Burkina Faso was one of the first countries to conduct a national baseline LF prevalence survey at the health district level in order to estimate the burden of the disease. It resulted in an estimated 29.2% national prevalence with the highest district-level prevalence being 74% (Programme National d’Elimination de la Filariose Lymphatique Burkina Faso, 2012). Shortly following these results, Burkina Faso launched their national programme for the elimination of LF and began, in 2001, implementing MDA within four health districts. Mass Drug Administration (MDA) then scaled up rapidly across the country, such that by 2006, therapeutic coverage levels were greater than 65% in all endemic districts (The Global Atlas of Helminth Infections, 2020). As LF is a vector-borne disease, entomological assessment of parasites circulating within vector populations has been a core component of disease surveillance. In the move towards elimination, direct estimation of the presence of microfilariae in vectors is the best method to document the interruption of W. bancrofti transmission. However, demonstrating the end of entomological transmission requires the search for microfilariae in a large number of vectors (Dorkenoo et al., 2018). Until now, it has been difficult to obtain sufficient numbers of vectors for entomological surveillance because the use of Human Landing Catch (HLC) for collection of lymphatic filariasis vectors is becoming less acceptable on practical and ethical grounds, and not least because it requires well-trained persons and remains the most expensive method to deploy. This situation has resulted in apathy towards the use of xenomonitoring for LF in some contexts. To avoid the ethical issues and the reliance on well-trained persons    from    central    laboratories,   methods   which facilitate vector collection by local communities and are simple, reliable and easy to implement must be developed. For example, the Double Net trap, comprising of a person sleeping inside a simple arrangement of nets was shown to be effective for collecting the vectors of W. bancrofti (Govella et al., 2011). In the context of onchocerciasis vector surveillance, the Esperanza Window Trap (EWT) has been used successfully for sampling Simulium species in Mexico (Rodríguez-Pérez et al., 2013a)and Burkina Faso (Toé et al., 2014).

Studies to evaluate the performance of these various trapping methods for collecting LF vectors have typically been conducted by experienced and well-trained entomologists. However, in the context of large-scale entomological surveillance of LF, use of qualified technical personnel poses an operational challenge due to the time and costs involved in training and maintaining entomological teams. There are also financial and logistical limits to the number of teams that can be deployed for activities in the field, reducing the number of localities that can be covered during surveillance (Sikaala et al., 2014). Thus, developing a similar programme of vector collection involving members of each local community could support enhanced surveillance, building on similar community-based initiatives, such as Community-Directed Treatment with Ivermectin (CDTI) under the African Programme for Onchocerciasis Control (APOC). Such a community-based direct entomological surveillance programme would promote an inclusive and economically viable assessment of the entomological transmission of this disease in space and time, which could not be achieved by the exclusive involvement of teams of qualified technicians from research centers. This approach has been successfully tested in Mexico, where the performance of the EWT trap was used for the collection of blackflies by members of the community of Chiapas for 60 days(Rodríguez-Pérez et al., 2013b). This study showed that, even though the number of Simulium vectors collected during this period by the community members was lower than that obtained by a team of qualified technicians, it was statistically sufficient to validate the evaluation criteria given by the WHO to estimate the interruption of transmission of the disease in the community (Rodríguez-Pérez et al., 2013b). However, there is the need to test whether this strategy could also be applied in the African context. The objective of this study was to assess the deployment of traps for collection of Anopheles vectors of W. bancrofti by a community-based approach in the Southwestern region, of Burkina Faso West Africa. Considering the endemicity of LF in the area, the prevalence of the parasite also was determined in the Anopheles vector, alongside the species composition of An. gambiae sensu lato.

