The mosquito vectors that sustained malaria transmission during the Magude project despite the combined deployment of indoor residual spraying, insecticide-treated nets and mass-drug administration.

The "Magude project" aimed but failed to interrupt local malaria transmission in Magude district, southern Mozambique, by using a comprehensive package of interventions, including indoor residual spraying (IRS), pyrethroid-only long-lasting insecticide treated nets (LLINs) and mass-drug administration (MDA). Here we present detailed information on the vector species that sustained malaria transmission, their association with malaria incidence and behaviors, and their amenability to the implemented control interventions. Mosquitoes were collected monthly between May 2015 and October 2017 in six sentinel sites in Magude district, using CDC light traps both indoors and outdoors. Anopheles arabiensis was the main vector during the project, while An. funestus s.s., An. merus, An. parensis and An. squamosus likely played a secondary role. The latter two species have never previously been found positive for Plasmodium falciparum in southern Mozambique. The intervention package successfully reduced vector sporozoite rates in all species throughout the project. IRS was effective in controlling An. funestus s.s. and An. parensis, which virtually disappeared after its first implementation, but less effective at controlling An. arabiensis. Despite suboptimal use, LLINs likely provided significant protection against An. arabiensis and An. merus that sought their host largely indoors when people where in bed. Adding IRS on top of LLINs and MDA likely added value to the control of malaria vectors during the Magude project. Future malaria elimination attempts in the area could benefit from i) increasing the use of LLINs, ii) using longer-lasting IRS products to counteract the increase in vector densities observed towards the end of the high transmission season, and iii) a higher coverage with MDA to reduce the likelihood of human infection. However, additional interventions targeting vectors that survive IRS and LLINs by biting outdoors or indoors before people go to bed, will be likely needed to achieve local malaria elimination.

Introduction

Despite the remarkable reductions in the malaria burden in sub-Saharan Africa over the last two decades [1,2], no country in this region has managed to eliminate malaria. The lowest malaria burden of all sub-Saharan Africa is observed in its southern part, namely in Namibia, Botswana, South Africa and eSwatini. In 2016, the World Health Organization (WHO) determined that eSwatini and South Africa had the potential to achieve zero indigenous cases by 2020. Albeit several regional malaria elimination efforts over the last few years [3,4], neither country was able to reach this target [5]. The importation of malaria cases from neighboring Mozambique, a country with considerably higher malaria transmission levels, has been highlighted as one of the causes [6]. In South Africa and eSwatini, malaria transmission is primarily concentrated in areas bordering Mozambique [5] and is driven by cases among migrant populations. Reducing or eliminating malaria in Mozambique, especially in its southern provinces, is therefore crucial to achieve malaria elimination in Southern African and eSwatini. The southern part of Mozambique has been targeted by initiatives to stop malaria transmission since the 1960’s, but, unfortunately, none have led to local malaria elimination [6-9].

The first attempt to eliminate malaria in southern Mozambique took place between 1960 and 1969 in the context of the Global Malaria Eradication Program (GMEP) [7]. The second initiative, the Lubombo Spatial Development Initiative (LSDI), was implemented between 1999 to 2011 [8-10]. Both were based on indoor residual spraying (IRS), which aims to kill mosquitoes resting on walls and ceilings with insecticides, although the second elimination attempt combined IRS with targeting the parasite reservoir using artemisinin combination therapies (ACTs). More recently, in 2015, the Mozambique, South Africa and eSwatini (MOSASWA) regional initiative [6] and the Mozambican Alliance Towards the Elimination of Malaria (MALTEM) [9] were established. MOSASWA aimed to strengthen regional collaboration and efforts to accelerate progress towards achieving malaria elimination in the region. MALTEM aimed, among other objectives, to create the necessary knowledge to inform an operational elimination plan and roadmap for malaria elimination in Mozambique [9], which was piloted during the Magude project.

The Magude project was designed to evaluate the feasibility of eliminating malaria in southern Mozambique with a package of interventions available at the time, namely a combination of interventions targeting the vector (IRS and long-lasting insecticidal nets, or LLINs) and the parasite reservoir (mass drug administration, or MDA, and standard diagnosis and treatment) simultaneously. In addition, it implemented strong community engagement campaigns to maximize the acceptance and coverage of all interventions. The project was expected to reduce vector densities with a combination of the killing effects from IRS and LLINs, and then reduce transmission by surviving infectious vectors through the prophylactic effect of the MDA drugs alongside the prevention of vector-human contact by LLINs, thereby closing the gap towards elimination [9].

While all of the initiatives listed above were successful at reducing the local burden of malaria [6-8], none of them achieved malaria elimination. Learning from past experiences is critical to guide future malaria elimination efforts in Mozambique and hence, to achieve elimination in the region. All these initiatives relied heavily on vector control, as do current and future malaria control efforts [11]. Therefore, 1) identifying the vectors that sustained malaria transmission despite the implemented vector control interventions and 2) evaluating the vectors’ amenability to the implemented vector control products, are crucial to understanding the shortcomings of the piloted approaches and, hence, to guide the design of future malaria elimination efforts in southern Africa.

The outcome of the two aims above are presented here. We first describe anopheline species composition, densities, host-seeking behavior (time and place) and P. falciparum sporozoite rates during the course of the project. We then evaluate relative vector importance by exploring the association between densities of different vector species and malaria incidence, accounting for the implemented interventions. We subsequently combine these findings with previously published data on i) the efficacy of the three core interventions, LLINs, IRS and MDA [12], ii) susceptibility of the vectors to the used insecticides, and iii) the overlap between human and vector behaviors, to examine the ability of the implemented interventions to control the local vector populations. Finally, we use our new understanding to provide vector control recommendation for future malaria elimination efforts in the area.

Materials and methods

The Magude project in the Magude district

Magude district is a rural district located in Maputo province (southern Mozambique) that borders with Bilene district to the north and with Mpumalanga Province in South Africa to the West (Fig 1). The district has an area of 6,961 km2 and is divided into five administrative posts, Magude Sede, Motaze, Panjane, Mahele and Mapulanguene. It had 48,448 residents in 2015 [13]. The vegetation is dominated by open forests and savannahs, and three main rivers cross the district. Most of the population relies on subsistence agriculture, fishing or working as sugar cane cutters in the sugar plantations in Magude, Xinavane or the neighboring Manhiça district. Houses are traditional round-shaped or rectangular-shaped huts constructed with cane, cement, mud brick or reeds and covered by adobe or cement. A comprehensive description of the district demographic, socio-economic and health characteristics, is provided elsewhere [13]. Two distinct climatological seasons are observed in southern Mozambique, a rainy season extending from October to March and a dry season from April to September. The high malaria incidence season occurs from November to April [13].

Fig 1

Map of Magude district highlighting the administrative posts and sentinel sites for entomological surveillance.

Dots represent the houses where mosquitoes were collected. Map borders were obtained from the Humanitarian Data Exchange under license "Creative commons attribution for Intergovernmental organizations. (CC-BY-IGO). https://data.humdata.org/faqs/licenses.

The Magude project started in 2015 by establishing a health and demographic platform to obtain the necessary information to guide the implementation of the core interventions during the project [13]. The district was covered (i.e. one net for every two people in a household) with LLINs that were distributed by the National Malaria Control Program (NMCP) through a mass distribution campaign in May 2014. Between August 2015 and December 2017, the project delivered an annual single-round of IRS with DDT and pirimiphos-methyl in the first year and only pirimiphos-methyl in the second and third year, and two annual rounds of MDA that were one month apart. During each MDA round, the de-facto population of Magude (including visitors but excluding infants <6 months, women in the first trimester of pregnancy or severely ill individuals) received a full 3-day course of dihydroartemisinin–piperaquine (DHAp). Community mobilization campaigns were implemented to increase the uptake of the interventions. In parallel, the NMCP provided diagnosis (HRP2-based RDT and microscopy) and treatment (artemether–lumefantrine as first line drug for clinical cases) to patients presenting at health facilities or community health workers. To monitor malaria incidence and prevalence, the Magude project established a rapid case reporting system and conducted annual cross-sectional malaria prevalence surveys [9,12]. The epidemiological results of the project, including information on the MDA campaigns, have been published elsewhere [12], as well as detailed information on the implementation and coverage of IRS [14], access and use of LLINs [15], the susceptibility of the main local vector species to insecticides [14] and a detailed evaluation of the overlap between human and mosquito behaviors [16]. This paper combines novel and detailed data on mosquito bionomics with those data that are analyzed previously to improve our understanding of the effectiveness of the project’s approach.

Entomological surveillance design

Mosquitoes were collected monthly between May 2015 and October 2017 in six sentinel sites in Magude district (Fig 1): Magude Sede, Muginge, Panjane, Mapulanguene, Chicutso and Motaze (Fig 2). These six sentinel sites were selected to represent the range of environmental and land use characteristics of the district.

Fig 2

Indoor and outdoor anopheline species composition before and after the implementation of the first IRS campaign during the Magude project.

