Entomological Monitoring and Evaluation: Diverse Transmission Settings of ICEMR Projects Will Require Local and Regional Malaria Elimination Strategies

The unprecedented global efforts for malaria elimination in the past decade have resulted in altered vectorial systems, vector behaviors, and bionomics. These changes combined with increasingly evident heterogeneities in malaria transmission require innovative vector control strategies in addition to the established practices of long-lasting insecticidal nets and indoor residual spraying. Integrated vector management will require focal and tailored vector control to achieve malaria elimination. This switch of emphasis from universal coverage to universal coverage plus additional interventions will be reliant on improved entomological monitoring and evaluation. In 2010, the National Institutes for Allergies and Infectious Diseases (NIAID) established a network of malaria research centers termed ICEMRs (International Centers for Excellence in Malaria Research) expressly to develop this evidence base in diverse malaria endemic settings. In this article, we contrast the differing ecology and transmission settings across the ICEMR study locations. In South America, Africa, and Asia, vector biologists are already dealing with many of the issues of pushing to elimination such as highly focal transmission, proportionate increase in the importance of outdoor and crepuscular biting, vector species complexity, and “sub patent” vector transmission.


INTRODUCTION
The unprecedented global efforts for malaria elimination in the past decade have resulted in the reduction of malaria cases in several settings, 1 but also in dramatic increases in resistance to pyrethroids and other insecticides, 2 changes in the relative importance of outdoor (residual) malaria transmission, and major shifts in biting time, for example, Anopheles farauti in the Solomon Islands 3 and Anopheles funestus in Benin and Senegal. 4,5 Together these new trends have already resulted in quantifiable changes in human-vector interactions in several endemic areas, and threaten to jeopardize future gains. Longlasting insecticidal nets (LLINs) and indoor residual spraying (IRS) and have been the mainstays of malaria control and have had a major impact on reducing global malaria, particu-larly where vectors are primarily endophagic (indoor biting), endophilic (indoor resting), and anthropophilic. 6 As such, the goal of global malaria elimination will require additional interventions and improvements in both the application of current control measures and entomological monitoring. 7 The single biggest threat to sustainable malaria control is insecticide resistance, which has reached alarmingly high levels in some vector populations of Africa, India, and China (M. L. Quiñones and others, unpublished data). 2 Second, there are indications of local adaptation in vector biting behavior, possibly in response to reliance on LLINs and IRS. 3,5,8,9 Whether this reflects a lack of vector ingress because of physical barriers, that is, mosquito-proof houses, adaptation of endophagic vectors to exophagy (outdoor feeding), or selection on phenotypic plasticity, is unknown. 10,11 It has been hypothesized that in some areas endophagic populations may have been eliminated, leaving the inadequately controlled exophagic population. 12 Also, in the Solomon Islands during the 1970s malaria eradication campaign, late night biting of Anopheles koliensis and Anopheles punctulatus, which had been common, virtually *Address correspondence to Jan E. Conn 13 Similarly, a switch from endophily to exophily (outdoor resting) has been documented in areas under intense IRS (R. Sloof, unpublished data). 14,15 The third major issue is the recognition that transmission is both focal and heterogeneous and that we urgently need to incorporate, for example, ecological context of mosquito foraging behavior and vector diversity into our transmission models to improve predictive accuracy. 16,17 Fourth, the use of LLINs at high coverage, although extremely effective overall, can alter species composition, which could change transmission patterns and possibly the entomological inoculation rate (EIR) because of different vectorial capacities, biting times, and behaviors, for example, a decrease in Anopheles gambiae and a concurrent increase in the relative proportion of Anopheles arabiensis. 6,18,19 Vectorial systems vary dramatically across regions and countries, 20 -22 and this variation will be reflected in how well malaria transmission responds to control. A suggested benchmark for adequate vector control is the decrease and maintenance of EIR below 1, together with epidemiological measures of malaria in humans. 23 Achieving EIRs of < 1 is especially challenging in those endemic areas where the current approach of IRS and/or LLINS may not be adequate to cover changing transmission scenarios.
The United States' National Institutes of Health funded 10 International Centers of Excellence in Malaria Research (ICEMR) in 2010 with a series of common aims including a concerted effort to closely link epidemiology and transmission metrics with vector biology. One of the unique features of the ICEMR program is the focus on longitudinal surveillance sites in diverse epidemiological settings across the globe (Figure 1). Each ICEMR uses similar approaches and metrics to quantify transmission. The strengths of this approach are our ability to incorporate seasonal and multiyear variation in routine entomological monitoring that can quantify temporal changes in insecticide susceptibility, EIR, vector species composition, and the effects of epidemiology interventions. Such data, when incorporated into malaria transmission models, should increase accuracy and predictive power.
