Malaria vector species composition and entomological indices following several years of indoor residual spraying in regions bordering Lake Victoria, Tanzania

Background Vector control through long lasting insecticidal nets and focal indoor residual spraying (IRS) is a major component of the Tanzania national malaria control strategy. In mainland Tanzania, IRS has been conducted annually around Lake Victoria basin since 2007. Due to pyrethroid resistance in malaria vectors, use of pyrethroids for IRS was phased out and from 2014 to 2017 pirimiphos-methyl (Actellic 300CS) was sprayed in regions of Kagera, Geita, Mwanza and Mara. Methods WHO Cone bioassays were conducted monthly on interior house walls to determine residual efficacy of pirimiphos-methyl CS. Indoor CDC light traps with or without bottle rotator were hung next to protected sleepers indoors and also set outdoors (un-baited) as a proxy measure for indoor and outdoor biting rate and time of biting. A sub-sample of Anopheles were tested by PCR to determine species identity and ELISA for sporozoite rate. sites compared to 0.8% (95% CI: 0.5-1.3) in sprayed sites (p=0.003). Sporozoite rates also appeared to be lower for An. funestus collected in sprayed sites. the Mwanza at Cone tests on conducted effect of Actellic 300CS in each When mortality in was between and results using Abbot’s formula, and and repeated(15). cone


Background
In sub-Saharan Africa, recent gains in malaria control have been mostly accomplished through a substantial boost in vector control using long-lasting insecticidal nets (LLINs) and indoor residual spraying (IRS). These tools have significantly contributed to a 50% reduction of Plasmodium falciparum infection prevalence in endemic countries between 2000 and 2015 (1). IRS has been reported to successfully reduce malaria prevalence and incidence in several African countries in the past decade(2-4). In mainland Tanzania Geita districts. Pyrethroid resistance was documented by Kisinza et al (2017) in all districts that were tested in 2015 near Lake Victoria, including Musoma Rural, Magu, and Muleba.
Due to the detection of pyrethroid resistance in malaria vectors, the use of pyrethroids for IRS was gradually phased out in accordance with WHO guidance that pyrethroids should be preserved for LLINs (8). The carbamate insecticide, bendiocarb (Ficam®, 80% WP) was used alongside the pyrethroid deltamethrin K-Othrine® (WG 250) from 2011 to 2013 (5).
From 2014 to 2017, a long-acting organophosphate formulation of pirimiphos-methyl (Actellic® 300CS) was sprayed annually in all targeted areas of the Lake Victoria basin in the regions of Kagera, Geita, Mwanza and Mara.
Despite widespread pyrethroid resistance being detected in malaria vectors throughout Tanzania(9,10), IRS in combination with pyrethroid LLINs has proven effective in mainland Tanzania (5,11) and also in Zanzibar (12). Partly due to vector control, reported malaria deaths in mainland Tanzania reduced by ~32%, from 15,819 in 2010 to 5,045 in 2016(13).
We report results of entomological surveillance covering ten sprayed sites and four unsprayed control sites in the Lake Victoria Basin. The main objective was to evaluate the entomological impact of IRS with pirimiphos-methyl CS against malaria vectors.
Specifically, entomological data was collected to assess the persistence of residual efficacy of pirimiphos-methyl CS on sprayed walls, determine vector species composition, seasonality, feeding behaviour and malaria infectivity.

Study area and duration
Entomological surveillance was conducted in regions around Lake Victoria, north-western Tanzania. For three years between 2015 and 2017, between eight and ten districts of the Lake Victoria basin were sprayed annually with pirimiphos-methyl CS and entomological monitoring was conducted in sprayed and unsprayed sites. A list of districts and annual spray status is presented in Table 1 and Figure 1.

Residual efficacy of Actellic 300CS (pirimiphos-methyl)
Cone bioassays were conducted on interior wall surfaces according to WHO protocols to determine the quality of spray within 14 days of application and the duration of residual efficacy, which was monitored monthly until mortality was lower than 80% for two consecutive months(14). Batches of two to five days-old non-blood-fed female Anopheles gambiae s.s. (Kisumu strain) were tested by exposing them to sprayed surfaces under WHO plastic cones for 30 minutes, after which they were transferred to clean paper cups and kept in a field insectary for recording delayed mortality. An.
gambiae Kisumu strain were known to be fully susceptible to pirimiphos-methyl and were reared in the NIMR Mwanza insectary at 27 ± 1°C, and 60-80% relative humidity before being transported to the field in cool boxes for the assays. Knock-down and mortality were recorded 60 minutes postexposure and after 24-hours holding. Portable untreated surfaces (approximately 30cm by 30cm) were constructed of cement, mud, burnt brick, whitewash and painted substrates and used as negative controls. Cone tests on untreated portable surfaces were conducted outdoors (to avoid the airborne effect of Actellic 300CS indoors) in a shaded area in parallel for each sprayed house. When mortality in negative controls was between 5% and 20%, the results were corrected using Abbot's formula, and those above 20% were discarded and the tests repeated (15). A summary of cone bioassay tests conducted is shown in Table 2 below.  (18). The number of sites, houses used for trapping, duration of sampling and outcomes are presented in Table 3. Rainfall data during the period of monthly entomological monitoring were accessed from an online database system (19).  (20). The person(s) was requested to sleep under an intact untreated mosquito net(s) provided by the project. CDC light traps were set to operate from 18:00 to 06:00. In the morning, captured mosquitoes were transferred into labeled paper cups and taken for preliminary morphological identification in the field office. All mosquitoes from traps were killed before conducting morphological identification and recording results according to species, sex and abdominal status.