In each sentinel site, mosquitoes were collected in fifteen households during two consecutive nights. They were collected indoors in 10 households and outdoors in another 5 households using CDC miniature light-trap (Model 512, John W Hock, Florida, USA). A Collection Bottle Rotator (Model 1512, John W Hock, Florida, USA) was added to six traps (three indoors and three outdoors, every night) to assess the time of mosquito host-seeking activity. The same houses were visited every month, but each month they were randomly assigned a trap type (i.e. CDC-light trap with or without a rotator) and a collection location (indoor or outdoor). Indoors, the CDC light-trap was hung at the foot-end of a bed with the trap opening approx. 1.5m above the ground. One or two adult (>15 years old) volunteers from the selected household were asked to sleep in the bed under an LLIN during the night. Participants not owning a net were provided with a WHO-pre-qualified pyrethroid-only LLIN. Outdoors, CDC light traps were baited with a BG-Lure cartridge (Biogents AG, Germany) and CO2 (generated through a mixture of 10g commercially available yeast (Instant yeast, Smart Chef), 100g white refined household sugar and 1L of regular tap water) to mimic indoor conditions (i.e. a human sleeping next to the trap). The outdoor traps were placed in the safest possible outdoor environment: under a tree close to the house, but away from animals and children. Due to suspicion of arboviral disease transmission in Mozambique, which was later confirmed [17,18], no comparisons against human landing catches (HLC) were done. Hence, we discuss here exposure to host-seeking mosquitoes rather than providing human biting rates.

Every morning after a collection night, the team visited the house to retrieve the collected mosquitoes and to record information on the quality of the collection using a digital structured questionnaire. This served to exclude collections that did not match our inclusion criteria (listed in S1 File). Data were collected with tablets (Huawei, Model S7-701u) using Open Data Kit. The collected mosquitoes were taken to the laboratory. Anopheles mosquitoes were selected and identified morphologically to species using a stereomicroscope and the keys of Gillies and Coetzee [19]. Individuals belonging to the An. gambiae and An. funestus species complex were transferred to the lab and identified to species level using the polymerase chain reaction (PCR) [20,21]. The presence of P. falciparum sporozoites in individual mosquito samples was analyzed through screening enzyme-linked immunosorbent assays (ELISA) conducted on mosquitoes’ grinded head and thorax [22]. The presence of sporozoites of other Plasmodium species was not tested because P. falciparum is known to account for over 90% of all diagnosed malaria infections in Mozambique [23] and for almost all in the neighboring district of Manhiça [24], and because very low positivity rates were expected given the elimination context. Positive samples were confirmed through a second ELISA test. ELISA lysates were not heated before running the test and positive samples were not confirmed by PCR or gene sequencing.

Indoor resting vectors

Indoor resting mosquitoes were not collected systematically, however, indoor resting blood-fed mosquitoes were collected for insecticide resistance monitoring purposes. Mosquitoes were collected from 6 am to 10 am using a mouth aspirator and a torch from April to September and in December of 2015, from February to August of 2016 and from August to November of 2017. A descriptive analysis is provided, as indoor resting behaviors are closely linked with the success of IRS campaigns (i.e. IRS products kill susceptible mosquitoes that rest on sprayed surfaces indoors).

Data analysis

The analysis aimed to 1) evaluate anopheline composition and densities over time; 2) quantify P. falciparum sporozoites per mosquito species and the impact of interventions on those sporozoite rates over time; 3) evaluate the association of different vector species with reported malaria cases, 4) evaluate the location and time of vector host-seeking activity and 5) identify the vector species that rested indoors. With these aims, the analysis makes use of the data collected through the surveillance system described in detail above as well of previously published datasets on malaria incidence, the efficacy of IRS, on ITN and MDA coverage and climatic data [12]. The results are discussed considering results from previously published analyses (see section below) to better understand the impact of the implemented interventions on local vector populations. Since the Magude project did not have a control area, nor a sufficiently long entomological baseline, this examination is mostly qualitative in nature, although we tried to quantify the impact where possible (e.g. the impact of interventions on sporozoite rates).

Evaluation of Anopheles composition and densities over time

We first evaluate Anopheles species compositions by calculating the relative abundance of each Anopheles species (i.e. the proportion of mosquitoes that belonged to each Anopheles species out of the total number of anophelines collected). We calculate each species’ relative abundance for the period before the first intervention of the project (May to July 2015, before the first IRS campaign), and for the full intervention period (August 2015 to October 2017). For the intervention period, we calculate the relative abundance of each species indoors and outdoors separately. We then calculate the ratio of mosquitoes collected indoors to those collected outdoors to evaluate the overall endophagy of each species during the intervention period. Since the number of collections indoors and outdoors per month was different, we normalized the number of mosquitoes collected indoors and outdoors in each month by dividing those numbers by the number of collections conducted indoors and outdoors, respectively, that month.

We calculate the number of host-seeking Anopheles mosquitoes per person per month as the mean value of the number of host-seeking Anopheles mosquitoes per person in each collection within that month, separated by species. The number of host-seeking Anopheles mosquitoes per person in each collection was collected as the number of Anopheles collected divided by the number of people who slept under a net next to the trap in the collection room that night. For outdoor collections, we assumed that our artificial lure mimicked a single person, and hence the number of mosquitoes that was collected is divided by 1. The number of monthly host-seeking mosquitoes per person for each Anopheles species from May 2015 until October 2017 is plotted alongside intervention coverage, use and/or efficacy, malaria incidence, as well as temperature and rainfall data, with the aim to visually explore potential associations between vector bionomics, interventions, climate and malaria incidence. Interventions coverage, use and efficacy data were obtained from previous publications [12,14,15]. Rainfall data were obtained from the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS). Data from every raster file per month were extracted for every household in Magude and aggregated to obtain monthly representative values [25]. Temperature data was obtained from the National Oceanic and Atmospheric Administration (NOAA) collected by the Maputo Weather Station (station ID 673410).

P. falciparum sporozoite rates and impact of interventions on these rates over time

Since very few mosquitoes were found carrying sporozoites throughout the entire project, we present the overall sporozoite rate (i.e. the number of P. falciparum positive mosquitoes over all mosquitoes analyzed) and the number of Anopheles mosquitoes of each species that were found carrying P. falciparum sporozoites, separated by those that were collected indoors or outdoors. Overall sporozoite rates are subsequently provided for specific project periods related to the time of implementation of MDA and IRS, with the aim to understand the potential impact of these intervention on sporozoite rates. The impact of both MDA and IRS on sporozoite rates can already be observed two to three weeks after implementation. MDA with DHAp immediately eliminates gametocides from humans, which prevents feeding mosquitoes from ingesting gametocytes and becoming infective. The time between gametocyte ingestion and sporozoite migration to mosquito salivary glands can be two or three weeks depending on temperature. By then, a proportion of older infected mosquitoes will have died, only a few younger mosquitoes will be infected and hence sporozoite rates will be lower than before MDA. IRS immediately reduces vector densities through mortality-inducing effects, reducing parasite transmission success (from human to mosquitoes and vice versa). In addition, new mosquitoes that emerge during the two or three weeks after each IRS round are unlikely to become infected due to the lower levels of circulating parasites in the human population. As a result, sporozoite rates are expected to decrease. Since the temporal resolution of our data is monthly, we considered the following periods for sporozoite analysis to assess how both interventions may have impacted transmission: 1) prior to the first IRS campaign (May-July 2015), 2) between the start of the first IRS and the start of MDA 1 (August–October 2015), 3) during MDAs 1 and 2 (November 2015-February 2016), 4) at the end the high transmission season 2016 (March-July 2016), 5) between the start of second IRS and the start of third MDA (August to November 2016), 6) During MDAs 3 and 4 (December 2016 to March 2017), 7) at the end of the high transmission season 2017 (April to July 2017) and 8) from August to October 2017. Sporozoite rates are presented together with their 95% confidence intervals (CIs), calculated as confidence intervals of a population proportion assuming the sample meets the Central Limit Theorem (Table 1). Because we had low numbers of mosquitoes and low vector sporozoite rates, no further statistical analyses were undertaken.

Table 1

Plasmodium falciparum sporozoite rates of the five Pf positive malaria vector species during the Magude project at distinct relevant periods defined in relation to interventions.

The proportions with 95% CI, and number of positive with respect to total mosquitoes collected in the period are shown.