The objective of this article is to provide an introduction to the broad-sense ecology of vectors in the 10 major geographic regions covered by the ICEMR projects and to discuss how similarities and contrasts between the areas will build to a comprehensive view of malaria transmission globally. We are not intending to provide a detailed historical review of transmission ecology in each setting, and as a consequence the extant literature has been sampled broadly but with only limited depth. Vector biologists from each ICEMR selected references that they believed to be the most pertinent to the objective of this article.

ENTOMOLOGICAL METRICS AT ICEMR SITES
Across the ICEMR sites, there is diversity in vector species and their contributions to malaria transmission of, principally, Plasmodium falciparum and Plasmodium vivax are variable. Although challenging, this is also an extraordinary opportunity to identify commonalities that may lead to new integrated approaches to control and eliminate malaria. Investigations in the 10 major regions are described in brief below, with summary vector biology data (Table 1) together with a corresponding figure of the 33 individual sites geographically located (Figure 1). The wide range of primary vector species, and putative new vectors that several ICEMR studies have detected, is illustrated in Figure 2.
Africa. The four ICEMRs in Africa (Malawi, west Africa, southern Africa, and east Africa; Figure 1) are near exclusively P. falciparum transmission settings where malaria is vectored by one or more of the four major African vectors (An. gambiae s.s., Anopheles coluzzii, An. arabiensis, and An. funestus s.s.) ( Figure 2). Vector control is reliant on LLINs, IRS, or some FIGURE 1. Location of the 33 field sites across 10 International Centers of Excellence in Malaria Research (ICEMR) (in red). The field sites (yellow circles) are numbered consecutively and are described in Table 1.     (Table 1). In west and east Africa, vectors are primarily endophagic and endophilic (but see An. arabiensis, Table 1), implying that nets should be highly effective, assuming high levels of local coverage. For two of the higher transmission sites in southern Africa, An. funestus is similarly expected to be endophagic and endophilic.
Although not yet reported for all the African sites, EIRs are, with the exception of the high transmission site in Tororo, Uganda (annual PfEIR~125; Table 1), 23 moderate relative to earlier studies. 1 A major concern for these ICEMRs is how vectors will respond to the massive rollout of LLINs in sub-Saharan Africa. Already, some populations of An. gambiae s.s., such as those in Guinea, have shifted from primarily endophagic biting to primarily exophagic 1,8 ; whereas in other situations, such as the highlands of Kenya, endophilic An. gambiae have been dramatically reduced, with low transmission maintained by An. arabiensis and novel anopheline species that are primarily exophilic and bite early in the evening when people are generally unprotected. 12 East Africa ICEMR. Comparing three transmission sites (low, moderate, and extremely high; Table 1) from this Uganda-based ICEMR, only at the low site, Jinja, there was a suggestion of a reduction in An. gambiae s.s. and a concurrent increase in the abundance of An. arabiensis compared with previous findings. 25,26 The study site in Jinja is peri-urban, and it is possible that the predominance of An. arabiensis reflects its adaptation to this disrupted environment as has been observed in west Africa. [27][28][29] Malawi ICEMR. This ICEMR is undertaking studies of Plasmodium transmission and malaria risk and prevention in various environments of southern Malawi, particularly in the districts of Blantyre, Chikwawa, and Thyolo ( Figure 1). 30,31 Study sites differ considerably, ranging from predominantly rural lowlands in Chikwawa where transmission is intense and essentially year-round, to the rural highlands of Thyolo with moderate seasonal transmission, to sites in and around urban Blantyre City, with apparently lower-level, heterogeneous infection (Table 1). Until the ICEMR-supported research began, little had been published about Anopheles vectors in Malawi. One of the first studies by Spiers and others 32 undertaken in Chikwawa reported that the predominant vectors were An. arabiensis and An. funestus s.s., although An. gambiae s.s. was also present in this Shire River valley area. Other work on filariasis vectors by Merelo-Lobo and others 33 also found these three species, with An. funestus being the most abundant. However, little is known about the relative abundance and role of Anopheles species in relation to malaria patterns from Malawi.
Vector ecology and infection studies are now underway as part of two ICEMR-supported projects involving a) health facility-identified, case-control comparisons of urban/peri-urban households in/around Blantyre City and b) cross-sectional, household visit-based sampling across districts of Chikwawa, Thyolo, and Blantyre. In all of these settings, Prokopack-style aspirators and Centers for Disease Control and Prevention (CDC) light traps are being used to test for the presence of indoor adult mosquitoes that are identified by microscopy and confirmed by polymerase chain reaction (PCR) ( Table 1). In rural settings, An. funestus s.s. is most abundant, with fewer An. arabiensis and rare Anopheles quadriannulatus. In urban/ peri-urban Blantyre, year-round aspiration of 511 households during April 2012 through March 2014 showed that 64% of houses had mosquitoes, with Culex spp. representing 98.7% of the sample (M. Wilson, personal communication). Very few Anopheles spp. (12 males, 29 females) were found. Nevertheless, more Anopheles were captured in households of cases (4.2%) than of controls (1.9%). In this urban setting, it remains very difficult to find many Anopheles using aspiration and light traps.
