In spite of important and diverse efforts towards control, malaria remains a challenge for the global community and for sub-Saharan countries in particular. The efficacy of current control strategies has stagnated in recent years and, more worryingly, for the first time in a decade, malaria incidence is increasing [1–3]. A number of factors may contribute to this including inadequate financing [2, 4], gaps in control management [5], parasite resistance to drugs [6, 7], and vector resistance to the insecticides used on insecticide-treated bed nets (ITNs) and for indoor residual spraying (IRS) [8–10]. The development of new tools or approaches and the improvement of integrated control strategies are some of the proposed solutions to this resurgence. Novel, genetic control approaches using genetically-modified malaria vectors to achieve population suppression or trait replacement [11, 12] are considered some of the most promising new approaches in an integrated malaria control strategy [13, 14].

West Africa, where high malaria transmission is expected to persist even under optimistic malaria control funding scenarios, might benefit substantially from the development of these novel technologies [15]. Burkina Faso, is one such West African country where the impact of current vector control tools promoted by the Roll-Back Malaria (RBM) initiative led by the World Health Organization (WHO), such as ITNs, IRS and improved malaria diagnostics and treatments, is not expected to bring the country close to elimination [1, 15, 16]. In recent years (2010, 2013 and 2016), wide-scale distribution of long-lasting insecticidal nets (LLINs) has been intensified resulting in coverage ranging from 90 to 97% by 2016 [17]. Although improved diagnosis and reporting structures contribute somewhat to this, the almost doubling of reported malaria cases between 2010 and 2016, rising from 5.7 to 9.8 M [17, 18], suggests that new vector control tools are needed.

One of the first steps towards implementation of any genetic vector control tool consists of gathering key baseline field information: identifying the different vector species present at field sites and evaluating their relative contribution to seasonal patterns of malaria transmission. This understanding is crucial for the planning of mosquito release strategies with optimal health impacts and for demonstrating their effect. The vector species composition patterns identified can match three distinct scenarios: in the first, one mosquito species is strongly dominant and is consequently the target for focussed intervention [19]; in the second case, a dominant species co-occurs consistently with one or more secondary vector species; the third case is that of a less-dominated community in which two or more vectors are more equal in abundance. The second and third scenarios may require more complex release strategies that could include the simultaneous or sequential release of genetically-modified mosquitoes of different taxa.

In western Burkina Faso, members of the Anopheles gambiae complex are the dominant malaria vectors [20, 21]. In these Sudano-Sahelian areas, populations of Anopheles coluzzii, An. gambiae and Anopheles arabiensis occur in varying proportions between sites, with large seasonal changes in abundance and a great reduction in numbers, or possibly even complete disappearance, recorded during the dry season [22, 23]. The specifics of an intervention using genetic control tools aiming for vector replacement or suppression [24] would thus depend on the population dynamics of each species and on associated patterns of disease transmission.

The level of malaria transmission can be estimated from many parameters which can be entomological [25], parasitological [26, 27], clinical [27, 28] or immunological [29, 30]. The most common entomological method used is the estimation of the entomological inoculation rate (EIR) by combining an estimated number of mosquito bites per human per day (HBR) [31], which itself is calculated from human landing catch (HLC) data, with information on the prevalence of Plasmodium falciparum infections in the salivary glands of mosquitoes, or sporozoite infection rates (SIR). Thus, this estimate of EIR indicates a theoretical maximum number of people potentially infected with P. falciparum during a given period (day, month or year) in a given place. The local and seasonal EIR estimates can then be used to inform and focus malaria control strategies.

Baseline entomological collections necessary for informing future strategies and for assessing their potential in reducing malaria transmission were established in three villages typical of Western Burkina Faso. The vectors responsible for malaria transmission were identified and their seasonal abundance was determined using both larval site surveys and pyrethroid spray catches (PSC) indoors. Malaria transmission dynamics were globally estimated by calculating the EIR from HLC and SIR data without digging into specific species contribution to this. Finally, local rainfall data were measured using a permanent weather station at each site. The results highlight the challenges resulting from very high vector densities characterizing West African Sudano-Sahelian regions during the rainy season.