Species	Before first IRS campaign(May-Jul 2015)	Between the start of the first IRS and the start of MDA 1(Aug-Oct 2015)	During MDAs 1 and 2(Nov-Feb 2016)	End the high transmission season 2016(Mar-Jul 2016)	Between the start of second IRS and the start of third MDA(Aug-Nov 2016)	During MDAs 3 and 4(Dec-Mar 2017)	End the high transmission season 2017(Apr-Jul 2017)	From Aug 2017 to October 2017
An. arabiensis	5.4%[3–9.5] (12/222)	2.9%[1.1.-6.9] (5/174)	0.0%[0–2.8] (0/168)	0.2%[0–0.7] (2/1165)	0%[0–4.2] (0/109)	0.3%[0.1–1.2] (2/655)	1.3%[0.6–2.9] (27/525)	11.8%[3.8–28.4] (4/34)
An. funestus s.s	2.7% [0.1–15.8] (1/37)	0%[0–37.1] (0/9)	Sporozoite presence not analyzed (n = 3)	0.0%[0–28.3] (0/13)	0%[0–60.4] (0/4)	0.0%[0–37.1] (0/9)	0.0%[0–17.2] (0/24)	0.0%[0–53.7] (0/5)
An. parensis	1.6% [0.1–9.7] (1/63)	0%[0–12.6] (0/34)	No An. parensis collected	No An. parensis collected	0%[0–94.5](0/1)	No An. parensis collected	0%[0–69](0/3)	0.0%[0–34.5] (0/10)
An. squamosus	0%[0–40.2] (0/8)	20%[1.1–70.1] (1/5)	0%[0–80.2] (0/2)	0%[0–80.2] (0/2)	No An. squamosus collected	Sporozoite presence not analyzed (n = 235)	Sporozoite presence not analyzed (n = 121)	Sporozoite presence not analyzed(n = 14)
An. merus	No An. merus collected	No An. merus collected	No An. merus collected	0.%[0–9.4] (0/47)	0.0%[0–94.5] (0/1)	0.0%[0–12.6] (0/36)	0.0%[0–10.7] (0/41)	20%[1.1–70.1] (1/5)
All five species combined	3.7% [2.1–6.2] (14/330)	2.5%[1–5.6] (6/222)	0%[0–2.5] (0/186)	0.2%[0–0.6] (2/1295)	0.0%[0–3.9] (0/120)	0.3%[0–1.1] (2/757)	1.3.%[0.6–2.7] (0/611)	7.5%[2.8–17.3] (5/67)

Association between vector species densities and malaria incidence

To understand the relative importance of each potential vector in malaria transmission and given the fact that the number of sporozoite positive mosquitoes were too low to obtain accurate estimates of the entomological inoculation rates, the association between the number of host-seeking anophelines of each species collected per month per person and the monthly malaria incidence was explored through a negative binomial multivariate regression model. The results of this model are later combined with data on the relative vector abundance, vector behaviors and how those overlap with human behaviors, to understand which species were most likely the main malaria vectors during the project.

The model correlated monthly malaria cases, as diagnosed by RDTs, with the monthly number of mosquitoes collected per person per month of all Anopheles species that represented at least 1% of the total vector population. The analysis was not restricted to those vectors carrying sporozoites, as other anopheline species collected during the project are known vectors in surrounding countries [26], and the low number of mosquitoes collected of some of these species may have prevented us from detecting sporozoites in their population. Monthly LLIN use, IRS residual efficacy and MDA coverage were included in the analysis because these interventions confound the effect of vector densities on malaria cases, as they can prevent mosquito entry into houses, reduce contact between vectors and humans and reduce the proportion of infected vectors. Since our temporal resolution is a month but the effect of the implemented interventions on malaria cases can be observed within two-three weeks (given the biological cycle of malaria transmission and the vector’s lifecycle) two models were fitted: one considering unlagged covariates and one considering covariates lagged one month. In addition, we fitted these two models for the entire vector surveillance period from May 2015 to October 2017 (i.e. including the baseline period May to July 2015), but also for the period August 2015 to October 2017 (intervention period only). The number of visits to a health facility was used as the offset, to account for variations in care seeking behaviors over time. Malaria incidence data were obtained from the DHIS2-based rapid case reporting system established in Magude district in January 2015 [12]. MDA coverage was assumed to be equal to the campaign coverage during the two months that each campaign lasted and 0 for the other months. MDA coverage estimates were obtained from previously published analyses [12]. Monthly IRS efficacy estimates for the 2016 and 2017 campaigns were obtained by fitting a logistic binomial Bayesian model to the observed mosquito mortality data 24h-post exposure in WHO standard cones bioassays. More details on the collection of residual efficacy data, the residual efficacy data themselves, and the data analysis are provided elsewhere [14]. Since the residual efficacy of the products used during the 2015 IRS campaign was not monitored, we assumed that the residual efficacy of Actellic 300CS and DDT in 2015 was similar to the efficacy of Actellic 300CS alone in 2016. Similar residual efficacies for both products have been observed in other campaigns [27], and due to the overall high variability of the residual efficacy of IRS products across geographical locations [27-29], we refrained from including DDT residual efficacy data from other sites. To calculate the residual efficacy of the 2015 IRS campaign, the residual efficacy of Actellic 300CS in 2015 was adjusted for the observed pace of spraying and coverage (94.5%) of the campaign that year (following the exact same method described elsewhere for the 2016 and 2017 IRS campaigns [14]. For the models that included the baseline period (May-July 2015), the residual efficacy of the 2014 IRS with deltamethrin was considered zero, as only Motaze (that accounts for 13.5% of the Magude population) received IRS and because the optimal residual efficacy of deltamethrin has been observed to be between 3–6 months in other settings and is therefore expected to have waned by May 2015 when the project started [27]. LLIN use was measured several times during the Magude project (details on the methodology and data collection are provided elsewhere [15] and followed a seasonal pattern, which was modelled using a sinusoidal function, where x is the month, A is the amplitude of the variation which we modeled as , B is the period, which for months is , C was adjusted for the sinusoidal function to follow the seasonality of LLIN use and D is the minimum observed use (39.1%) plus the amplitude of the variation (the function is represented in S2 File). The goodness of fit was evaluated by checking the distribution and autocorrelation of the residuals. Models were compared using the Akaike Information Criterion. The model with the lowest AIC was considered to be the best performing model, provided there was no autocorrelation in its residuals. Regression log transformed coefficients are reported together with their 95% confidence intervals. The predicted cases are shown along with true malaria cases. Detailed model results are provided in S4 File.

Vector host-seeking activity

We evaluated indoor and outdoor host-seeking times during the project’s intervention period (August 2015 to October 2017) by calculating the number of host-seeking mosquitoes of each species collected per person for each 2 hour time interval (period of rotation of the CDC bottle rotator) from 18:00 to 06:00, before 18:00 and after 06:00, separating indoor and outdoor collected mosquitoes. Then we evaluated the composition of host-seeking vectors at the different collection time intervals by calculating the relative percentage of the total host-seeking mosquitoes per person that each species represented.

All data cleaning and analysis was conducted using R version 4.1.0.

Ethical clearance

Ethical approval was obtained from the Manhiça Health Research Center’s Institutional Bioethics Committee for Health (CIBS-CISM/043/2015 for our entomological surveillance) and local administrative authorities (52/SDSMASS/024.1). Verbal informed consent was obtained from an adult member of each household where a mosquito trap was placed indoors or outdoors. All participating households were free to withdraw from the studies at any given time. All other studies were approved by CISM’s institutional ethics committee, Hospital Clinic of Barcelona’s Ethics Committee, and the Mozambican Ministry of Health National Bioethics Committee. The study protocol to implement and evaluate the impact of MDAs was also approved by the pharmaceutical department of the Ministry of Health of Mozambique and registered as Clinical Trial NCT02914145. More details on the ethical considerations of the population census, household surveys, cross-sectional prevalence surveys and MDAs are provided elsewhere [12,13]

Results

Mosquito collections

A total of 5,361 trap-night collections were performed between May 2015 and October 2017. Of those, 513 collections were discarded for not complying with the inclusion criteria (S1 File). As a result, 4,848 trap-night collections were considered in the present analysis, 3,329 indoors (933 in CDC light traps with collection bottle rotators, and 2,396 in CDC light traps without rotators) and 1,519 outdoors (808 in CDC light traps with collection bottle rotators, and 711 in CDC light traps without). Only 18.4% of the trap-night collections yielded at least one female Anopheles mosquito, with 81.6% resulting in zero mosquitoes caught. A total of 4,655 Anopheles female mosquitoes were caught, 4,107 indoors (1,015 in CDC light traps with bottle rotators and 3,092 with CDC light traps without) and 548 outdoors (243 with CDC light traps with bottle rotators and 305 in CDC light traps without). Accounting for the differences in the number of sampling nights indoors and outdoors, these numbers indicate indoor-outdoor ratios of roughly 3.4 to 1 respectively.

Anopheline species composition and densities over time

Ninety-seven percent (97.5%, n = 4,539) of all collected mosquitoes were identified morphologically. Of the indoor collected mosquitoes 1.9% (n = 93) could not be identified; of the outdoor collected mosquitoes 5.9% (n = 37) could not be identified because they were either too damaged or because the microscopists could not find a matching species in the dichotomous keys. Molecular identification was performed for 98% (n = 3,364) of mosquitoes belonging to the An. gambiae complex, and 87.3% (n = 332) of mosquitoes belonging to the An. funestus group.