Study sites

The study was carried out in three villages, Bapla (10°53'25.83"N; 3°15'46.74"W), Ouessa (11°02'19.59"N; 2°47'7.57"W) and Koudjo (9°53'50.41"N; 2°58'43.43"W), located in Southwestern region of Burkina Faso (Figure 1). This region is endemic for LF and the three villages were randomly selected. The rainy season in the region extends from June to October and the dry season from November to May. Average annual rainfall ranges between 1000 and 1300 mm with extreme temperatures of 12 and 38°C. The principal river is the Mouhoun, which originates in the Haut Bassins region and drains into Ghana. This part of the country is characterized mainly by a vector complex including An. gambiae s.l. (60%), An funestus (37%) and An nili (<5%). An. nili is more concentrated in the highly wooded and humid savannas of the Southwestern region, where it accounted for 30% of the Anopheles collected using Human Landing Catch in 2003 in the zone of Gaoua. Previous entomological studies undertaken in this area (Namountougou et al., 2013)indicate that Anopheles arabiensis is rare and found at low frequencies, often less than 10%. This region experiences permanent transmission of malaria, highest during the rainy season and at the beginning of the dry season, and historically it has been an area of intense transmission of lymphatic filariasis; and since 2001, mass drug administration has been instituted by National Programme of Lymphatic Filariasis Elimination (Kima, 2012).

Community engagement process and volunteer training

Community engagement was initiated following meetings between a group of staff from Institut de Recherche en Sciences de la Santé (IRSS), including entomologists, community engagement specialists, and institutional management and the local authorities of each study village. During this meeting, IRSS staff presented the protocol and the study objectives, highlighting the strategies used to survey vector of diseases, focusing on malaria and lymphatic filariasis, and the importance of the role that villagers could played in supporting its success. In each village, the local authorities then selected eight people to be involved in vector collection, on the basis of the criteria suggested by IRSS staff: i) males, ii) 20 years old, iii) resident in the village and in good apparent health, and iv) able to read and write. After selection of the volunteers, the IRSS staff provided them with a one-week training course. This consisted first of showing the collectors how to collect mosquitoes using the HLC technique, the CDC light trap technique, the Window Exit trap and the Double Net trap, tools used to collect mosquitoes in this current study, according to established protocols for each method (Boakye et al., 2019; Costantini et al., 1998; Epopa et al., 2019). After initial demonstrations of the different techniques, the volunteers were supervised in using all these methods to collect mosquitoes. Then, the team taught the volunteers how to identify the genera Anopheles, Culex and Aedes using morphological criteria and how to correctly sort, label and store them in tubes containing silica gel desiccant to preserve them before transport to the laboratory. Finally,  one  of  the   eight   volunteers   was   chosen   as   a  team supervisor and received additional training in packing, storing and shipping the samples to the laboratories at IRSS. Volunteers were provided chemoprophylaxis for the duration of the study. The sum of 4000 FCFA ($4) per day of mosquito collection was offered to the volunteers to compensate for their loss of time and effort.

Mosquito sampling by local collectors

Mosquitoes were sampled by HLC, CDC light trap, Double Net Trap and Exit Trap methods of collection for six months from June to November 2016 in each village. Collections were carried out during two consecutive nights each month (the same nights in all villages) in four houses randomly selected for each collection method. For HLC two collector stations (one indoor and one outdoor) at each house collected all skin landing (pre-biting) mosquitoes from 20:00 to 06:00 the next morning. Collectors were regularly rotated to reduce collector-mediated bias in the results and supervision was provided to ensure collectors stayed awake thus reducing any potential for biting (Figure 2A). The mosquitoes collected in each house were stored by collection origin (indoor and outdoor) and in hourly tranches. In the morning, the mosquito collectors stunned the mosquitoes by exposition to the sun and pooled them by genus into groups of 10, which were placed together in a tube with silicagel. For CDC light trap method, two  traps  were  installed (one indoor and one outdoor) (Figure 2B). Indoor CDC light traps were suspended approximately at 1.5 to 2 m from an occupied untreated bed net generally at the foot end of the bed 1.5 m from the floor. When nets were torn or absent, new nets were provided. The outdoor CDC light trap was placed under the roof at 0.2-0.3 m from the outer wall, at approximately 1.5 m from the ground. All CDC light traps operated from 20:00 to 06:00 h. Exit traps (ET) were fitted to windows outdoor (placed in the path of flying mosquitoes) to catch mosquitoes exiting the human dwelling (Figure 2C). These traps allowed collection of mosquitoes according to their degree of normal or induced exophily. Traps were installed before sunset (5:00 PM) and emptied after sunrise (8:00 AM). For Double Net traps, two traps were installed (one indoor and one outdoor), using a person on a camp bed as "bait" (Figure 2D). The bottom of the outer net was raised about 15 cm from the ground to allow entry of mosquitoes. The collections of mosquitoes trapped during the night within the two nets were made early in the morning by a collector using a flashlight and mouth aspirator.

