeln T (2014) Anthropophilic Anopheles species composition and malaria in Tierradentro, Córdoba, Colombia. Mem Inst Oswaldo Cruz 109: 384–387.
- 55. Naranjo-Diaz N, Rosero DA, Rua-Uribe G, Luckhart S, Correa MM (2013) Abundance, behavior and entomological inoculation rates of anthropophilic anophelines from a primary Colombian malaria endemic area. Parasit Vectors 6: 61. doi: 10.1186/1756-3305-6-61. pmid:23497535
- 56. Pinault LL, Hunter FF (2011) New highland distribution records of multiple Anopheles species in the Ecuadorian Andes. Malar J 10: 236. doi: 10.1186/1475-2875-10-236. pmid:21835004
- 57. Solarte Y, Hurtado C, Gonzalez R, Alexander B (1996) Man-biting activity of Anopheles (Nyssorhynchus) albimanus and An.(Kerteszia) neivai (Diptera: Culicidae) in the Pacific lowlands of Colombia. Mem Inst Oswaldo Cruz 91: 141–146. pmid:8736081
- 58. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965–1978.
- 59. Natural Resources Conservation Service (2015) Global Soil Regions Map. United States Department of Agriculture.
- 60. Worldclim (2015) Global Climate Data—Date for current conditions (~1950–2000).
- 61. United States Geological Survey (2015) HYDRO1K. Washington DC.
- 62. DeFries RS, Hansen MC, Townshend JRG, Janetos AC, Loveland TR (2000) 1 Kilometer Tree Cover Continuous Fields, 1.0 (1992–1993). College Park, Maryland: Department of Geography University of Maryland,.
- 63. Kramer-Schadt S, Niedballa J, Pilgrim JD, Schröder B, Lindenborn J, et al. (2013) The importance of correcting for sampling bias in MaxEnt species distribution models. Diversity Distrib 19: 1366–1379.
- 64. Young KE, Abbott LB, Caldwell CA, Schrader TS (2013) Estimating suitable environments for invasive plant species across large landscapes: A remote sensing strategy using Landsat 7 ETM+. Int J Biodivers Conserv 5: 122–134.
- 65. Brown JL (2014) SDMtoolbox: a python-based GIS toolkit for landscape genetic, biogeographic, and species distribution model analyses. Methods Ecol Evol 5: 694–700.
- 66. Liu C, White M, Newell G (2013) Selecting thresholds for the prediction of species occurrence with presence-only data. J Biogeogr 40: 778–789.
- 67. Laporta GZ, Linton YM, Wilkerson RC, Bergo ES, Nagaki SS, et al. (2015) Malaria vectors in South America: current and future scenarios. Parasit Vectors 8: 426. doi: 10.1186/s13071-015-1038-4. pmid:26283539
- 68. Hijmans RJ (2015) raster: Geographic Data Analysis and Modeling. R package version 24–20.
- 69. Middleton KM, Pruim R (2015) abd: The Analysis of Biological Data. R package version 02–8.
- 70. Wilkerson RC, Strickman D, Litwak TR (1990) Illustrated key to the female anopheline mosquitoes of Central America and Mexico. J Am Mosq Control Assoc 6: 7–34. pmid:2324726
- 71. de Arruda M, Carvalho MB, Nussenzweig RS, Maracic M, Ferreira AW, et al. (1986) Potential vectors of malaria and their different susceptibility to Plasmodium falciparum and Plasmodium vivax in northern Brazil identified by immunoassay. Am J Trop Med Hyg 35: 873–881. pmid:3532844
- 72. Bickersmith SA, Lainhart W, Moreno M, Chu VM, Vinetz JM, et al. (2015) A sensitive, specific and reproducible real-time PCR method for detection of Plasmodium vivax and P. falciparum infection in field-collected anophelines. Mem Inst Oswaldo Cruz 110: 573–576. doi: 10.1590/0074-02760150031. pmid:26061150
- 73. Marrelli MT, Floeter-Winter LM, Malafronte RS, Tadei WP, Lourenço de Oliveira R, et al. (2005) Amazonian malaria vector anopheline relationships interpreted from ITS2 rDNA sequences. Med Vet Entomol 19: 208–218. pmid:15958027
- 74. DIVA-GIS (2015) DIVA-GIS.