Public health efforts to reduce vector-human contact have used widespread distribution of LLINs and focused use of IRS. 34 Nationwide distribution of free LLINs has been enhanced since 2012 when ownership was at only 58% of households, and now all children born in health facilities receive an LLIN, as do pregnant women when they first visit an antenatal care clinic. Similarly, a free LLIN is now given to each child at her/his first Expanded Program on Immunization (EPI) visit. From 2008 through 2012, more than 6 million LLINs were distributed in Malawi. 34 However, Plasmodium infection measured by PCR among children under 5 years of age was still 43% overall, and 60.5% in the lowest wealth quintile. The scale-up plan for 2012-2015 aims to achieve one LLIN for every two people in each household. 34 Other control efforts using IRS are coordinated by the Malawi government, but this program is not nationwide, instead focusing on seven districts of particularly high disease burden: Karonga and Nkhata Bay (northern region), Nkhotakota and Salima (central region), and Mangochi, Chikwawa, and Nsanje (southern region). The 2012 MIS survey indicated that less than 10% of Malawi households had received IRS within the preceding 12 months, suggesting that this form of vector control is relatively less important. 34 Effectiveness of interventions to reduce vector-human contact depends on where, when, and on whom competent Anopheles are feeding, but again little is known about this in Malawi. Recent work in northern Malawi (Karonga) has shown that An. funestus and Anopheles rivulorum were mostly found indoors, but none were infected with either P. falciparum or P. vivax. 35 A new An. funestus-like species was also mostly collected indoors, but mainly had fed on animals and also was uninfected. Other studies from countries that border Malawi support the general pattern that both An. funestus and An. arabiensis predominantly feed indoors and on people. Investigations in southeastern Zambia (~500 km west of southern Malawi) have shown that An. funestus and An. quadriannulatus were captured both indoors and outdoors, but nearly all were found to have fed indoors, reinforcing the importance of LLIN use. 36 More generally, this pattern of An. gambiae complex and An. funestus group predominantly biting humans indoors at night seems to be common in eastern Africa. 37 West African ICEMR. Three malaria endemic countries: Mali, Senegal, and The Gambia, comprise the focus of this ICEMR ( Figure 1). Across this broad geographical area vector populations and malaria transmission differ in their complexity. The site types (riverine: Gambissara, Dangassa; urban: Medina Fall, and rice irrigation: Dioro) were chosen in part to explore differences in length of transmission season, EIR (from~5 to 51 bites/month; Table 1), and status of malaria control. 38 Anopheles gambiae and An. arabiensis are the major vectors in all the three countries ( Figure 2). However, other anopheline vectors are encountered such as An. funestus and Anopheles pharoensis inland and Anopheles melas on the coast. [39][40][41][42][43] In The Gambia and Mali sites, as in the other African ICEMR localities (Figure 1), An. gambiae is mainly endophagic, although since the inception of this study, it has been collected feeding outdoors 45-50% of the time (M. Coulibaly, personal communication), similar to the change in behavior documented on Bioko Island, Equatorial Guinea. 8 Anopheles arabiensis is primarily exophagic except in the urban site of Medina Fall where it feeds both indoors and outdoors (Table 1).
Overall malaria transmission is seasonal and coincident with the rainy season. The peaks of transmission occur toward the end of the rains when mosquito densities are waning. 38 Nevertheless, transmission is perennial in some areas where irrigated rice cultivation maintains anopheline breeding during the dry season. 44 The urban site of Medina Fall (Senegal) showed the lowest transmission level. The current large-scale vector control strategies in use in Mali, The Gambia, and Senegal are LLINs and IRS. Although the LLINs distribution is country wide at all the three sites through campaigns and routine antenatal consultation and EPI, IRS has been implemented only in targeted areas in the respective countries. None of the west African ICEMR study sites has received IRS to date, but insecticide resistance is widespread.