Before the scale up of IRS (May-July 2015, i.e. the dry season), mosquitoes from the An. gambiae complex (all identified as An. arabiensis) accounted for 56.8% (n = 225) of the Anopheles collected, and those from the An. funestus group accounted for 36.1% (n = 143). Most mosquitoes from the latter species group were identified as An. parensis (44.8%, n = 64), followed by An. funestus s.s. (23.8%, n = 37), An. leesoni (1.4%, n = 2) and An. rivulorum (4.2%, n = 6). The other 23.8% (n = 34) of the mosquitoes in this group could not be identified to species. An. squamosus accounted for 2.3% (n = 9) of the anopheline collected. Other Anopheles species accounted for 6.8% of the mosquito population, but less than six individuals of each of these other species were collected.

From the scale up of IRS (August 2015) onwards (i.e. intervention period), mosquitoes belonging to the An. gambiae complex continued to account for the majority of anophelines collected (75.3%, n = 3,206). Molecular identification revealed the following composition: An. arabiensis (91.6%, n = 2,938), An. merus (4.2%, n = 135), An. quadriannulatus (2%, n = 63), An. gambiae s.s. (0.1%, n = 3) and 2.1% (n = 67) could not be identified. An. squamosus accounted for 8.9% (n = 380) of the Anopheles collected followed by the An. funestus group (5.6%, n = 237). The species composition within this group was An. parensis (50.2%, n = 119), An. funestus s.s. (37.6%, n = 89), An. rivulorum (5.1%, n = 12), An. leesoni (0.8%, n = 2) and 6.5% (n = 15) could not be identified. Finally, An. ziemanni accounted for 2.7% (n = 114), An. pharoensis for 2.3% (n = 99), An. rufipes for 1.9% (n = 83) and several other vector species for 3.3%, with less than 12 mosquitoes of each of these other species collected.

The indoor vector composition during the intervention period was: An. arabiensis (76.6%, n = 2,873), An. squamosus (5.8%, n = 217), An. merus (3.5%, n = 133), An. parensis (2.5%, n = 92), An. funestus s.s. (1.9%, n = 71), unidentified mosquitoes of the An. gambiae complex (1.7%, n = 62), An. quadriannulatus (1.5%, n = 55), An. ziemanni (1.4%, n = 52), An. rufipes (1.3%, n = 49), An. pharoensis (0.9%, n = 32), unidentified mosquitoes of the An. funestus group (0.3%, n = 12), An. listeri (0.2%, n = 6), An rivulorum (0.1%, n = 5), An. coustani s.l. (0.1%, n = 4), An. gambiae s.s. (0.1%, n = 3), An. leesoni (0.1%, n = 2), An. tenebrosus (0.1%, n = 2), An. multicolor (0%, n = 1) and 2.1% (n = 79) could not be identified (Fig 2).

The outdoor vector composition during the intervention period was: An. squamosus (32%, n = 163), An. pharoensis (13.2%, n = 67), An. arabiensis (12.8%, n = 65), An. ziemanni (12.2%, n = 62), An. rufipes (6.7%, n = 34), An. parensis (5.3%, n = 27), An. funestus s.s. (3.5%, n = 18), An. quadriannulatus (1.6%, n = 8), An. rivulorum (1.4%, n = 7), An. tenebrosus (1.4%, n = 7), unidentified mosquitoes of the An. gambiae complex (1.0%, n = 5), An. listeri (0.8%, n = 4), unidentified mosquitoes of the An. funestus group (0.6%, n = 3), An. merus (0.4%, n = 2), An. coustani s.l. (0.2%, n = 1), An. garnhami (0.2%, n = 1) and 6.9% (n = 34) could not be identified (Fig 2).

After normalizing the total number of mosquitoes collected indoors and outdoors by the number of collection nights indoors and outdoors, respectively, the ratio indoor to outdoor collected mosquitoes for the different species during the intervention period was as follows: An. arabiensis (20.0 indoors to 1 outdoors), An. merus (30.1:1), An. quandriannulatus (3.1:1), An. coustani s.l. (1.8:1), An. funestus s.s. (1.8:1), An. parensis (1.5:1), An. listeri (0.7:1), An. rufipes (0.7:1), An. squamosus (0.6:1), An. ziemanni (0.4: 1), An. rivulorum (0.3:1), An. pharoensis (0.2: 1) and An. tenebrosus (0.1:1). This suggests that An. arabiensis and An. merus were highly endophagic, that An. coustani s.l., An. funestus s.s., and An. parensis were slightly more endophagic than exophagic and that An. listeri, An. rufipes, An. squamosus, An. ziemanni, An. rivulorum, An. pharoensis and An. tenebrosus were more exophagic.

The number of indoor host-seeking mosquitoes per person oscillated seasonally reaching a maximum in the month of April in both years, after the annual rainfall peak. More than 55% of the total host-seeking anopheline mosquitoes per person were collected during the months of February, March and April, with 27.7% of all mosquitoes being collected during the month of April (Fig 3). The greatest species richness was observed during the first half of 2017 (Fig 3), following the highest rainfall, and preceding the highest malaria incidence observed during the project.

Fig 3

Anopheline densities over time during the Magude project in relation to malaria cases, climate data, and relevant malaria control intervention indicators.

Number of anopheline mosquitoes collected (per person per month, colored bars), malaria cases (black line), modeled LLIN use over time (%, dashed grey line), modeled IRS efficacy over time (%, grey shaded areas), MDA coverage (%, blue shaded areas), amount of rainfall (mm, blue line) and temperature (°C, orange line).

Indoors, An. arabiensis was the predominant vector before and during the intervention period, accounting for between 48% and 100% of all host-seeking anophelines collected per person in any given month until September 2017. The proportion of host-seeking An. funestus s.s. and An. parensis decreased markedly after the first IRS campaign. An. parensis disappeared and was only collected again in November 2016 and from July 2017 onwards, accounting for 62.2% and 27.0% of host-seeking anophelines in September and October 2017, respectively. Anopheles funestus s.s. was occasionally present during the period when IRS was effective, but only accounted for 0.6% up to 12.1% of indoor host-seeking anophelines when it was found. An. merus was only collected after the first IRS campaign and accounted for between 1.2% and 26.0% of host-seeking anophelines when it was found. Indoor host-seeking An. squamosus accounted for between 0.2% and 4.0% of host-seeking anophelines before and after the first IRS campaign until February 2017, when it accounted for 17.5%. Its relative presence remained high in subsequent months (14.5% and 14.9% in March and April 2017, respectively). An. ziemanni was only present in larger numbers in 2017, but still accounted for less than 5% of the vectors collected indoors until May and June 2017, when its relative abundance increased to 7.2% and 7.8%, respectively (Fig 3).

Outdoors, different species dominated in different months. There was no single species that predominated over time during the intervention period, and several species were only collected outdoors in specific months. Two peaks in the number of outdoor host-seeking anophelines per person were observed. The first peak was in April 2016, the second in February 2017, with 15.1% and 31.0% of all mosquitoes collected after the first IRS campaign, respectively. The peak in April 2016 was dominated by An. pharoensis and preceded an increase in malaria cases. The peak in February 2017 was dominated by host-seeking An. squamous and was also followed by an increase in malaria cases. Outdoor host-seeking An. arabiensis were collected outdoors throughout the project. Outdoor host-seeking An. parensis accounted for a high proportion of outdoor host-seeking anophelines before the first IRS campaign (17.6% to 55.5%) and were absent until September and October 2017, when they accounted for 61.9% and 27.8%, respectively. Outdoor host-seeking An. funestus s.s. were sporadically collected throughout the study. Outdoor host-seeking An. squamosus accounted for between 15.4% and 37.5% of all outdoor host-seeking Anopheles before the first IRS campaign, but were mostly absent until December 2016. Between December 2016 and July 2017, they accounted for between 18.1% and 66.6% of the monthly outdoor host-seeking Anopheles collected. An. ziemanni was absent from outdoor collections from the first implementation of IRS until December 2016, when its relative abundance increased up to 42.8% in June 2017, preceding an increase in malaria incidence, and remained substantial until the end of the implementation period (October 2017) (Fig 3).

Sporozoite rates

The presence/absence of Plasmodium falciparum sporozoites was determined for 3,656 specimens (78.5% of all mosquitoes collected). A total of 37 (0.8%) mosquitoes were sporozoite positive; 35 collected indoors and two collected outdoors. These belonged to five species: An. arabiensis (32 positive of 3,052 tested, 1%), An. merus (1/128, 0.8%), An. parensis (1/111, 0.9%), An. funestus s.s. (1/101, 1%), An. squamosus (1/17, 5.9%), and one unidentified mosquito from the An. gambiae complex (1/62, 1.6%). Of the 59 An. quadriannulatus, 46 An. pharoensis, 15 An. rivulorum, 8 An. rufipes, 6 An. coustani s.l., 5 An. ziemanni, 3 An. gambiae s.s., 3 An. marshallii complex, 2 An. leesoni, 1 An. demeilloni, 1 An. garnhami, 1 An. pretoriensis and 1 An. tenebrosus analyzed, none were positive. None of the An. listeria, An. multicolor or An. salbaii were tested for sporozoites.