- 75. Ruiz-Lopez F, Wilkerson RC, Conn JE, McKeon SN, Levin DM, et al. (2012) DNA barcoding reveals both known and novel taxa in the Albitarsis Group (Anopheles: Nyssorhynchus) of Neotropical malaria vectors. Parasit Vectors 5: 44. doi: 10.1186/1756-3305-5-44. pmid:22353437
- 76. Loaiza JR, Scott ME, Bermingham E, Sanjur OI, Rovira JR, et al. (2013) Novel genetic diversity within Anopheles punctimacula s.l.: Phylogenetic discrepancy between the Barcode cytochrome c oxidase I (COI) gene and the rDNA second internal transcribed spacer (ITS2). Acta Trop.
- 77. Obaldia N, Baro NK, Calzada JE, Santamaria AM, Daniels R, et al. (2015) Clonal outbreak of Plasmodium falciparum infection in eastern Panama. J Infect Dis 211: 1087–1096. doi: 10.1093/infdis/jiu575. pmid:25336725
- 78. da Silva-Nunes M, Moreno M, Conn JE, Gamboa D, Abeles S, et al. (2012) Amazonian malaria: Asymptomatic human reservoirs, diagnostic challenges, environmentally driven changes in mosquito vector populations, and the mandate for sustainable control strategies. Acta Trop 121: 281–291. doi: 10.1016/j.actatropica.2011.10.001. pmid:22015425
- 79. Bousema T, Okell L, Felger I, Drakeley C (2014) Asymptomatic malaria infections: detectability, transmissibility and public health relevance. Nat Rev Micro 12: 833–840.
- 80. Alves FP, Durlacher RR, Menezes MJ, Krieger H, Silva LH, et al. (2002) High prevalence of asymptomatic Plasmodium vivax and Plasmodium falciparum infections in native Amazonian populations. Am J Trop Med Hyg 66: 641–648. pmid:12224567
- 81. Alves FP, Gil LH, Marrelli MT, Ribolla PE, Camargo EP, et al. (2005) Asymptomatic carriers of Plasmodium spp. as infection source for malaria vector mosquitoes in the Brazilian Amazon. J Med Entomol 42: 777–779. pmid:16363160
- 82. Coura JR, Suárez-Mutis M, Ladeia-Andrade S (2006) A new challenge for malaria control in Brazil: asymptomatic Plasmodium infection—a review. Memórias do Instituto Oswaldo Cruz 101: 229–237. pmid:16862314
- 83. Waltmann A, Darcy AW, Harris I, Koepfli C, Lodo J, et al. (2015) High Rates of Asymptomatic, Sub-microscopic Plasmodium vivax Infection and Disappearing Plasmodium falciparum Malaria in an Area of Low Transmission in Solomon Islands. PLoS Negl Trop Dis 9: e0003758. doi: 10.1371/journal.pntd.0003758. pmid:25996619
- 84. Stresman GH, Kamanga A, Moono P, Hamapumbu H, Mharakurwa S, et al. (2010) A method of active case detection to target reservoirs of asymptomatic malaria and gametocyte carriers in a rural area in Southern Province, Zambia. Malar J 9: 265. doi: 10.1186/1475-2875-9-265. pmid:20920328
- 85. Silveira P, Damatta RA, Dagosto M (2009) Hematological changes of chickens experimentally infected with Plasmodium (Bennettinia) juxtanucleare. Vet Parasitol 162: 257–262. doi: 10.1016/j.vetpar.2009.03.013. pmid:19345020
- 86. Grim KC, Van der Merwe E, Sullivan M, Parsons N, McCutchan TF, et al. (2003) Plasmodium Juxtanucleare associated with mortality in black-footed penguins (Spheniscus demersus) admitted to a rehabilitation center. J Zoo Wildl Med 34: 250–255. pmid:14582786
- 87. Kissinger JC, Souza PCA, Soares CO, Paul R, Wahl AM, et al. (2002) Molecular phyogenetic analysis of the avian malarial parasite Plasmodium (Novyella) juxtanucleare. J Parasitol 88: 769–773. pmid:12197128
- 88. Earle RA, Huchzermeyer FW, Bennett GF, Little RM (1991) Occurrence of Plasmodium juxtanucleare in greywing francolin: short communication. South African Journal of Wildlife Research 21: 30–32.