Shipping of the collected mosquitoes to the IRSS laboratory

At the end of the two nights of collection, each supervisor packed mosquitoes collected in a cool box and shipped them to IRSS laboratory located in Bobo-Dioulasso by local bus transporters.

Morphological and molecular identification of collected Anopheles

In the IRSS laboratory, morphological identification of collected mosquitoes was made according to the keys as previously described (Gillies and de Meillon, 1968, Gillies and Coetzee, 1987). After this morphological identification, only sibling species of the An. gambiae complex were identified by polymerase chain reaction (PCR) as previously described (Santolamazza et al., 2008).

W. bancrofti detection in Anopheles species

Detection of W. bancrofti was performed on whole individuals of An. gambiaes., An. funestus and An. nili. Parasite DNA was extracted from whole mosquito by CTAB 2%. Positive samples were determined using the standard polymerase chain reaction (PCR)-based diagnostic tool previously described (Ramzy et al., 1997).

Data analysis

Data were entered and cross-checked in Windows Excel 2010. Statistical analyses were performed using R.4.0.3 with a significance level of 5%. Two main variables were analysed: density of mosquitoes per night and human biting rate. The biting rate from HLC, CDC light trap and Double Net trap was expressed as the number of Anopheles per sampling method per night. The result was obtained by the number of unfed Anopheles female collected at each sampling point divided by the total number of sampling days and the average number of volunteers. ANOVAs were conducted to compare the mosquito density and the biting rate between the different collection methods.

Ethical considerations

The protocol of the study received approval from the institutional ethics committee of Institut de Recherche en Sciences de la Santé (N/Réf. A16-2016/CEIRES). Engagement visits were carried out in study sites to present the project and to request the participation of villagers. During these visits, the objectives of the protocol and expected results were explained and discussed with the villagers, as well as the implications for the households willing to take part in this study. A written consent form was signed or marked with fingerprint by the head of the households before any activity could take place in his compound. At the end of the activity, local communities were informed of the results of the study in town-hall meetings. In addition, education has been provided to the local communities regarding mosquitoes biting prevention.

Mosquito abundance in three study sites

A total of 1,706 mosquitoes was collected during the study period in Bapla with 1,199 (70.3%), 278 (16.3%), 165 (9.7%) and 64 (3.7%) by HLC, CDC light trap, Double Net trap and Exit trap, respectively (Table 1). The majority of the mosquitoes collected were Anophelinae (74.2%). Among Anopheles species, proportions of 53, 14, 2.8, 2.2, 1, 1.1% were found for An. gambiae s.l., An. funestus, An. coustani, An. nili, An. pharaonsis, and other Anopheles species, respectively. The remaining Culicidae  were Culex (23%), Aedes (0.6%) and Mansonia species (2.2%).

In Ouessa, a total of 1,906 mosquitoes were collected. Broken down across the trapping methods, a number of 876 (46%), 301 (15.8%), 713 (37.4%) and 16 (0.8%) by HLC, CDC light trap, Double Net trap and Exit trap, respectively (Table 1). The Anophelinae subfamily represented a proportion of 77.5% of the samples where A. gambiae s.l. (42.7%), An. funestus (8.7%), An. nili (5.5%), An. coustani (13%), An. pharaonsis (0.3%) and other Anopheles species (7.3%) were found. Other species collected were Culex spp (10.4%), Aedes spp (4.9%) and Mansonia spp (7.2%).

In Koudjo, a total of 625 mosquitoes were collected: 426 (68%) by HLC, 58 (9.28%) by CDC light trap, 70 (11.2%) by Double Net trap and 71 (11.36%) from Exit trap (Table 1). The majority of mosquitoes belonged to Anopheles genera: An. gambiae s.l. (88.2%), followed by An. funestus (3.4%), An. nili (3.7%), An. coustani (0.8%) and other Anopheles (1.8%). Other Culicine mosquitoes were collected in very low proportions, specifically Culex (1.1%), Aedes (1%) and Mansonia (0.2%) species.

