The results that follow summarizes the semi-structured interviews and represent the current perceived advantages and disadvantages for various commonly used Anopheles surveillance tools and techniques by the global community.
Sampling adult mosquitoes
The tools used to sample blood seeking adult mosquitoes were classified into five categories: the human landing catch (HLC); CDC an trap (with or without lights or lures including placement near an occupied bednet) [11]; human-baited traps (including host decoy [12], odour baited entry [13], Ifakara tent [14], Flavela tent [15] and electrocuting grids near humans [16]); animal-baited traps (i.e., animal baited nets [17], animal baited huts, Magoon stable traps [18], and barrier fences around animals); other traps (including UV light, updraft, the BG-Suna [19] and the BG-Sentinel [20]) (Tables 1 and 2). Each tool category had unique strengths and weaknesses identified by the 40 informants.
Human landing catch
The HLC was the most frequently used technique (Table 1). It was considered to be highly effective and was used to monitor more indicators (determination of peak biting behaviours of indoor and outdoor biting abundance by season and hour) than any of the other methods (Table 2). The HLC was a preferred sampling method that is compatible with analyses for sporozoites and species identification. The HLC was the only technique used by informants that directly estimates the epidemiologically relevant indicator: the exposure of humans to biting mosquitoes. Thus, estimates of the biting rate from the HLC are directly used to calculate the entomological inoculation rate (EIR) by multiplying by the sporozoite rate. This technique requires limited training before implementing and is thus compatible with community recruitment for monitoring all human biting mosquitoes.
The potential exposure of collectors to vector borne diseases limits the use of HLC in some countries [21]. The HLC also requires a high level of supervision to maintain quality and sample size is a function of collector attractiveness and collector efficiency, thus impacting sampling reproducibility (e.g., high variances in catch numbers makes it hard to standardize). Another limitation is the logistics of getting the supervisory team and supplies to study sites.
CDC light traps
CDC light traps were the second most frequently used tool for monitoring biting vectors (Table 1). CDC light traps were used to provide data on adult presence/densities both indoors and outdoors. In addition, community householders are easily trained to operate CDC light traps, thus minimizing the need for supervision and the tool is a good measure of tracking anopheline indoor numbers (Table 2). Informants’ opinions varied s about the range of indicators monitored by light traps and the effectiveness of light traps.
Data from CDC-LT are not directly epidemiologically relevant (i.e., the number of mosquitoes collected must be transformed to a biting rate which will vary by species, human biting habit and geographic area and requires simultaneous catch comparisons in any new area to HLC to validate interpretation of catch abundance). When used outdoors, the interpretation of the number sampled is uncertain (i.e., the range that mosquitoes are attracted to the CDC-LT is unknown).
Human and animal baited traps
Human and animal baited traps were used less frequently in studies of biting mosquitoes and were, limited to providing data on biting rates and times (Table 1). Human-baited traps were considered to be effective, while animal baited traps were reported to have variable effectiveness.
Both human and animal baited traps were limited by labour, costs and ease of use. The relationship of the number of sporozoite positive mosquitoes captured when attracted to animal baited traps to the sporozoite rate in mosquitoes attracted to humans was uncertain to informants and thus concerning, as was the relative attractiveness of mosquitoes to animals and humans; thereby making the calculation of the EIR based on these techniques problematic without area specific comparisons to HLC collections.
Resting mosquitoes
Two basic surveillance categories sampled resting mosquitoes: sampling natural resting habitats with aspirators and knockdown spray catches and the construction of sites from which resting mosquitoes are collected (e.g., barrier screens [22], pit traps and window traps) (Tables 3 and 4).
Sampling natural habitats
Searches for resting mosquitoes with oral and battery-powered aspirators or collections following knockdown (i.e. pyrethrum) sprays were the most common tools for sampling indoor resting mosquitoes. None of these tools were considered to be particularly effective (Table 4). These three tools generate data on indoor density and for determining the human blood index and to provide samples for analyses for insecticide resistance frequency. However, these tools are labour intensive (and thus costly) and logistically not easy to implement (Table 2).
Sampling man-made habitats
The second approach is to create or provide suitable resting habitats prior to collection through building of pit traps, provision of resting boxes and clay pots [23] and construction of barrier screens [24]. Pit traps often catch a lot of mosquito specimens (Table 4). However, the technique is labour intensive to set up and to maintain. Resting pots/boxes are relatively immobile and capture few mosquitoes.
Barrier screens differ from both pit traps and resting boxes and pots in that the barrier screen intercepts mosquitoes in transit to likely resting habitats or when transitioning between behaviours (e.g., while seeking oviposition sites, blood or sugar meals or mating swarms). Barrier screens are inexpensive to construct from locally sourced materials, and community members are easily trained to construct and operate barrier screens. Barrier screens are labour intensive to operate and involve some effort to transport and assemble materials in remote sites.