- 89. Bennett GF, Warren MW, Cheong WH (1966) Biology of the Malaysian strain of Plasmodium juxtanucleare Versiani and Gomes, 1941. II The sporogonic stages in Culex culex sitiens Wiedmann. J Parasitol 52: 647–652. pmid:5969102
- 90. Bennett GF, Eyles DE, Warren MW, Cheong WH (1963) Plasmodium juxtanucleare a newly discovered parasite of domestic fowl in Malaysia. Singapore Med J 4: 172–173.
- 91. Lourenço de Oliveira R, Castro FA (1991) Culex saltanensis Dyar, 1928—Natural vector of Plasmodium juxtanucleare in Rio de Janiero, Brazil. Mem Inst Oswaldo Cruz 86: 87–94.
- 92. Paraense WL (1944) Infecção experimental do Culex quinquefasciatus pelo Plasmodium juxtanucleare. Mem Inst Oswaldo Cruz 41: 435–440.
- 93. Kangoye D, Noor A, Midega J, Mwongeli J, Mkabili D, et al. (2016) Malaria hotspots defined by clinical malaria, asymptomatic carriage, PCR and vector numbers in a low transmission area on the Kenyan coast. Malar J 15.
- 94. Nourein A, Abass M, Nugud A, El Hassan I, Snow R, et al. (2011) Identifying residual foci of Plasmodium falciparum infections for malaria elimination: the urban context of Khartoum, Sudan. PloS ONE 6: e16948. doi: 10.1371/journal.pone.0016948. pmid:21373202
- 95. Mirghani S, Nour B, Bushra S, Elhassan I, Snow R, et al. (2010) The spatial-temporal clustering of Plasmodium falciparum infection over eleven yearsin Gezira State, The Sudan. Malar J 9.
- 96. Badu K, Gyan B, Appawu M, Mensah D, Dodoo D, et al. (2015) Serological evidence of vector and parasite exposure in Southern Ghana: the dynamics of malaria transmission intensity. Parasit Vectors 8.
- 97. Patil A, Gething P, Piel F, Hay S (2011) Bayesian geostatistics in health cartography: the perspective of malaria. Trends Parasitol 27: 246–253. doi: 10.1016/j.pt.2011.01.003. pmid:21420361
- 98. Bousema T, Stresman G, Baidjoe A, Bradley J, Knight P, et al. (2016) The impact of hotspot-targeted interventions on malaria transmission in Rachuonyo South District in the western Kenyan highlands: a cluster-randomized controlled trial. PLoS Med 13: e1001993. doi: 10.1371/journal.pmed.1001993. pmid:27071072
- 99. Marshall J, Touré M, Ouédraogo A, Ndhlovu M, Kiware S, et al. (2016) Key traveller groups of relevance to spatial malaria transmission: a survey of movement patterns in four sub-Saharan African countries. Malar J 15.
- 100. Ruktanonchai N, DeLeenheer P, Tatem A, Alegana V, Caughlin T, et al. (2016) Identifying malaria transmission foci for elimination using human mobility data. PLoS Comput Biol 12: e1004846. doi: 10.1371/journal.pcbi.1004846. pmid:27043913