The density of mosquitoes collected per night is represented in Figure 3. The data shows a significant difference according to the sampling method; in Bapla village, the HLC caught more than the three other traps (F = 4.265, P< 0.05, Figure 3A) and no significant difference was found between CDC, Double Net trap and Exit trap (F= 0.9170, P = 0.42). In Ouessa village, significantly more mosquitoes were collected with HLC than Exit trap (t = 2.497, P< 0.05) but no significant difference between CDC and Double Net trap was found (F = 0.4646, P = 0.63). Furthermore, mosquito density did not differ significantly between CDC, Double Net trap and exit trap (F = 2.422, P = 0.13, Figure 3B). In Koudjo village, the density of mosquito did not differ significantly between the four sampling methods (F = 2.217, P = 0.13, Figure 3C).

Indoor and outdoor biting rate of Anopheles mosquito on human estimated by HLC, CDC light traps and net trap

In Bapla village, the numbers of mosquitoes per night vary significantly according to the sampling method. In this village, HLC caught the largest number of three vectors mosquitoes indoors (F = 4.536, P< 0.05) and outdoor (F = 9.345, P< 0.05) (Figure 4A to C). Considering the HLC collection method, the mean number of mosquitoes/person/night (m/p/n) did not vary significantly between indoor and outdoor for An. gambiae (t = 0.308, P = 0.77), An. funestus (t = 0.728, P = 0.51) and An. nili (t = 0.754, P = 0.49). The mean number of mosquitoes/trap/night (m/t/n) did not differ significantly between CDC light and double Net traps both for indoor (t = 1.534, P = 0.16) and outdoor collections (t = 1.603, P = 0.12)  of  An.  gambiae   and   An.   funestus  respectively (Figure 4A, B); no An. nili were collected with the Double Net trap indoor (Figure 4C).

In Ouessa village, the sampling method affected also the number of mosquitoes per night. The HLC again collected the greatest number of three vectors indoor (F = 9.952, P< 0.05) and outdoor (F = 55.94, P< 0.05) (Figure 4D, E, F). However, the mean number of m/p/n was relatively similar between indoor and outdoor collection for An. gambiae (t = 1.582, P = 0.19), An. funestus (t = 0.736, P = 0.50) and An. nili (t = 0.323, P = 0.76). Concerning the CDC and double Net trap, no significant difference was found in the number of m/t/n between these two collection methods indoor and outdoor for An. gambiae  (F =  0405, P = 0.75) and for An. funestus (F = 2.018, P = 0.19), (Figure 4DE). An. nili was only collected outdoor with Double Net trap in this village.

In Kodjo village, the sampling method and position was not significant impacted by the biting patterns for An. funestus (F = 0.835, P = 0.54) and An. nili (F = 2.322, P = 0.08). However, for An. gambiae mosquito, the mean number of mosquitoes per night indoors was higher with HLC compared to CDC and double net traps both indoor (F = 5.119, P< 0.05) and outdoor (F = 10.77, P< 0.05). The mean number of m/t/n did not differ significantly between CDC and double Net trap indoor (t = 0. 266, P = 0.82) and outdoor (t = 1. 576, P = 0.17).

Sibling species of A. gambiae s.l and W. bancrofti detection

In Bapla village, An. gambiae species was predominant (80.21%) followed by An. coluzzii (15.41%) and An. arabiensis (4.38%). In the same range, An. gambiae remained the predominant species in Ouessa (67%), followed by An. Coluzzii (24.14%) and An. arabiensis (8.04%). Analysis of the species composition according to the collection methods shown that HLC, CDC light trap, double Net trap and Exit trap (Figure 5) were able to collect all species of An. gambiae complex present in Bapla and Ouessa villages.

A total of 1,018 Anopheles were analysed for W.bancrofti infection from Bapla (928 samples) and Ouessa (90 samples). The overall estimated rate of infection of all collected mosquitoes with W. bancrofti parasites was 0.4% (Table 2). The infection rate was significantly higher in Ouessa (2.2%) compared to Bapla (0.2%) (P< 0.05). Positive samples were observed only in the An. gambiae species. In term of collection position, the infection rate was significantly greater outdoor (0.3%) compared to indoor collection from the two villages (0.1%) (P< 0.05).

Although the most standard and widely-used method of collecting entomological data for LF is the HLC, the ethical concerns and cost efficacy ratio (Sikaala et al., 2013)of this tool hassled the scientific community to develop new sampling methods, such as the Ifakara tent trap (Govella et al., 2009). Some of them aimed at improving  the  cost  which  being  less  than that  of  HLC (Sikaala et al., 2014).