Indoors, sporozoite rates were 1% for An. arabiensis (31/2987), 0.8% for An. merus (1/126), 1.1% for An. parensis (1/87) and An. funestus s.s. (1/88) and 33.3% for An. squamosus (1/3). The two outdoor sporozoite positive mosquitoes included one An. arabiensis (1/65, 1.5%) and one mosquito from the An. gambiae complex that could not be identified to species (1/5, 20%).

Sporozoite-positive An. arabiensis were detected for the duration of the project while other species only tested positive sporadically. After the first MDA campaign, only An. arabiensis and An. merus were found positive for P. falciparum. The sporozoite rates for the five vector species during specific points in time (described in the methods and related to the timing of our malaria and vector control interventions) are shown in Table 1. More details on sporozoite rates per species and month are provide in S3 File.

Overall, vector sporozoite rates decreased with the implementation of IRS and the two first rounds of MDA but increased again at the end of the high transmission season of 2017 when the highest malaria incidence of the entire project was observed (Table 1). An unexpected increase in sporozoite rates was subsequently observed in August 2017 (12.5% [5.2–25.9]) and September 2017 (20.0% [6.6–44.3]) (S3 File) and only An. arabiensis and one An. merus mosquito were sporozoite positive during those months.

Association between host-seeking Anopheles per person and residual malaria incidence

Only species accounting for more than 1% of the anophelines collected were included in the analysis. Of the models used to correlate the number of host-seeking mosquitoes per person per month and per species, and controlling for MDA coverage, LLIN use and IRS efficacy, the best-performing model was the one including covariates lagged one month and which only considered the intervention period (August 2015 to October 2017). It had an AIC of 318.55 compared to 348.4, 377.2, 399.4 for the other models (detailed results from each model are presented in S4 File). In this best performing model, An. arabiensis, An. funestus s.s. An. parensis, An. squamosus, An. merus, An. rufipes and An. quadriannulatus were significantly associated with malaria incidence. An. arabiensis at <1% significance, An. parensis and An. quamosus at 1% significance, An. funestus s.s., An. merus and An. quadriannulatus at 5% significance and An. rufipes at 10% significance. The coefficients, reflecting the likely increase in malaria cases expected with the respective presence of each mosquito if positive, were: An. arabiensis 2.93 [95% CI: 2.06–4.19], An. parensis 23.4 [4.41–248.58], An. squamosus 30.21 [4.21–218.55], An. funestus s.s. 6.76×10−6 [3.64×10−10–0.11], An. merus 2.75×10−2 [1.14×10−3–0.74], An. rufipes 0.04 [1.42×10−3–1.26], An. quadriannulatus 4.31×10−7 [2.7×10−12–0.08]. Fig 4 shows the actual reported malaria cases and the cases predicted by the best performing model.

Fig 4

True malaria cases versus model predicted cases.

Shaded areas represent 95% CI.

Place and time of host-seeking anopheline mosquitoes

Fig 5 shows both the composition of anopheline host-seeking mosquitoes and their densities for each 2h interval during the evening, night and early morning during the intervention period (August 2015 to October 2017). Indoors, most of the host seeking mosquitoes were An. arabiensis. Host seeking activity was concentrated in the evening and nighttime hours before 02:00 and showed the greatest peak between 00:00 and 02:00. Outdoors, host-seeking activity was concentrated between 18:00 and 00:00 and between 02:00 and 04:00. Early hours saw a marked presence of host-seeking mosquitoes of the An. funestus group and An. squamosus mosquitoes. The peak between 02:00 to 04:00 was dominated by An. squamosus and An. pharoensis.

Fig 5

Density of host-seeking mosquitoes over different time intervals during the evening, night and early morning.

Identification of indoor resting vectors

A total of 1,042 blood-fed An. funestus s.l. mosquitoes were collected indoors in 2015, before the first IRS campaign was implemented. Very few mosquitoes of this group (insufficient for insecticide susceptibility testing) were collected indoors during the years following the first IRS campaign. Blood-fed mosquitoes of the An. gambiae complex were found resting indoors in sufficient numbers for resistance testing during the entire intervention period (1,024 in 2015, 3,753 in 2016 and 508 in 2017). A subset of An. funestus s.l. mosquitoes collected indoors in 2015 were identified, and the following species detected: An. parensis, An. funestus s.s., An. rivolorum and An. vaneedeni. Among the An. gambiae complex, most individuals were An. arabiensis, a few were An. merus. Although indoor resting behavior was not assessed through pyrethrum spray catches or similar methods, these data suggest that a part of the population of members of the An. funestus group and of An. arabiensis may rest indoors during a part of their gonotrophic cycle.

Discussion

The present study aimed to i) identify the anopheline species that sustained malaria transmission during the Magude project, ii) qualitatively evaluate the impact of the interventions on those species, and iii) identify the potential gaps in vector control during the project, to provide recommendations for future malaria elimination efforts in the area.

We believe that five species sustained transmission during the Magude project—An. arabiensis, An. merus, An. funestus s.s., An. parensis and An. squamosus—with others potentially playing a very minor role, if any. These five species were found positive for P. falciparum sporozoites and were significantly associated with malaria incidence. In southern Mozambique, An. funestus s.s. and An. arabiensis have historically been identified as the major malaria vectors [7,8,30]. Anopheles merus was first incriminated as a vector in year 2000 in Boane, located in Maputo province [31] and later during the Lubombo Spatial Development Initiative in the same province [8]. In contrast, and to our best knowledge, this is the first time that An. squamosus and An. parensis are found carrying P. falciparum sporozoites in southern Mozambique. Anopheles squamosus has been incriminated as a malaria vector in southern Zambia, but its exact role in malaria transmission in that area is unknown [32]. In 2017, An. parensis was identified as a minor vector species in Kwazulu-Natal, a province in South Africa that borders the southern part of Mozambique [33,34]. Although P. falciparum sporozoites were not detected in any of the An. rufipes (8 analyzed out of 83 identified) and An. quadriannulatus (59 analyzed out of 63 identified), these two species were significantly associated with malaria cases, albeit with the weakest associations and very small correlation coefficients. Although it has been demonstrated that An. quadriannulatus can carry P. falciparum sporozoites in laboratory studies [35], this species has never been incriminated as a malaria vector in nature and therefore unlikely played a role in malaria transmission during the Magude project. Anopheles rufipes, on the other hand, is a secondary malaria vector in equatorial countries of Africa (Burkina Faso, Cameroon, Gambia, Ghana, Kenya, Mali, Nigeria, Senegal and Togo) as well as in Zambia [26,36-44]. As such, this vector could have played a minor role during the project.

Other anopheline species that were found in Magude district are known vectors of malaria in southern Mozambique, i.e. An. tenebrosus [45], or elsewhere in Africa, namely An. coustani s.l [46-49], An. ziemanni [50-53], An. rivulorum [49,54,55], An. leesoni [44,56-58] and An. pharoensis [39,49,59-61]. However, none of the few analyzed specimens of these species were found carrying P. falciparum sporozoites. In addition, several non-vector Anopheles species were collected, namely An. demeilloni, An. garnhami, An. listeri, An. marshalli s.l., An. multicolor and An. salbaii. To our knowledge, this is the first time that An. garnhami, An. multicolor and An. salbaii are collected in Mozambique [26]. Of these, only An. garnhami has been detected in Mozambique’s neighboring countries (i.e. South Africa, Tanzania and Zimbabwe) [26]. In contrast, An. multicolor and An. salbaii have been detected in equatorial Africa, namely Niger and Sudan (An. multicolor) and in Ethiopia, Kenya, Niger, Somalia and Djibouti [26]. The detection of these other vector and non-vector Anopheles species could, however, be an artifact of wrong morphological identification resulting from damaged mosquito specimens [49,62]. Unfortunately, none of these species were molecularly identified to species (e.g. using ITS2 and COI analysis [49]) and the presence of sporozoites was not assessed for several Anopheles species. Confirming the presence of these species and investigating their role in persistent malaria transmission, especially of those known to be vectors in other countries of Africa, should be a priority in future studies in southern Mozambique.

Three specimens of An. gambiae s.s., a very competent vector of malaria transmission in sub-Saharan Africa, were found in Magude district. Although this vector is nowadays rarely found in southern Africa, it was found resting indoors in Magude’s neighboring district of Chokwe back in 2000–2002 [63] and it was collected in 2017–2018 in the neighboring’s province of Limpopo, South Africa [64]. The detection of this vector could have been the result of misidentifications. In 2018, Erlank et al. [62] showed that An. squamosus, An rufipes and An. pretoriensis showed amplicons similar to An. gambiae s.s. when the PCR protocol for identification of species of the An. gambiae complex was applied to them. Hence, if mosquitoes of any of these three species were wrongly identified as belonging to the An. gambiae complex, this could have led into false identification of An. gambiae s.s. The current presence of An. gambiae s.s. in southern Mozambique and its potential role in transmission should be further investigated. It is recommended that if this species is reported in southern Africa in the future, the species identification be confirmed with sequencing of the ITS2 or COI regions.