For over 3 consecutive years (July 2012–November 2015), monthly mosquito field collections to track vector composition, abundance and related malaria transmission, took place in three villages in Burkina Faso’s Sudano-Sahelian region near Bobo-Dioulasso. Three collection methods were used each time: larval site surveys, indoor pyrethroid spray catches (PSC) and human landing catches (HLC).

The malaria vector dynamics in these three study sites contrast in several ways: Species composition, abundance, larval site occupancy and in SIR and EIR. Although the sites are close to one another within the same province, Houet, there are contrasts in the local ecological conditions that may contribute to this. In Bana and Souroukoudingan, the majority of larval sites were rain puddles and tyre tracks which are ephemeral (short-term) larval sites and highly rain-dependent, though Bana has some longer-term larval habitat availability in a semi-permanent river which persists for longer into the dry season and a quasi-permanent overflow pond at the village pump. In contrast, in Pala the most common larval sites were small water pools located along the beds of the two small rivers that cross the village. These pits are also quasi-permanent and remain wet year-round. Larval site occupancy was higher in Pala than in the two other villages, possibly due to the availability of these small, enduring and numerous potential larval sites.

The water relations in the villages may also influence the species composition of the vector community. The malaria vectors identified in this study were mostly within the An. gambiae complex (An. coluzzii, An. gambiae, and An. arabiensis), and other vector species known in Burkina Faso, Anopheles funestus [21, 35, 36] and Anopheles nili [37], were found occasionally. This is consistent with expectations of this ecological zone as the type of larval sites found in the Sahelian climate do not favour the development of An. funestus or An. nili, which are usually found where there is running water [20, 38]. There are ecological differences in the larval habitat preferences within the An. gambiae complex too. Anopheles coluzzii larvae are often associated with larger, more permanent breeding sites while An. gambiae larvae are associated with small and short-term breeding sites [39, 40]. In Bana and Souroukoudingan, where An. coluzzii dominated the catch, the majority of breeding sites were short-term, though generally of large size (3–10 m²). In Pala, where An. gambiae dominated the catch, they were mainly quasi-permanent but usually of small size (0.02–0.5 m²). This suggests that the size rather than the level of permanency may be more influential on the mosquitoes’ development at larval stage and thus on species composition. In fact, the size of larval habitat is known to have an important influence on the response of Anopheles larvae to predation [41]. The contrast in predation response between these species may thus influence larvae colonization of larval habitats [40, 41]. Some other factors such as water turbidity, colour or sunniness could have also contributed up stream to this colonization. Unfortunately these factors were not described in this study.

Mosquito abundance patterns identified in HLC in the three villages agree with those observed by PSC collections: An. gambiae s.l. constitutes the principal suite of malaria vectors in the three villages and there is a high correlation between the techniques for estimates of vector densities through the year. Both techniques provide a good index of relative abundance and species composition in a given area and the differences observed between villages, such as the very high wet-season populations in Bana and the residual dry-season populations in Pala were supported by both techniques. Sahelian anopheline populations are strongly seasonal, peaking in the rainy season [22, 23] and all of the study sites strongly conform to this pattern. The presence of a dry-season population in Pala reflects and responds to the presence of enduring larval habitat in this village.

A vector’s biting rate is one of the most important parameters to influence malaria transmission in a given locality [42–44] and the most direct way to estimate this parameter is through human landing catches (HLC). The use of HLC allows for exploration of another aspect of mosquito behaviour that PSC does not give: endophagy versus exophagy. Here, no difference in landing rate was observed between indoor and outdoor collections and suggests that in this region, malaria vectors have an opportunistic biting behaviour. Endophagy is usually the expected dominant behaviour in an An. gambiae s.l. mosquito population and the degree of exophagy observed here has been recorded previously in South Africa [45, 46]. This may result from a modification in host-seeking behaviour induced by years of vector control using ITNs [47, 48]. Mosquito species of the An. gambiae complex are known to be heterogenous in their biting behaviour. Unfortunately, species composition of the HLC collections are not available in this study. This could have given a more precise information about the contribution of each species in the observed general biting behaviour of the mosquitoes caught. However, at least in a context of high dominance of one species, as observed in the villages of Bana and Pala, it could be considered that species composition of PSC collection should not be too much different from the HLC ones.