CDC light trap with collection bottle rotator (CBR) (2017)
One CDC light trap with automatic collection bottle rotator (CBR -John Hock model 1512) was set indoors and one outdoors at ten randomly selected houses per site for 10 nights per month. CBR traps were set from March to December (10 months) in 2017 and sampling was scheduled on nights near a new moon to minimize the effect of moonlight on the outdoor light-trap collection, and to reduce bias when comparing species distribution across seasons.
An estimate of the presence and period of moonlight was calculated using an online lunar calendar (21). Indoor CBRs were set up in sleeping areas of houses, while outdoor CBRs were set up within a 10-meter radius of the house. Ethical concerns restrict use of human landing collection (HLC) for mosquito collection. Therefore, the CBR trapping was considered a proxy for human landing collection targeting host seeking vector mosquitoes. Indoor and outdoor human-biting rate of Anopheles and time of biting were determined in the selected sentinel sites. All mosquitoes from traps were killed before conducting morphological identification and recording results according to species, sex and abdominal status.

Indoor Prokopack aspirator (2017)
The improved Prokopack aspirator (John Hock model 1419) was used for sampling indoor resting mosquitoes from 10 houses per day over 20 days within each selected sentinel site per month in 2017 (18,22). Aspiration was carried out in the morning between 06:00 and 08:00am and was conducted in all rooms in the house, moving the aspirator across walls, ceiling and near furniture. To standardize the collection, the sampling was conducted for a total of 30 minutes per house, by two assistants working simultaneously in the same house for fifteen minutes each.

Laboratory analysis
All collected samples were identified to species morphologically using the systematic key of Gillies

Residual efficacy of Actellic 300CS (pirimiphos-methyl), 2015-17
Overall densities were recorded between January and February, which is 10-12 months after the previous IRS cycle, by which time insecticide efficacy had decreased substantially.

Biting rate for Anopheles gambiae s.l. using CDC light trap fitted with bottle rotator (CBR)
In there also appeared to be more outdoor biting, but only late at night between 10pm and 3am.
However, it should be noted that the majority of An. gambiae s.l. in all sites were collected later in the evening when the majority of people are likely to be indoors and protected by LLINs. Nevertheless, we observed a greater degree of outdoor biting risk early in the evening in sprayed sites compared to unsprayed sites ( Figure 4).

Indoor resting densities of Anopheles gambiae s.l. in 2017 using Prokopack aspirators
The mean number of An. gambiae s.l. collected by Prokopack aspirator resting indoors, was greater in the 4 unsprayed sites of Biharamulo, Bukombe, Busega and Tarime than in the 4 sprayed sites of Bukoba Rural, Chato, Sengerema and Musoma Rural. In general, the highest peak in resting density was observed between May and August after the long rain season, with Chato and Busega also having a smaller peak in March ( Figure 6). There were no An. gambiae s.l. collected throughout the 2017 collection period in the sprayed site of Musoma Rural ( Figure 5).

Blood meal analysis
A total of 194 Anopheles arabiensis (identified by PCR) that were collected from January to  Table S2).

Discussion
Cone bioassay results following indoor residual spraying (IRS) with Actellic CS show that it provided control of malaria vectors in north-western Tanzania for a mean of 7 months. Seven months residual duration lies in the higher end of performance for this insecticide formulation, considering a range of 2-9 months that was observed in 9 other PMI supported countries (29). Although An. arabiensis in the Lake zone of north-western Tanzania are resistant to pyrethroids, with high intensity resistance present observed in some sites, they were susceptible to pirimiphos-methyl during the study (7).
IRS campaigns were usually conducted in February and March, meaning that protection was provided through the year up to October/November. However, rainfall in north-western Tanzania is bi-modal, with a second peak of An. gambiae s.l. occurring in January and February, which is 10-12 months after the previous IRS cycle, by which time insecticide efficacy had decreased substantially. In response to entomology data from this study and District Health Information System 2 (DHIS2)-derived reports on peak malaria cases, the timing of IRS has since been changed to November in Kagera and Geita Regions in 2018 (30). Spraying towards the end of the year should provide better protection during the two major malaria peaks of December/January and June/July.