However, there are still debates about the most efficient method to be used for mosquito sampling and, depending on the country and its regulations, different sampling methods are used. For example, several countries in the East African region have already replaced HLC by more recent techniques such as versions of the Ifakara tent traps (Govella et al., 2009).

Communities may have a valuable role to play in supporting efforts to improve the geographical range over which LF surveillance takes places, and the duration of vector surveillance periods. This will be important in moves towards elimination, where enhanced entomological surveillance is required to detect ever lower infection rates. The reliance on trained staff from central laboratories in vector collection has mitigated against the use of entomological methods in disease monitoring and evaluation strategies. However, in the context of elimination of malaria and vector borne NTDs where large  numbers of vectors are required for analysis to determine if elimination indicators have been achieved, it is crucial to engage communities in this process. That means that different vector collection tools need to be tested in communities and by communities themselves to compare their efficiency and determine the community acceptance and use of the tools as a routine surveillance method. Implementation of this strategy in Brazil  showed that the number of Simulium vectors collected by community members was lower than that obtained by a team of qualified technicians (Rodríguez-Pérez et al., 2013b). However, this result can be explain by aninsufficiency in the training of local community and need to be tested in Africa.

The   present   study   tested   four   sampling  methods including HLC, CDC light trap, Exit trap and a local variation of a double Net trap. Our results showed that HLC remained the most efficient in terms of number of mosquitoes collected, whatever the village and the collection places (indoor vs outdoor). It was followed by the Double Net trap which was sometimes similar to CDC traps. Regarding the biting rate estimation, HLC showed high level of An. gambiae s.l. biting rate especially indoors but less efficient than other methods in areas where mosquito’s densities were reduced like Ouessa and Koudjo. The higher density of mosquitoes collected by HLC method can be explained by the direct access of mosquitoes to the volunteer with this method. In Ghana, a study showed that Double Net traps observed to be performing relatively better than HLC (Boakye et al., 2019). Several factors can explain this discrepancy including entomological and ecological features. In addition, our study was carried out in rainy season only and this may be a weakness of our study.

An important observation is that in area with low and medium densities of mosquitoes like Ouessa and Koudjo, the biting rates were better highlighted by the double Net Trap than did the HLC compared to the area with high mosquito  density  such   as  Bapla. This trap  provides  a relatively good performance similar or superior to HLC in collecting other culicidae than anophelinae. The exit traps which were placed in the tops of windows do not allow to evaluate the density of mosquitoes collected potentially in contact of humans indoors or outdoors (biting rates) but remain also a collection tool relatively less efficient but operational.

Interestingly, in term of species composition, our study showed that every collection method used in this study were efficient to collect both Malaria and LF vectors. The positive samples of filariasis infection were only found in An. gambiae. The absence of infection in the remaining species may be attributed to the small number of samples tested. Mass drug administration (MDA) occurred within communities in these areas.

Furthermore, in terms of communities’ acceptance, we did not perform a direct questionnaire to assess these aspects, but a feedback mission remains to be organized including at posteriori social questionnaire on the acceptance of such community surveillance of LF vectors. This is a limit of our study. Without statistics data, the collectors would prefer to use double Net trap followed by exit traps. The main reasons were that these techniques did not  need  all  night  movements  to  repair malfunctioning tools such as CDC traps or to remain awake to collect landing mosquitoes by HLC. The double Net traps provided them some protection and they need to collect mosquitoes only just the next early morning. They also found that the exit trap is simple to be used and did not need other technical skills. As we mentioned in the methodology, this point remains to be documented by posteriori questionnaire and focus group discussions need to be performed with the local collectors and some members of the communities on the acceptability and the perception to conduct such surveys routinely.

In conclusion, this study revealed that it is possible to involve local communities to vector surveillance and to obtain large sample sizes necessary to make validation of elimination thresholds and that community collectors preferred to use collection systems that did not involve many inconveniences at night such Double Net trap or exit traps. However, we will investigate on communities’ members perception of acceptability of these tools through systematic questionnaires.

The authors have not declared any conflict of interests.

The authors appreciate the laboratory staff and the field teams who worked so hard to gather these data, and the community of the villages of Bapla, Ouessa and Koudjo who have been very helpful in accepting and consenting to mosquito collection. This work received financial support from the Coalition for Operational Research on Neglected Tropical Diseases (COR-NTD).