There was not a preferred tool for sampling resting adults. The barrier screen, the most recently developed tool, measured more indicators associated with resting mosquitoes but was used by only 15% of informants. The need for a new more efficient and less labour-intensive tool for sampling resting mosquitoes was universally expressed.
Sampling immatures and gravid adults
Dipping [25] and emergence traps [26] were used to monitor immature populations and gravid traps for sampling ovipositing adults [27] (Tables 5 and 6). Dipping was the most commonly used tool and uses simple and inexpensive equipment but the effectiveness of dipping varies by site (Table 6), being effective in small larval habitats (though some habitats may be inaccessible or cryptic (i.e., hard to find)). On the other hand, samples from larger larval habitats are biased as only the perimeter is sampled and thus the distribution of larvae is not defined. Larval surveys by dipping are labour intensive (with efficiency of collections varying by individual collectors) and dipping procedures are not standardized.
While dipping for larvae can determine species presence, a significant limitation includes the uncertain relationship of larval numbers to epidemiologically relevant indicators as the density or prevalence of dips positive for larvae to larvae numbers by surface area, total larvae numbers in the habitat or the number of biting adults is unknown. A confounding factor for interpretation of results is that the consistency of dipping by species will vary by species larval behaviours and the efficiency of the collector.
Emergence and gravid traps provide data on mosquito species presence and density but were not used by vector control experts as they do not provide much information on other entomological indicators. These sampling tools are not easy to use and require training before use.
Analysis of mosquito specimens
Adult mosquito samples were most often analysed for species identification, malaria parasite species infection and adult mosquito age structure. While all the analysis techniques are potentially compatible with in-country analyses, almost all of the techniques require well trained staff, have complex protocols, have significant costs and associated logistic constraints, as well as requiring specimens in good condition. The interpretation of results from these laboratory techniques are susceptible to variability which can arise from the variation in the technical consistency of the laboratorian performing the test (Table 7).
Mosquito identification
Thirty eight percent of vector control experts commented on vector identification techniques.
Morphological Mosquito identification
Morphological mosquito identification is less expensive than molecular-based identifications, and countries have this capability. However, the training and retention of staff to use complex taxonomic keys are drawbacks. Morphological-based identifications are often only to species complexes even for good condition specimens or a high identification error rate can result (varies by site/person). Hence, morphological identifications almost always require molecular confirmation for members of species complexes. Ethanol storage for ease of molecular identifications, unfortunately, increases the difficulty of morphological identifications.
Molecular mosquito identifications
Polymerase chain reactions (PCR) for mosquito identifications can be done in some countries and PCR based analyses is compatible with multiple storage conditions and has high sample throughput [28]. However, the technique requires a high level of training and the protocol is complex with high costs. A prerequisite for PCR-based analyses is accurate morphological identifications prior to molecular analyses which requires high quality specimens. While PCR reactions can be done in-country, some countries send specimens outside the country for sequencing, (Table 7).
Parasite detection in mosquitoes
Ten vector control experts critiqued the circumsporozoite enzyme-linked immunosorbent assay (CS-ELISA) and PCR to identify malaria parasites in mosquitoes. Both techniques share the need for a high level of training, have high laboratory establishment costs, are labour intensive techniques with complex analyses protocols. However, both techniques are compatible with multiple mosquito storage methods prior to analyses.
CS-ELISA
The CS-ELISA strengths are species specificity, lower laboratory establishment and sample run costs relative to PCR and compatibility with analyses of pools of mosquitoes [29]. Once established, the CS-ELISA was judged to be more robust and reliable than PCR by the key informants, (Table 7). However, the need for PCR analysis for mosquito species identifications is a significant limitation as CS-ELISA analyses does not negate the expense of conducting PCR based assays to identify the vector species in species complexes.
PCR sporozoite detection
PCR can identify sporozoites to species [30] and is compatible with PCRs for mosquito species identifications and insecticide resistance mechanism determination, a significant advantage over the CS-ELISA.
The PCR for sporozoite detection is more costly than the CS-ELISA to process samples and to establish the laboratory. The technique is species but not stage specific, so, like the CS-ELISA, mosquito heads and thoraxes need to be separated from abdomens before analysis. There is also the potential for interference by blood contamination if abdomens are not removed prior to analyses.
Age grading
The current techniques to age grade field collected mosquitoes are dissections for parity and ovarian dilatation counts [31]. A third technique, near infra-red scanning, has not yet been evaluated on wild mosquitoes [32]. Seventeen vector control experts provided input on these three-mosquito age-grading techniques.