Although accurate estimates of sporozoite rates or entomological inoculation rates could not be established due to i) the low numbers of mosquitoes collected, and ii) the fact that the parasite reservoir was targeted by several interventions, An. arabiensis was most likely responsible for most of the residual malaria transmission during the Magude project. This is because it was the most abundant mosquito species, accounted for the highest proportion of human exposure to bites [16], it presented one of the strongest statistical associations with malaria incidence and it was the only vector found carrying sporozoites after the first IRS campaign, except for one An. merus. Anopheles funestus s.s. and An. merus may have played a minor but continuous role, based on their significant association with malaria incidence and consistent presence through the intervention period albeit with a low relative abundance. An. squamosus may have played a minor but more erratic role. While it was significantly associated with malaria incidence, it was only present in the months surrounding the observed high malaria incidence peak. An. parensis may have played a role sporadically. Despite its significant association with malaria incidence, it disappeared after the implementation of the first IRS campaign and practically didn’t appear again until July 2017. If other anopheline species contributed to residual malaria transmission, such as the two other vectors that were significantly associated with malaria incidence (An. rufipes and An. quadriannulatus) or those that are known vectors elsewhere, they likely also played a very minor role due to their very low relative abundances and densities. Evaluating the exact role of each species will be critical to guide vector control strategies but is cumbersome in a project like the Magude project due to the high pressure exerted on both vector (LLINs, IRS) and parasite populations (MDAs). The relative importance of each species should be ideally assessed in similar nearby areas that are not subjected to intense control interventions.

The project’s interventions managed to significantly reduce sporozoite rates from 3.7% before the first IRS campaign, to 0.1%, after the first MDA campaign before the second IRS campaign started, but sporozoite rates later increased to 1.1% after the MDA rounds in year 2. These patterns are in line with the decrease in malaria prevalence observed through the cross-sectional prevalence surveys conducted in May 2015 and May 2016, which revealed that malaria prevalence went from 9.1% to 1.5% after the implementation of the first IRS campaign and the two first MDA rounds, but later increased to 2.6% in May 2017 [12]. The fluctuation in sporozoite rates and malaria prevalence shows that, although the intervention package managed to reduce transmission to very low levels during the first year, it was not further reduced after the second year despite a similar coverage of all interventions. Only LLINs may have been less effective around that time (approx. two years after the mass campaign in 2014) as their integrity and insecticide bio-efficacy is known to decrease over time [65,66]. This, alongside the heavy rains in 2017, may be a reason for the slightly increased incidence rates observed that year.

IRS was likely very effective at controlling An. parensis and An. funestus s.s. because they were susceptible to the insecticides use in IRS (DDT and pirimiphos-methyl) [14] and the numbers in our CDC light trap collections decreased dramatically after the implementation of the first IRS campaign. An. parensis practically disappeared for approx. 2 years, whereas An. funestus s.s. remained present but in very low densities. In addition, very few (insufficient numbers to conduct insecticide susceptibility assays) blood-fed An. funestus s.s. and its sibling species were found resting indoors after the first IRS campaign, whereas mosquitoes of that species group were abundant prior to the first round of IRS. Our analysis suggests that An. arabiensis was not affected as much by IRS compared to An. parensis and An. funestus s.s., even though An. arabiensis was also susceptible to the IRS insecticides [14]. This is because the relative abundance of An. arabiensis increased after IRS, it remained the predominant species throughout the project, its population managed to increase rapidly every year when rains increased and as the effect of IRS waned off, and we continued to find large enough numbers of this species indoors for our insecticide susceptibility bioassays throughout the project. Elsewhere it has been observed that An. arabiensis is much less affected by IRS with pirimiphos-methyl than other important vector species, such as An. funestus s.s. [67], possibly due to the fact that a proportion of its population rests outdoors after feeding avoiding contact with sprayed wall surfaces [68-70]. Actually, following intense implementation of IRS in southern Africa over the last eight decades, An. arabiensis has become the main malaria vector in South Africa and eSwatini [4,71]. Nonetheless, IRS presumably limited its population growth, as densities are expected to increase after the rains (because this vector readily breeds in small, temporary and shallow pools [72]), yet the population remained at similar densities after the heavy rains in 2017, compared to the drier preceding year. The effect of IRS on An. merus cannot be analyzed, as no mosquitoes of this species were collected before the first IRS campaign. However, studies conducted in Mozambique during the Global Malaria Eradication Campaign (1960–1969) showed that An. merus entered houses to feed, but rested outdoors, thereby avoiding contact with IRS [7]. A similar behavior is expected in Magude district, as very few An. merus (insufficient numbers for the insecticide susceptibility assays) were collected during indoor manual collections. The effect of IRS on An. squamosus cannot be properly examined due to their very low numbers, but the limited data suggest that it is likely little affected by IRS. The relative proportion of An. squamosus increased after the first IRS campaign and, although it maintained a very low relative abundance or was not detected during most months of the project, its population increased rapidly in February 2017, when the efficacy of IRS was waning but estimated to still be around 80%.

Looking at all species together, around 50% of mosquitoes were collected between February and April and Anopheles densities increased rapidly from January onwards, approximately five months after the start of each IRS campaign. This coincides with the duration of IRS’s optimal residual efficacy, which was estimated to be between 3.5 months and 5.5 months, when considering mosquito mortality 24h post-exposure or delayed mortalities, respectively. It seems that IRS was effective at controlling mosquitoes during the initial months after implementation (and during the start of the rainy season), but that its residual efficacy did not prevent the growth of vector populations during the entire malaria transmission season [14]. A second round of IRS or using a product with longer residual efficacy will be needed to effectively cover the high transmission season. However, since a second round of IRS is unlikely to be operationally feasible, given costs and the logistical challenges during a rainy season, products with a longer lasting residual efficacy will be preferred.

LLINs likely provided good protection against An. arabiensis (both in terms of killing susceptible mosquitoes, and reducing human-vector contact), as this species was largely endophagic, active when people were already sleeping [16] and susceptible to pyrethroids [14]. Our previous analysis showed that LLINs prevented 41.8% of human exposure to An. arabiensis and could have prevented 67.4% if all residents would have used a LLIN to sleep. LLINs likely provided lesser protection against An. funestus s.s. and An. parensis, compared to An. arabiensis. These two vector species were resistant to pyrethroids [14] and nets could only prevent 21.9% and 13.9% of exposure to host seeking mosquitoes of these species, respectively [16]. Although we could not evaluate the level of pyrethroid resistance, if any, in An. squamosus and An. merus, LLINs prevented less than half of the human exposure to these vectors (32.0% and 45.4%, respectively). As shown previously, LLIN use was suboptimal, especially during the low transmission season [15]. LLIN personal protection against host seeking mosquitoes of species that were P. falciparum positive during the project could have increased by between 18% to 30%, depending on the vector species, if all residents would have used the net to sleep [16]. We therefore conclude that LLINs did not achieve their full protection potential during the Magude project, but that they did complement the protection provided by IRS, mainly by providing a certain level of protection against An. arabiensis, An. merus and An. squamosus. Improving LLIN use will be especially important for controlling the An. arabiensis and An. merus that survive IRS, as they mainly bite indoors while people are in bed.

The results presented here, combined with previous analyses [16] show that there is a proportion of vectors that either seek their host outdoors, or seek their host indoors at times when people are not yet under a net. In addition, some species could not be found resting indoors whereas others were still found resting indoors after the implementation of IRS. This highlights that there are gaps in protection by both LLINs and IRS.

The complex changes in vector composition over time, the diversity in feeding behaviors, combined with suboptimal levels of LLIN use and short IRS realized efficacy, suggest that ITNs are IRS alone would not have been sufficient to fully control local vector populations. Additional interventions and stronger community engagement campaigns would likely have helped to achieve optimal vector control. House screening, eave tubes and lethal lures [73] could tackle those vectors that seek their hosts indoors before people are under a net. In addition, larviciding or other forms of environmental management could reduce local vector populations, including those that are not affected by current adult vector control interventions. Future interventions to kill or prevent outdoor host-seeking vectors from finding their host, including repel and lure devices and attractive targeted sugar baits, may become suitable options to increase human protection outdoors during elimination efforts in southern Mozambique.