Southern Africa ICEMR. In Macha, Choma District, southern Zambia (Figure 1), there is marked spatial and temporal heterogeneity in the foraging behavior of the main vector An. arabiensis and previously undocumented high anthropophily in secondary vectors Anopheles coustani s.l. and Anopheles squamosus (Figure 2). [45][46][47] Choma District has the potential to become a malaria-free zone, in part because the formerly primary vector, An. funestus s.s., was locally eradicated in 2004, possibly by a drought. 48 Even though the Macha population of An. arabiensis is highly anthropophilic with foraging times that extend from dusk until dawn with a greater tendency for exophagy (Table 1), no specimens have been detected positive for Plasmodium since 2006 and transmission in the district remains very low (D. Norris, personal communication). 46,49 Despite the low overall risk, it should be recognized that individual and household risk is very unevenly distributed and spatially clustered, and most importantly, this heterogeneity may be further exacerbated by anti-vector interventions and multiple host feeding by the vector. 16 In contrast, both endophagic An. funestus s.s. and An. gambiae s.s. are responsible for very high levels of transmission despite reported coverage of 1.73 LLINs per person (2012) and greater than 90% coverage with IRS (2011) in Nchelenge District, northern Zambia (S. Das, unpublished data). 50 The inability to achieve control here is likely due to high levels of insecticide resistance to dichlorodiphenyltrichloroethane and deltamethrin and an inability to apply effective control measures to the vector populations that are physically difficult to access or may reside in transient households (D. Norris, unpublished data), a condition further limited by resources. 50 In Nchelenge District, the two vector species exhibit enormous temporal and spatial heterogeneity, which is hypothesized to exacerbate the observed perennial year-round transmission (S. Das and D. E. Norris, unpublished data) ( Table 1). High rates of feeding on multiple human hosts in a single gonotrophic cycle (S. Das and D. E. Norris, unpublished data) and human movement into malaria risk zones are seen as added challenges to control in this area (K. M. Searle and W. J. Moss, unpublished data).
The third site for this ICEMR is Mutasa, eastern Zimbabwe (Figure 1), where resurgent malaria occurs seasonally and An. funestus s.s. appears to be the only significant vector. Although current loss of vector control here is likely due largely to insecticide resistance of An. funestus s.s. (M. Coetzee, and others, unpublished data), historically An. gambiae s.s. was the primary vector in this region. 51 This change in primary vectors may be attributed to gaps in malaria control because of economic constraints that allowed mainly endophagic An. funestus s.s. to invade from nearby Mozambique or emerge from unknown refugia. Insecticides used subsequent to this event, to which the An. funestus s.s. population were likely already resistant, would have helped this invasive vector population to fully establish and thrive.
Latin America. In marked contrast to the African ICEMR sites, most malaria in Latin America is caused by P. vivax (~70%), except for relatively uncommon hot spots such as Haiti, Guyana, and gold-mining areas across the Amazon, where P. falciparum case numbers are higher than the aver-age~30%. 1 In the vast area of the basin drained by the Amazon and its tributaries, Anopheles darlingi is the main vector, 52,53 but, as is evident in the Latin American study sites, several other vector species contribute to transmission, and much less is known about their ecologies and entomological metrics (Table 1). Many vector species in the neotropics are exophagic and exophilic (Table 1), with the notable exceptions of An. darlingi, Anopheles albimanus, and Anopheles nuneztovari (see the summary below for Latin American ICEMR), which display endo/exophagy depending on host availability and environmental characteristics. 65 Therefore, control by IRS has been a mainstay for many years, and, partly for reasons of logistics and distribution, the use of LLINs has spread more slowly in Latin America than Africa, Asia, or the southwest Pacific. 52,55,56 An unresolved issue is the relatively high use of IRS combined with very low levels of insecticide resistance (M. Quinones, personal communication).
Amazonian ICEMR. There are very few reports in Latin America where An. darlingi is no longer the dominant malaria vector, for example, Suriname. 53 Infrequent, extensive flooding that coincided with the beginning of the interventions in Suriname likely contributed to the local collapse of An. darlingi. 53 Anopheles darlingi is the predominant vector in study sites near Iquitos, Peru, and near Acre, western Brazil, in this ICEMR. In these localities An. darlingi is the main vector, the most abundant, highly seasonal, exo-and endophagic, and nearly exclusively exophilic (M. Moreno and others, unpublished data) ( Table 1).