Ovarian dilatations
The advantage of ovarian dilatations for age grading mosquitoes is the sensitivity of age determinations based on the number of oviposition events which provides epidemiology relevant data (i.e., the potential to differentiate mosquitoes that have lived long enough to transmit pathogens from younger mosquitoes) (Table 7).. Ovarian dilatation dissections are labour intensive and challenging: significant training/practice is required, and variability in scoring the number of dilatations can be high.
Parity dissection
Parity dissections are easier to perform than ovarian dilatation dissections and parity determination is thus a field friendly technique, enabling a skilled technician to age-grade many mosquitoes by distinguishing parous from non-parous mosquitoes. Weaknesses of parity dissections were the lack of specificity (mosquitoes either have never laid eggs or they have laid eggs), a microscope is required, and live mosquitoes must be dissected (limitations shared with the ovarian dilatation technique). While easier than ovarian dilations, a level of experience and practice is required; the technique is labour intensive and needs a dedicated team.
Near infra-red
The relatively new technique of using near infra-red scanning to age-grade mosquitoes has advantages of being compatible with high throughput, is inexpensive to analyse mosquitoes (after the initial establishment costs) and is compatible with species identification [33]. However, there is significant variance in the age estimations of laboratory specimens of known age. Specimens must be carefully handled and oriented prior to scanning. Establishing the age of a mosquito requires calibration against mosquitoes of known chronological age from each area and time when the technique is used. The calibration curve is labour intensive to establish and its stability in time and space is unknown. Thus far, the reliability and usefulness of near infra-red analyses for determining mosquito population age-structures has not been verified by studies of wild mosquito populations.
Insecticide resistance and quality assurance
Current techniques for measuring insecticide resistance phenotypes in mosquitoes were discussed by 15 vector control experts. Training requirements, protocol complexity, logistics including transporting specimens to laboratories and the need to rear larvae from field collections to adult mosquitoes were key weaknesses for both the WHO tube test and the CDC bottle bioassay.
WHO tube test
The WHO tube test is the historical gold standard and is compatible with both laboratory and field-based analyses [34]. The existence of extensive historic data enables comparisons and tracking of resistance trends. However, WHO tube test analyses are limited to fast acting contact insecticides for whom the resistance thresholds are poorly defined. The stock supply of WHO insecticide-treated papers, particularly papers with different concentrations of an insecticide for intensity assays, varies with papers sometimes arriving with a limited use window before exceeding the expiry dates (Table 7). Furthermore, it is uncertain if the discriminating dose as applied to the papers are appropriate for all species. The construction of the tubes and slides is poor, making manipulation of mosquitoes difficult and the duration of mosquito contact with the insecticide paper is uncertain.
CDC bottle bioassay
The CDC bottle bioassay measures shifts in population insecticide resistance phenotypes and is more adaptable to local conditions than the WHO tube test. The available SOPs lack sufficient detail and would benefit from videos on defining live and dead mosquitoes. Assays using the PBO synergist are challenging in that they require a 1-hour pre-exposure to PBO before exposing mosquitoes to pyrethroids, which damages mosquitoes and thus impacts survival. Errors are easily made in the dilution series and acetone procurement is difficult in some countries.
Bioassay for insecticide concentrations on treated surfaces
The current technique to determine biologically active insecticide concentrations on treated surfaces is the WHO cone test attached to a wall or ITN [35]. Weaknesses identified by 5 informants were inconsistent results by mosquito strains (i.e., lack of a standard mosquito reference strain) which is compounded by lack of a standard mosquito rearing technique (Table 7). The need for colony mosquitoes in assays carries the high operational cost associated with maintenance of an insectary. The cone bioassay technique requires a high degree of training and is labour intensive. Use of WHO cones on uneven mud walls often results in mosquitoes escaping (as maintaining an adequate seal with the wall surface is difficult). As with the WHO tube test, the actual contact time of mosquitoes on the treated surface is unknown.
Next generation entomological surveillance tools
Twenty-three vector control experts provided ideas and attributes for Target Product Profiles (TPP) for seven next generation vector surveillance tools based upon the identified strengths and limitations of current surveillance tools, as summarized above. Overall, it was expressed that the next generation of surveillance field tools and surveillance techniques should require be simple to deploy with minimal training and manpower and have low establishment and operational costs to collect and process specimens. Sampling tools should be applicable for both indoor and outdoor applications and be amenable to a standardised deployment strategy. New tools should sample populations using representative sampling algorithms to measure epidemiologically relevant parameters that can be quickly and easily interpreted by program managers for decision-making (Supplemental files 1-7).
Seven vector sampling methods were identified as priority targets for improving vector surveillance in two broad categories: vector sampling (i.e., HLC alternatives, automated adult traps, quantitative larval sampling, vector age-grading and identifying malaria parasites) and insecticide monitoring (i.e., quantitative non-bioassay methods for surface active compounds and insecticide resistance phenotyping).