The landing time data over the 4 years of this study does not indicate a shift in host seeking towards dusk or dawn when people are more mobile and less likely to be protected by their bed nets. Each village has a locally consistent pattern and, in each, most landings were observed between 22:00 and 05:00. Extending the duration of the HLC an hour in either direction would provide the means to detect possible behavioural change towards day biting. In the absence of evidence of a local behavioural shift, the night landing patterns observed here confirms the crucial importance of ITNs as a malaria vector control tool.

An early-biting pattern could potentially explain a high sporozoite infection rate (SIR) and related EIR [47, 48], but this was not seen to be the case at the sites studied during this survey. Another factor which may contribute to residual EIR in the dry season around Pala is a change in host behaviour; with increasing urbanization there is a shift in human activity that leads to increased host availability into the dark hours [49].

The estimates of SIR and related EIR in this study emphasize the magnitude of the remaining malaria burden, despite good local implementation of the national vector control strategy. Although ELISA-CSP is known to overestimate SIR by a factor of about 1.1–1.9 [50], there is discussion surrounding the accuracy and level of inference to be drawn from these traditional calculations of EIR [51, 52], especially with the low sample size used in this study, these values remain surprisingly high in this local context [53]. This region has one of the highest rates of ITN coverage in Burkina Faso with a mean coverage of about 92% and a usage rate over 72% [17, 18].

Three successive ITN (or LLIN) mass distribution campaigns occurred in Burkina Faso in 2010, 2013 and 2016 with excellent results in term of household coverage [18]. The expectation of a significant reduction in EIR via action on its components: landing rate or sporozoite infection rate, or both, has not been observed in the years of this study. This agrees with national patterns of malaria incidence and morbidity which have continuously increased during the same period, rising from about 5.7 M malaria cases in 2010 to about 9.8 M cases in 2016 [17, 18]. No significant increase in EIR was observed during this study which may suggest at least a stabilization of malaria transmission in this region and these findings may indicate a qualified success (though insufficient) of the currently implemented malaria control tools in this region.

Potential reduction in efficacy of current malaria vector control tools is not specific to the areas studied during this survey; an increasing number of malaria endemic countries of diverse mosquito ecology retain high malaria transmission parameters despite a good implementation of ITNs and/or IRS. This is the case in an Ethiopian region where residual transmission is observed in spite of extensive coverage by IRS and LLINs [54]. In a forest village of Cameroon (Central Africa), Ndo et al. [55] observed recently a SIR of 3.87% in An. gambiae mosquitoes. In Mali, the African IRS project has described long-term non-effects of IRS and LLINs in a number of areas [56].

Over these last two decades, LLINs and IRS have had an exceptional impact in reducing malaria and other vector-borne disease transmission worldwide [3]. Even now, their contribution to the reduction of malaria incidence and morbidity represents the largest proportion of global impact in malaria control [3]. Unfortunately, a reduction of their efficacy is increasingly observed at different sites. Insecticide resistance in mosquitoes [55, 57] and biting behaviour changes [47, 48] are thought to be the main contributors. All these results sustain the thesis of Shaukat et al. [25] which suggests that current vector control methods cannot ultimately eradicate malaria. So, with an aim to malaria eradication in the Sudano-Sahelian region, it is increasingly urgent to research and develop novel vector control tools particularly designed to suppress its very large and seasonal malaria vector populations. In that context, novel genetic control approaches based on male mosquito release programmes may prove a crucial additional tool in the toolbox of future integrated malaria control programmes.

This study has allowed a better understanding of seasonal dynamics of malaria vectors in the villages studied and the main factors that influence these. The vectors found in these three villages belong mostly to the An. gambiae complex (An. coluzzii, An. gambiae, An. arabiensis), but with variable relative proportion from one locality to another. Local geographic and hydrographic differences could explain, at least partially, the differences in vector abundance from one locality to another. As anticipated, cyclic variation in observed vector abundance followed that of the wet and dry seasons. Nevertheless, while in the village of Bana and Souroukoudingan, the vector abundance and associated EIR were highly correlated with seasons, in the village of Pala, the particular local hydrography allowed a low vector density and malaria transmission through the dry season. Additional studies will be needed to have a better understanding of the specific contribution of different malaria vector species in the local malaria transmission. The level of annual malaria transmission remained generally similar from one locality to another (Bana, Pala and Souroukoudingan) and from 1 year to another and remains high despite recent bed net mass-distribution campaigns. Additional vector control tools are urgently needed to complement current malaria control interventions.