Despite Ministry of Health and international (e.g., Control de la Malaria en las Zonas Fronterizas de la Región Andina: Un Enfoque Comunitario [PAMAFRO]) efforts to distribute LLINs in Brazil and Peru from 2006 to 2011, unimpregnated net use remains common in some localities, although an integrated approach of LLINs combined with IRS has been recommended. [56][57][58][59] Many populations of Amazonian An. darlingi, including those in our study sites (M. Moreno and others, unpublished data) display multimodal biting times. 60 A crepuscular peak (~19-21 hours) is common, well before most people retire for the evening reducing the potential impact of LLINs. 61 Major issues in Peru are the correlation of deforestation with significantly high human biting rates along highways 62 and in riverine settlements (W. Lainhart, unpublished data), and hyperendemic malaria hot spots related to occupational travel. 62,63 In western Amazonian Brazil, deforestation linked to agricultural settlements and gold mining is of primary concern. 56 In study sites in Madre de Dios region, southern Peru, An. darlingi was not common, and both Anopheles rangeli and Anopheles benarrochi B were detected infected with P. vivax for the first time in this region (Table 1, Figure 2), but sample size was so small that the actual role of these species in transmission could not be evaluated (M. Moreno and J. E. Conn, unpublished data). These data suggest that elimination efforts might be concentrated more usefully on the detection and rapid treatment of occupational malaria transmission hot spots. Initial blood meal data (M. Moreno and J. E. Conn, unpublished data) from barrier screens support previous findings from eastern Amazonian Brazil that An. darlingi feeds opportunistically, and strongly suggest that host availability is the prime driver of blood meal preference. 64 It remains to be seen how best to exploit these new findings to improve vector control in this region.
Latin American ICEMR. Latin American countries in the Centro Latino Americano de Investigación en Malaria (CLAIM) include Guatemala, Panama, Colombia, and northwestern Peru. 52 In this broad area, the most important regional vectors are An. darlingi, An. nuneztovari s.l., and An. albimanus (Table 1, Figure 1). 21 Nevertheless, in one field site in the Pacific region, Tumaco (Figures 1 and 2, Table 1), the species Anopheles calderoni was found infected with P. vivax (M. L. Quiñones, unpublished data). It was also infected with P. falciparum in specimens from a palm-oil plantation in the same region. 55 These data infer that An. calderoni may be relatively important in local transmission, and ecological and biological investigation in addition to control efforts should be increased. There are many critical information gaps for these species, such as lack of data on vector ecology, vector competence, and effects of environmental change on vectors. 54 The most common malaria control methods have been IRS, LLINs, and early detection, diagnosis, and treatment. 23 Regrettably, for LLINs, there have been basically no evaluations of the potential suppression of vector populations, vector behavioral changes, transmission level, or location. 54 Behavioral changes toward increased exo-and endophagy in the northern part of the geographical range of An. albimanus (Mexico and Panama) may have been induced by early adoption of IRS. 3 Overall, the predominant behavior among these species is exophagy and exophily (Table 1); however, in several localities in the Amazon, An. darlingi is mainly endophagic and An. albimanus displays considerable plasticity, exhibiting both behaviors, depending on host availability and locality. 21,65 Broadly distributed across northern South America, An. nuneztovari is more exophagic in the Amazon (where it may be Anopheles goeldii) and more endophagic in Colombia and western Venezuela. 66 As such, decisions on the most appropriate intervention(s) differs across its geographic range necessitating more locally tailored control than that seen in Africa. 65 New findings of a cross-sectional study using human-landing catch (HLC) in 70 localities in western Colombia (Cordoba, Narino, and Valle), where most transmission occurs, found that An. albimanus and An. nuneztovari together constituted approximately 80% of the 12,052 adult mosquitoes collected and identified, and these were the only two species positive for Plasmodium by enzyme-linked immunosorbent assay (ELISA). Furthermore, 35% of these adults were endophagic. Of all An. albimanus collected,~22% were endophagic, compared with~45% endophagic An. nuneztovari (M. L. Quiñones, unpublished data). A survey of breeding sites found that most positive water bodies were either fish ponds or small reservoirs for domestic use; 70% of all larvae were An. nuneztovari. At least where An. nuneztovari is most abundant and endophagic, continued use of LLINs combined with focal application of larvicides might be the most effective tools, even though malaria elimination in the near term may be difficult to achieve.
South Asia ICEMR. Entomological results from this ICEMR have implicated Anopheles stephensi, collected during 85 nights from multiple urban and rural localities in Goa, western coastal India, as a vector of P. falciparum (Figures 1 and 2). Panaji City, within Goa, had an EIR of 18.1 compared with an overall EIR of 2.35 for all of Goa (Table 1). 67 In this city, An. stephensi is endophagic, but rests outdoors. Most An. stephensi (N = 55) were actively biting between 03:00 and 06:00, although there were seasonal differences. Both Anopheles fluviatilis (N = 75) and Anopheles culicifacies (N = 32) were collected biting humans but neither species was positive for Plasmodium. Mosquito control in and around Goa relies on larval suppression using fish and larvicides (Table 1).