The present study has several limitations that may have hampered our ability to fully understand the vectors that sustained malaria transmission during the project, and the impact of the interventions on those vectors. First, P. falciparum sporozoite rates were not assessed for all mosquito species collected, and, consequently, other species may have contributed to sustaining local malaria transmission. Second, the ELISA protocol that was followed has been observed to yield false positive results in some mosquito species, especially those that exhibit zoophilic tendencies [74]. Although we ran a confirmatory test for each positive specimen, since 1) An. arabiensis, An. parensis, An. merus and An. squamosus are all known to be partially or opportunistically zoophilic and 2) the presence of sporozoites was not confirmed by PCR, some species may have been falsely identified as carrying sporozoites. Third, the low number of mosquitoes one collects during an elimination campaign due to the high pressure on the vectors (LLINs and IRS), combined with high pressure on the parasite reservoirs (MDA), hamper the possibility to reliably estimate sporozoite rates. Fourth, the human blood index was not determined for any of the vector species identified as host-seeking mosquitoes, which were mostly unfed. Understanding this is important, as the sporozoite rate combined with the preference to feed on humans and overall human biting rates determine the entomological inoculation rate, the gold standard metric to understand the relative importance of each vector species in malaria transmission. Fifth, since our CDC light traps started collecting mosquitoes several hours before people went to bed (i.e. while there was no human bait under the net next to the trap), the number of host-seeking mosquitoes per person reported for the time interval 18:00–20:00 may be an underestimation. Sixth, the residual efficacy of DDT was not measured after the 2015 campaign, which affects the accuracy of our IRS residual efficacy estimates for this campaign. Finally, due to the absence of a sufficiently long baseline of mosquito collections prior to the implementation of the first interventions, and/or the lack of a control district monitored simultaneously, the magnitude of the effect of vector control interventions on local vector population densities could not be quantified through robust statistical models. Future projects should include a baseline of at least one year, or a representative control district, to properly and quantitatively evaluate the effectiveness of the interventions on local vector populations. This is critical to identifying and subsequently addressing the gaps in the protection offered.

Conclusions

Anopheles arabiensis was the main vector species during the Magude project. Anopheles merus, An. parensis, An. funestus s.s. and An. squamosus likely played a secondary and minor role. Further investigation into the possible role of other collected vector species (i.e. Anopheles species that are known secondary vectors elsewhere in Africa) is needed. The deployment of MDA and IRS, in addition to LLINs, successfully reduced vector sporozoite rates during the first year of implementation but no further reduction in sporozoite rates was observed despite similar intervention coverages in the second year. IRS most likely controlled An. funestus s.s. and An. parensis and was also effective at controlling An. arabiensis, but its effect was limited by its short residual efficacy that went below optimal levels (80% mosquito mortality in WHO cone bioassays) before the end of the high transmission season. Its effect on An. merus and An. squamosus could not be assessed due to low mosquito numbers but should be investigated as these species were incriminated as malaria vectors in Magude. LLINs complemented the protection provided by IRS, especially by providing protection against the indoor and late-evening host-seeking An. arabiensis and An. merus vectors. Therefore, the combination of IRS and LLIN is likely to have brought added value to the control of malaria vectors during the Magude project, compared to the implementation of one intervention alone. However, the effect of LLINs was compromised by their suboptimal use and the pyrethroid resistance in An. funestus s.s. and An. parensis populations. Future progress towards malaria elimination could be made by increasing LLIN use and distributing dual active ingredient LLINs to prevent transmission by the An. arabiensis and An. merus that survive IRS, by sustaining IRS to maintain control of An. funestus s.s. and An. parensis, and by using longer lasting residual insecticides for IRS to prevent the increase in vector densities observed at the end of the rainy season. Additional interventions will nevertheless be needed to tackle those vectors that transmit malaria outdoors and early indoors if we are to close the gap towards malaria elimination.

Exclusions criteria for mosquito collections.

(DOCX)

Click here for additional data file.

Sinusoidal function used to simulate ITN use based on observed values of ITN use.

(DOCX)

Click here for additional data file.

Detailed Sporozoite detection results.

(DOCX)

Click here for additional data file.

Detailed model results.

(DOCX)

Click here for additional data file.

(XLSX)

Click here for additional data file.

22 Jul 2022

12 Aug 2022

Dear Editor,

We like to thank the two reviewers for their constructive comments and excellent attention to detail. We were very pleased to see that the reviewers had a very good understanding of the topic, which lead to the improvement of our manuscript. We have largely accepted the recommendations and have answered all questions below (in blue bold text).

We acknowledge that we cite three of our other papers that are currently in press (PONE-D-21-29269R2), in revision (PONE-D-21-40329), and in review (PONE-D-22-14587). We could work with PLOS ONE and e.g., publish those manuscripts on BioRXiv so we can properly cite them. But your suggestions are much appreciated.

Kind regards,

Lucia Fernandez Montoya, on behalf of all co-authors

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

We feel our manuscript now adheres to PLOS ONE’s style requirements but do let us know if more changes are required.

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

"We acknowledge support from the Spanish Ministry of Science and Innovation through the “Centro de Excelencia Severo Ochoa 2019-2023” Program (CEX2018-000806-S), and support from the Generalitat de Catalunya through the CERCA Program. CISM is supported by the Government of Mozambique and the Spanish Agency for International Development (AECID). LLK is supported by a DST/NRF South African Research Chairs Initiative Grant (UID 64763). ESS is funded by a UKRI Future Leaders Fellowship from the Medical Research Council (MR/T041986/1) and acknowledges funding from the MRC Centre for Global Infectious Disease Analysis (reference MR/R015600/1), jointly funded by the UK Medical Research Council (MRC) and the UK Foreign, Commonwealth & Development Office (FCDO), under the MRC/FCDO Concordat agreement and is also part of the EDCTP2 programme supported by the European Union; and acknowledges funding by Community Jameel."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"This study was supported by the Bill and Melinda Gates Foundation and Obra Social “la Caixa” Partnership for the Elimination of Malaria in Southern Mozambique (INV-0084830). LLK is supported by a DST/NRF South African Research Chairs Initiative Grant (UID 64763). ESS is funded by a UKRI Future Leaders Fellowship from the Medical Research Council (MR/T041986/1) and acknowledges funding from the MRC Centre for Global Infectious Disease Analysis (reference MR/R015600/1). Abt Associates Inc. provided support in the form of salaries for author NC, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. This does not alter our adherence to PLOS ONE policies on sharing data and materials."

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

We have edited the Acknowledgment statement to remove any reference to funding sources.

Please update our funding statement as follows:

"This study was supported by the Bill and Melinda Gates Foundation and Obra Social “la Caixa” Partnership for the Elimination of Malaria in Southern Mozambique (INV-008483). LLK is supported by a DST/NRF South African Research Chairs Initiative Grant (UID 64763). ESS is funded by a UKRI Future Leaders Fellowship from the Medical Research Council (MR/T041986/1) and acknowledges funding from the MRC Centre for Global Infectious Disease Analysis (reference MR/R015600/1), jointly funded by the UK Medical Research Council (MRC) and the UK Foreign, Commonwealth & Development Office (FCDO), under the MRC/FCDO Concordat agreement and is also part of the EDCTP2 programme supported by the European Union; and acknowledges funding by Community Jameel. Abt. Associates Inc. provided support in the form of salaries for author NC, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. This does not alter our adherence to PLOS ONE policies on sharing data and materials. The specific role of this author is articulated in the ‘author contributions’ section."

3. Please ensure that you refer to Figure 4 in your text as, if accepted, production will need this reference to link the reader to the figure.

We have inserted a reference to this figure at the end of the results section “Association between host-seeking Anopheles per person and residual malaria incidence”, line 554.

4. Please upload a copy of Figure 6, to which you refer in your text on page 20. If the figure is no longer to be included as part of the submission please remove all reference to it within the text.

There is actually no Figure 6, and we mistakenly wrote Fig 6 instead of Fig 4 at the end of the results section “Association between host-seeking Anopheles per person and residual malaria incidence”. We have corrected this.

5. We note that Figure 1 in your submission contain [map/satellite] images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

We created the map ourselves using administrative boundaries from the Humanitarian Data Exchange under license "Creative commons attribution for Intergovernmental organizations. (CC-BY-IGO). https://data.humdata.org/faqs/licenses.

We have indicated this in the caption of Fig 1.

a. You may seek permission from the original copyright holder of Figure 1 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an ""Other"" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

6. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

We have revised all references to ensure that they meet PLOS ONE criteria. In addition, we have removed two references (former references number 11 and 34) and added one (number 23) to respond to reviewer’s comments. We cite three of our other papers that are currently in press (PONE-D-21-29269R2), in revision (PONE-D-21-40329), and in review (PONE-D-22-14587). We would like to work with PLOS ONE to find a solution so that we can cite them properly (e.g publish those manuscripts on BioRXiv). Your suggestions will be much appreciated.

***

Reviewers' comments:

Reviewer #1: This paper presents interesting results that are relevant for many malaria control programmes in southern/eastern Africa. My only major problem was that it seems very detailed and I wondered if it could be more focused, for example presenting the data in tables instead of in the text as is currently the case in a lot of the Results section. Thanks a lot for this suggestion. At the time of writing the manuscript, we tried to present the results of the first four paragraph of section “Anopheline species composition and densities over time” in different tables and graphs. Most of the co-authors found Fig 2 easier to interpret and more visually appealing than any other tables. Hence, we prefer to keep the results as they are now.