Vector sampling
HLC alternatives: Alternatives to the HLC are needed to sample human host seeking adult vectors that can be calibrated to historical HLC data without placing trap operators at risk and are safe for use in or near human dwellings. Any new method must maintain the functionality of the HLC (see
Table 1) and monitor without bias the presence of all malaria vector species in an area, ir-regardless of their density or behaviours (time of biting, anthropophagy, resting habits, etc) (see Additional file 2). Alternatives to the HLC could include automated sensitive and specific traps to attract, collect, count and identify mosquitoes to species to improve data collection and specimen quality. Like any HLC alternative, automated trap catches should be compatible with calibration to current trapping techniques but capable of collecting all malaria vector species both indoors and outdoors while recording the number of adult female vectors collected per unit time (see Additional file 2).
Quantitative larval sampling: A quantitative larval sampling tool to estimate adult malaria vector populations is needed to determine mosquito larval vector composition and densities. Larval sampling methods must fully characterize larval habitats to identify productive sites and to predict habitat creation following rainfall. Autonomous sampling would be beneficial and could be based on drone or other technologies that could sample larvae and identify species with multi-imaging capacities to find water bodies based on species specific characteristics across ecosystems (see Additional file 2).
Vector age-grading: A technique to determine mosquito age in 1-day increments is required to calculate median mosquito age with a 95% confidence interval (see Additional file 2).
Identifying malaria parasites: A novel rapid diagnostic test to identify all Plasmodium species in vector species that can be calibrated to current Plasmodium detection and identification techniques is required to reliably identify sporozoites in mosquitoes at a sensitivity of <1000 sporozoites with a 95% precision in duplicate readings of mosquito cohorts (see Additional file 2).
Insecticide monitoring
Quantitative non-bioassay methods for surface active compounds: A technique to quantify surface-accessible active compounds as a cost-effective alternative to WHO cone bioassays or chemical extraction and analysis techniques is needed to circumvent the cost of maintaining mosquito colonies. The technique needs to be field applicable, rapid and accurate in detecting all active ingredients and chemical stereoisomer ratios and content on all surfaces and sufficiently robust to operate in the presence of common wall contaminates (e.g. dirt or smoke) with operational costs below current costs to determine surface insecticide concentrations (see Additional file 2).
Insecticide resistance phenotyping: A cost-effective method to define insecticide resistance phenotypes (frequency and intensity) as well as molecular or biochemical resistance markers in adult mosquitoes that reduces human and budget resources over current WHO tube and CDC bottle assays is needed (see Additional file 2).
Research required for effective vector surveillances
Vector control experts noted that little is known about some mosquito behaviours that are being targeted by novel control measures (e.g., outdoor resting and blood feeding, sugar feeding, mating, etc). Thus, there is a need to develop next generation vector surveillance tools to monitor these behaviours. The resulting data will then better inform interventions targeting these behaviours. The following priority areas were identified as critical for improving vector surveillance.
Vector behaviour data to inform surveillance deployment
More information is needed on the movement of adult vectors from emergence to blood feeding and oviposition sites to inform representative sampling over a geographic area and not biased to areas with high densities. Such representative sampling can then be used to optimise intervention deployment more effectively. Improved surveillance would benefit from understanding 1) where and when mosquitoes are exposed to control tools, 2) the distribution of daytime biting mosquitoes; 3) host attractancy/biting rates of mosquitoes to different blood meal hosts, 4) novel attractants to replace carbon dioxide, human odours and other blood seeking lures to improve trap performance, 5) receptivity (its definition and measurement), 6) outdoor resting site characteristics by species to improve interventions targeting exophilic adults, 7) larval habitat characteristics by species to improve larval source management. Detailed understanding of the biology of vectors will improve both vector surveillance and malaria control programs.
Data management
Faster data entry systems with a uniform formats for data capture and recording across locations will enable vector surveillance data to guide program decisions in real time to improve vector surveillance.
Algorithms for representative sampling
Historically the need to maximise mosquito collection samples sizes (e.g., determination of sporozoite rates, age structure and peak biting time) encouraged sampling bias to collect maximum numbers of mosquitoes with minimal effort. Algorithms for representative sampling across geographic areas for adults and larvae as well as defining /stratifying receptivity are essential.
Representative sampling methods for resistance testing of mosquito populations are needed. More spatially explicit representative sampling for resistance phenotypes are required along with guidance on interpreting results for resistance management (i.e., when to increase insecticide concentrations or to switch insecticide classes in the face of changing resistance profiles). Algorithms are needed to correlate vector insecticide resistance bioassay and genetic data to intervention impact and to correlate hut data to intervention efficacy at scale.