In Wardha (central India) and Ranchi (eastern India) (Figure 1), An. culicifacies and An. fluviatilis are exophagic and exophilic. Here, IRS is the only vector control, and transmission is perennial with peaks during the rainy season. In Ranchi, An. culicifacies transmits year long, with peaks during the rainy season, and An. fluviatilis transmits primarily during February and March. In Assam state, northeastern India, the vectors are Anopheles baimai and Anopheles minimus. This is an atypical part of India in which four Plasmodium species circulate and are transmitted (Table 1). Malaria is also perennial here. To date, EIRs have not been determined for the localities in Wardha, Ranchi, or Assam.
India ICEMR. Urban and rural sites with contrasting transmission dynamics are the main focus of this ICEMR. In urban Chennai, India, a consistent hot and humid climate supports stable, low level transmission of predominantly P. vivax malaria by a single vector species, An. stephensi (Figure 1, Table 1). Malaria control in Chennai follows strategies adopted by the Urban Malaria Scheme of the national program of India wherein vector control is based on antilarval measures such as the use of abate (temephos), application of Bacillus thuringiensis israelensis formulations, and, to a certain extent, larvivorous fish, Gambusia affinis. Despite a reduction in malaria prevalence 68 the disease persists, possibly due to rapid urbanization, regular reintroduction, large numbers of breeding sites, a submicroscopic and/or asymptomatic parasite load, and the difficulty of targeting dormant stages of P. vivax. Mosquito control tends to be restricted to application of larval insecticides, targeting known breeding habitats of An. stephensi such as wells, overhead tanks, and other water storage containers. Use of interventions against the adult vectors within domestic dwellings, such as LLINs or IRS, is minimal. The reasons for the lack of adult mosquito control are varied but include the extremely dense and complex nature of the housing within urban slum settings (it is logistically challenging to access every house), discomfort in using nets in the hot and humid conditions, low transmission rates (there are many challenges at the household level above and beyond occasional infection with generally nonlethal P. vivax), and very low density of adult populations of An. stephensi. Indeed, determining where and when local transmission occurs is very difficult. The adult vectors are highly zoophagic (human blood index [HBI] = 0.028) and almost exclusively found in cattle sheds (Table 1). However, these biting and resting behaviors create the potential for novel control strategies targeting the more limited focal sites (i.e., cattle sheds) with tools such as toxic sugar baits, or possibly treating cattle and other livestock directly with insecticides or antihelminthics. Nonetheless, given that the majority of adult malaria vectors are not feeding on humans or resting in domestic dwellings, such focal interventions targeting zoophagic and exophilic behavior could have a dramatic impact on local transmission, which appears to be almost a secondary foraging "spillover" phenomenon.
Vector control practices in forested tribal areas such as Raurkela, follow more established approaches with intensive IRS and LLIN programs. The perennial vector in these rural settings is An. culicifacies, which tends to be zoophagic and exophilic. The most common sibling species, B, is refractory to P. vivax. 69,70 The primary vector responsible for peak P. falciparum transmission is An. fluviatilis. This species is restricted to breeding in slow-moving fresh water that occurs post-monsoon, and so exhibits highly seasonal dynamics. The predominant sibling species, S, is an efficient vector, with previous studies showing it to be highly anthropophagic and endophagic, with a HBI up to 0.90. 71,72 Recent results may suggest a shift in An. fluviatilis (M. B. Thomas, unpublished data) feeding behavior, mirroring changes observed elsewhere in response to wide-scale use of IRS and LLINs. 8,73,74 Studies suggest that An. fluviatilis has shifted from resting within human dwellings to semi-enclosed animal sheds. 75 This apparent behavioral change is actually species replacement; the S type now comprises only 20% of the 2013-2014 population and the zoophagic and exophilic sibling species T, now comprises the majority. This replacement might be due to increased comparative fitness of T during control measures (i.e., some form of competitive replacement), or it could be that S is simply disproportionately affected by interventions with T, feeding in cattle sheds, unaffected by IRS and LLINs, and remaining at similar absolute levels but showing a relative increase. These patterns (including the challenges in interpretation) again mirror those reported elsewhere. 8,73,74 Southeast Asia ICEMR. The China-Myanmar and Thai-Myanmar border regions have been the geographical emphasis of this project (Figure 1). The Government of China has set a goal of malaria elimination by 2020, and Thailand is pursuing spatially progressive elimination and has a national goal to eliminate malaria from 80% of the country by 2020. 76 However, high malaria incidence in neighboring Myanmar and cross-border human movement present major challenges for malaria elimination in China and Thailand. 1 Therefore, understanding vectorial systems and developing site-appropriate transmission control methods in the border regions are crucial.