The following are minor corrections.

Abstract

Line 57 – “b,y” should be “by”, This is corrected.

Introduction

Line 102 – Ref #11 is a duplicate of ref #8. Delete #11. We have deleted the duplicate reference.

Results

Line 386 – delete “with” at the very end of this line. Done.

Lines 392-393 – why could these mosquitoes not be identified morphologically? Were they too damaged? This is because they were either too damaged or because the microscopists could not find a matching species in the used taxonomy keys. We have added this sentence in line 393 of the results section.

Line 395 – replace “were” with “to the” Corrected.

Line 510 – the number of An. garnhami is missing. Also, check the spelling of “garnhami” throughout. The number of An. garnhami tested for P. falciparum sporozoites is 1. We have added it to line 510. We have also checked the spelling of “garnhami” everywhere and corrected as needed.

Line 511 – it should be An. listeri (no ‘a’ at the end). Also, as far as I know, neither An. multicolor or An. salbaii occur so far south. I would delete all mention of both of these species. We have corrected An. listeri throughout the text. We prefer to keep mention to An. multicolor and An. salbaii in the manuscript, as An. multicolor and An. salbaii have been detected in equatorial and eastern Africa and we cannot rule out the possibility they inhabit southern Mozambique. In the discussion we are nuancing this by mentioning that the detection of these other Anopheles species could be an artifact of wrong morphological identification and that molecular validation will be required.

Line 513-514 – this is basically a repetition of what is said in the previous paragraph. The previous paragraph provides the total numbers and percentages of Pf sporozoite positive mosquitoes while this particular paragraph provides the percentage for indoor collected mosquitoes only. We reckon that since most of the mosquitos were collected indoors, the percentages are very similar, but we want to provide the figures for both indoors and outdoors separately, so we have kept both paragraphs.

Lines 548-549 – An. quadriannulatus is mentioned twice and no mention is made of An. rufipes. Thanks for spotting this. An. rufipes is associated at 10% significance. We have corrected this.

Line 553 – shouldn’t this be “Fig. 4”? Indeed, this should be Fig. 4, we have corrected it.

Discussion

Line 603 – reference #34 is about work in Benin, not South Africa. Thanks for spotting this mistake. We have removed reference #34

References

Ref #

11 – This is a repeat of ref 8. Delete. We have deleted reference 11 and corrected reference numbers along the manuscript accordingly.

15-17 – There is no journal information. I assume that these three papers have been submitted to PLoS One but not yet published? If so, then include “(submitted)” at the end of each article. We cite three of our other papers that are currently in press, in revision and in review respectively. We will work with PLOS ONE to find the best solution to cite these papers properly (e.g., publishing those manuscripts on BioRXiv).

18 – abbreviate the journal name as you have done for all the other refs. Corrected.

21 – it should be “(Diptera: Culicidae)”. Corrected.

22 – it should be “Brogdon”. Also, why are the first names given for the first two authors but not for Collins? You do not give first names for any other article. This was a mistake. We have corrected it to provide only surname and name’s initials.

39,40,46,55 – abbreviate the journal name. We have revised all reference to make sure that the journal names are abbreviated.

41 – is this author’s name correct? Corrected.

48, line 976 – give both ‘ anopheles’ a capital A. Corrected.

Reviewer #2: Dear Authors, please consider the following suggested edits:

Page 12, line 133: You say: “…a rainy season expanding from October to March…” I suggest replace “expanding” with “extending”. Thanks a lot for this suggestion. We have replaced “expanding” for “extending”.

Page 14, Line 200: You say: “…The presence of sporozoites in individual mosquito samples was analyzed through screening enzyme-linked immunosorbent assays (ELISA) conducted on mosquitoes’ grinded head and thorax…”. It would be useful to know which parasites you assayed for, whether it was only for P falciparum or if you also tested for P vivax. Vivax is known to occur in the region and because the focus is usually on testing for falciparum, little appears to be known regarding vivax prevalence. So just clarify which parasites you tested for and why you excluded others. We tested for P. falciparum only because we were expecting very few sporozoite positive mosquitoes (given the elimination context) and P. falciparum accounts for 90% of all infections in Mozambique and almost all in the district neighboring the study area. We have edited the sentence to add the parasite species that we looked for, i.e. “… The presence of P. falciparum sporozoites in individual mosquito samples was analyzed through screening enzyme-linked immunosorbent assays (ELISA) conducted on mosquitoes’ grinded head and thorax ...” and the justification for testing only this species, i.e. “The presence of sporozoites of other Plasmodium species was not tested because P. falciparum is known to account for over 90% of all diagnosed malaria infections in Mozambique [23] and for almost all in the neighboring district of Manhiça [24], and because very low positivity rates were expected given the elimination context.”

Page 16, Line 266 – 275: You say “…The impact of both MDA and IRS on sporozoite rates can potentially already be observed two to three weeks after implementation. MDA with DHAp immediately eliminates gametocides from humans, which prevents feeding mosquitoes from ingesting gametocytes and becoming infective. The tim….”. This text is under “Data Analysis” and strictly speaking is probably more appropriate under Results or Discussion, but is probably not too serious. Just be more careful to appropriately allocate material more appropriately in future. I can see that you were intending to create context for your analysis but this could also be done under Results or Discussion.

Thanks for this suggestion. Indeed, we want to provide rationale for the time lags used in the analysis. We tried to move it to the result section, but some co-authors have found it confusing to scroll up and down through the manuscript to get this information. Consequently, we have decided to leave it in the data analysis section.

Page 16, Line 270: You say “…migration to mosquito salivary glands can be of two or three weeks depen…”. I suggest delete the “of” before “two or three weeks”. Thanks for the suggestion. We have deleted “of”.

Page 19, Lines 353-356: You say “…We evaluated indoor and outdoor host-seeking times during the project’s intervention period (August 2015 to October 2017) by calculating the number of host-seeking mosquitoes of each species collected per person for each 2 hour time interval (period of rotation of the CDC bottle rotator) from 18:00 to 06:00, b…”. I just want to mention that earlier in the article you state that, indoors, the CDC traps were hung 1,5 metres above ground at the foot end of the beds, and therefore the primary attractant was human odour and thus the traps and numbers of mosquitoes caught only really become effective and valid once a person has gone to bed, presumably around 9pm. So one needs to exercise caution in apportioning host-seeking times based purely on trap catches when the “bait” is not in use in early evening. Thanks so much for raising this point. It is a very important observation. We have added it to the limitations section of the discussion.

Page 20, Lines 372-374: You say “…More details on the ethical considerations of the population census, household surveys, cross-sectional 374 prevalence surveys and MDAs are provided elsewhere…”. Where is “elsewhere? I don’t really care to be honest, and neither will most readers of the paper, but you cannot make a statement like this without giving some indication where the details are available; it is like saying some piece of information has been published “elsewhere” but you do not give the reference. Thanks for spotting this missing reference. We have added reference 12 and 13 here, which provide all details on the census, household surveys, cross-sectional prevalence surveys and MDAs.

Page 20, Lines 394-395: You say “…and 87.3% 395 (n=332) of mosquitoes belonging were An. funestus group.”. Fix the language please, by deleting “were” and replacing with “to the”. Corrected.

Page 24, Line 511: You say “…None of the An. listeria,…”, it should be An listeri. Corrected.

Page 29, Line 642: Please italicize An. gambiae. Corrected.

Page 31, Line 696: You say “…we continued to find large enough number of this species…” Either say “a large enough number” or “large enough numbers”. Corrected.

Page 36. References 8 and 11 are repetitions of the same publication, you need to delete one. Corrected.

Page 37. References 15, 16 and 17 have no year of publication or page numbers, please correct.

We cite three of our other papers that are currently in press, in revision and in review respectively. We will work with PLOS ONE to find the best solution to cite these papers properly e.g., publishing those manuscripts on BioRXiv so we can properly cite them.

Page 40. Reference 48. The genus name needs to be capitalized. Corrected.

Pge 45, Figure 2. “Indetermined” should be “Undetermined”. Same with Figures 3 and 5.We have corrected all three figures.

________________________________________

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

24 Aug 2022

The mosquito vectors that sustained malaria transmission during the Magude project despite the combined deployment of indoor residual spraying, insecticide-treated nets and mass-drug administration.

PONE-D-22-18501R1

Dear Dr. Fernandez Montoya,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Basil Brooke, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

31 Aug 2022

PONE-D-22-18501R1

The mosquito vectors that sustained malaria transmission during the Magude project despite the combined deployment of indoor residual spraying, insecticide-treated nets and mass-drug administration.

Dear Dr. Fernandez Montoya:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr Basil Brooke

Academic Editor

PLOS ONE