The study sites on either side of the China-Myanmar border area are separated by less than 10 km but have significantly different vectorial systems. In the Chinese sites (around Nabang town in Yingjiang County, Yunnan Province), the major malaria vectors in the 2010-2012 survey by CDC light traps were An. minimus, Anopheles sinensis, and Anopheles maculatus (Figure 2). Anopheles minimus and An. maculatus are endophagic and anthropophilic whereas An. sinensis is generally exophagic (Table 1). 77 Anopheline density was highly variable among sites and between seasons. Peak mosquito density was in May, during the rainy malaria transmission season. Molecular taxonomy of a subsample of An. minimus s.l. found that the only member of this complex present was An. minimus s.s. Moreover, of the anophelines examined, only An. minimus was detected infected with P. vivax, albeit only one positive individual out of 1,500 tested (0.07%). Blood meal analysis of a modest sample size (N = 104) revealed that humans were the main host (82.6%), followed by cattle, pigs, and dogs. 77 Mixed blood meals of human/pig and human/cow were detected at low frequency (0.9% each). In the Myanmar study sites near Laza Town, Kachin Special Zone, CDC light trap collections in 2013-2014 found An. minimus was predominant, constituting 89% of anophelines collected despite high anopheline species diversity (20 species collected). The sporozoite rate of An. minimus (1.8%) in Laza Town, Myanmar, was considerably higher than in Nabang Town on the Chinese side.
It is important to note that malaria vector species composition varied significantly by mosquito collection methods. For example, larval mosquito collection in the same study sites in Myanmar found that the four most abundant vectors were An. sinensis, Anopheles barbirostris, An. minimus, and Anopheles splendidus (G. Yan, unpublished data) rather than An. minimus. Larval and adult survivorship in life table studies found differential survival because of local climate conditions as well as land use and land cover variation (G. Yan, unpublished data). In any predictive model for malaria, the environmental variables that effect anopheline survivorship need to be measured and incorporated. 17 On the Thai-Myanmar border area, the main vectors are An. minimus, An. maculatus, and Anopheles annularis. All three species rest indoors and outdoors, so LLINs should exert some control on the former portion of their populations.
Southwest Pacific ICEMR. The geographic foci of the southwest Pacific ICEMR are the Madang and East Sepik Provinces in Papua New Guinea as well as Central and Western Provinces in the Solomon Islands. Current vector control in Papua New Guinea and the Solomon Islands relies on LLINs with limited use of IRS in the Solomon Islands. 78 The main malaria vectors across the region are members of the An. punctulatus group: An. punctulatus s.s., An. koliensis, and the An. farauti complex, which consists of eight cryptic species (Figure 1), and the diversity of these species varies across sites. [79][80][81] Two of the primary vectors in the southwest Pacific, An. koliensis and An. punctulatus, which were endophagic and more anthropophic than An. farauti, were dramatically reduced by IRS, and An. koliensis may have disappeared from the Solomon Islands. 13 In Papua New Guinea, two other species in the An. farauti group are important vectors: Anopheles hinesorum and An. farauti 4, with An. farauti 6 and An. farauti 8 considered to be minor vectors with circumsporozoite antigen detected by ELISA. 82 Other minor malaria vectors in Papua New Guinea are, either uncommon or with a limited geographical distribution, Anopheles longirostris, Anopheles bancroftii, Anopheles subpictus, and Anopheles karwari. 81,82 Collections made by HLC in Papua New Guinea found An. farauti s.s. was the predominant species in Mirap village whereas An. punctulatus was the most common species found in the inland villages of Yauatong and Wasab in the East Sepik Province. The minor vector, An. longirostris, had a relatively high population density in Wasab village. In the Solomon Islands, the only species caught biting humans at significant densities in Central Province was An. farauti s.s.
In Papua New Guinea, the populations of An. punctulatus and An. farauti were mostly zoophilic, late night biting, and exophilic but exhibited both exophagic and endophagic biting habits. 83 Recent data from the Central Province of the Solomon Islands reported An. farauti as highly anthropophagic with more than 90% of blood meals on humans (Table 1). 84 Both the timing of night biting behavior (from late to early) and location (from indoors/outdoors to predominantly outdoors) of An. farauti in the Solomon Islands shifted in response to selective pressure to avoid insecticides following IRS decades ago and these behavioral changes have persisted. 3 Physiological resistance to insecticides has not been found yet in Papua New Guinea or the Solomon Islands, with the temporal biting shift to earlier possibly providing a behavioral resistance mechanism to minimize exposure of this vector to insecticides. 85 This change in blood-feeding behavior has appeared independently on multiple islands in this archipelago, suggesting that LLINS and IRS will have a limited impact on malaria transmission for this important regional vector. 86 Elimination may require the use of supplemental and complementary interventions to be implemented with LLINs.

LESSONS LEARNED FROM VECTOR BIOLOGY ACROSS ICEMRS
Collection methods. The wide range of trap types and/or control methods favored by each ICEMR is a result of projectspecific research questions, trap type collector bias, and controversy surrounding the HLC method because of a perceived infectivity risk to collectors (see Gimnig and others 87 ). Furthermore, HLC is expensive and labor intensive. In several of the ICEMR sites, CDC traps combined with either pyrethroid spray catch or aspiration have replaced or supplemented HLC (Table 1). A recent study in the three east Africa ICEMR sites revealed statistically comparable EIRs for CDC compared with HLC such that the former can safely and effectively replace the latter. 25 On the other hand, the Amazonian ICEMR found that HLC resulted in significantly higher numbers of An. darlingi compared with Shannon traps and CDC traps, although the number of infected mosquitoes was so low that EIRs could not be compared (M. Moreno and others, unpublished data).
An important outcome of the ICEMR vector biology studies is a new push to standardize monitoring across sites by the use of barrier screens, recently developed in Indonesia and the south Pacific. 84 Advantages of this method include simplicity of construction and use of local material, collection of high numbers of both nulliparous and parous anophelines, collection of infected anophelines (to date only evidence from Iquitos; M. Moreno and others, unpublished data), no bias related to feeding preference (humans and animals are not involved as attractants), and physical integrity of specimens collected. The Amazonian, Latin American, southeast Asia, southern Africa, and southwest Pacific ICEMRs are currently testing these traps for effectiveness in collecting resting and host-seeking anophelines across the diverse settings, habitat types, and vector species.
New putative malaria vectors. Using standard incrimination criteria (presence of infected, correctly identified anthropophilic vectors concurrent with malaria transmission), evidence of new or potential vector species has been collected in four of the 10 ICEMRs (Figure 2). Although EIRs have been determined only for An. calderoni in Colombia, they are in progress for An. rangeli and An. benarrochi B in southern Peru, and will be calculated, pending vector confirmation, for the additional species and sites (i.e., An. subpictus in Goa, possibly An. coustani s.l. and An. squamosus in Macha). [45][46][47] In Western Province, Solomon Islands, Anopheles lungae is the most common anopheline collected by HLC in several villages. However, incrimination as a potential vector of human malaria awaits confirmation of the presence of sporozoites by PCR or ELISA. 81,82 How these and other putative new vectors will respond to changes in LLIN use or climate change is being actively addressed in the Amazonian ICEMR. 88 Value of longitudinal surveillance sites to malaria elimination. Highlights of the value of these sites thus far include the following: An. darlingi is more locally opportunistic than anthropophilic.
Each of these discoveries contributes to more accurate EIR values and provides feedback to parallel epidemiological and parasitological studies ongoing in the ICEMR sites.
Management of outdoor (residual) transmission. Overall, malaria vector control in the ICEMR study sites is reliant on the two insecticide-based interventions for which there exists a strong, primarily Africa-derived evidence-base: LLINs and IRS. 89,90 However, the ICEMR sites reveal a variety of transmission scenarios that will require a more tailored approach that can be monitored and modified rapidly as the need arises. Where endophagy remains dominant and vectors are resistant to pyrethroids, the deployment of attractive toxic sugar baits (ATSB) indoors in combination with LLINs, is one possibility 91,92 although indoor ATSBs may be temporally unsustainable (S. Lindsay, personal communication). Reduction of crepuscular human-vector contact outside houses might be accomplished by the use of ATSB outdoors, as this intervention was predicted to be especially effective against An. arabiensis, which is primarily exophilic. 91 Furthermore, if these exophilic populations are also mainly zoophilic, treating nearby animal hosts as suggested for An. fluviatilis, in India, as mentioned above, could be an effective part of an integrated vector control plan. 93,94 Additional important options to prevent outdoor transmission in the context of integrated control include larviciding, as used in the ICEMR urban sites of Tumaco, Goa, Chennai, and Nadiad (Table 1), and environmental management. 95,96 More broadly, new interventions could include transgenic mosquitoes, sterile male releases, or cost-effective consumer products.

CONCLUSIONS
Data from the ICEMRs clearly illustrate that malaria transmission and vectors are highly spatially and temporally heterogeneous. In addition, behaviors exhibited by many vector species involved are diverse, and although they can be broadly categorized as endophagic, exophagic, endophilic, and exophilic, most vectors exhibit a mix of behaviors (e.g., some "outdoor feeding" vectors will occasionally blood feed indoors). Local behavioral adaptations will require new combinations of sampling, surveillance, and control tools. For example, at one location a program may have to address the control of multiple species, but also the control of a single species that can present multiple behaviors.