Evaluation of long-lasting microbial larvicide for malaria vector control in Kenya

The present study examined a United States Environmental Protection Agency approved, slow-release FourStar™ briquets 180-days formulation manufactured by Adapco Inc [21]. The 30 g formulation consists of 6% by weight of Bacillus sphaericus (Bs) Serotype H5a5b strain 2362, 1% by weight of Bacillus thuringiensis sub-species israelensis, (Bti) strain BMP 144 and other ingredients that make the formulation to release slowly the bacterial toxins.

To test the efficacy and effective duration of the slow-release microbial larvicide formulation in killing malaria mosquitoes under semi-natural conditions, the 30-g FourStar™ briquets 180-days formulation was placed in 560 L of rainwater in a 1000-L container. This amount of water was calculated based on the recommended dose of one FourStar™ briquet in a 9.3-sq metres of surface area. The briquet was dissolved, stored in the container and covered with insect-proof net, as the stock solution. Three stock solutions were prepared. Microcosms were prepared in a plastic basin (30 cm diameter and 20 cm deep) using 2 kg soil from the rural community in Kisian village, western Kenya and 2 L of stock solution. Control microcosms used 2 kg soil and rainwater without any larvicide.

Sixty-second-instar field collected An. gambiae larvae were introduced to each microcosm basin each month for a period of 6 months. At each time point, ten microcosms were used, including five basins with microbial larvicide and five control basins without any larvicide. Water in basins was replenished from the stock solution or rainwater to compensate for water loss due to evaporation. The number of surviving larvae and pupae were monitored and recorded daily. Pupae productivity was used as a proxy measure for adult mosquito emergence, and it was used as the key measurement of the effectiveness of microbial larvicide here [22].

After the promising result in the semi-natural conditions, further experiments were conducted to test the efficacy and effective duration under the field conditions. Eighty An. gambiae larval habitats were selected in a highland site of Iguhu and 50 in a lowland site of Kisian in western Kenya to determine their stability and productivity, for a period of 5 months from September 2009 to January 2010. Habitat stability was the time period during which larval habitats remained aquatic to support larval development. Habitat productivity was the number of pupae produced in a habitat. These habitats were monitored every 2 weeks for the entire duration of the study. Sampling was done to determine the larval and pupal density per habitat. After the period of monitoring, stable and productive habitats were selected for either treatment or no treatment with the briquet formulation. Fifty-six habitats in the highland site of Iguhu and 23 habitats in the lowland site of Kisian were stable and productive, so they were randomized into intervention and non-intervention habitats, stratified by highland and lowland sites. There were 28 habitats in the intervention arm, and 28 in the control arm in the highland site, and 13 in the intervention arm and ten in the control arm in the lowland site (Fig. 1). Because of an extremely large amount of rainfall due to El Nino event in western Kenya from end of January to April 2010, the FourStar™ briquets 180-days microbial larvicide was applied in larval habitats of intervention arm in May 2010, using the recommended dosage of one 30-g briquet per 9.3-m² of surface area. Habitats in the control arm were not treated by microbial larvicide. All breeding habitats were monitored monthly from June to October 2010 to determine pupal productivity. Rainfall data were obtained from local meteorological stations for both sites.

Six 2 × 2 km² sites from lowland, highland fringe and highland regions in western Kenya were used to determine the effect of slow-release microbial larvicide on adult malaria vector abundance reduction (Fig. 2). Two sites were located in malaria-endemic lowland (Kisian and Kombewa, 1150–1200 m above sea level) [3]. Two sites were in malaria-endemic highland fringe region (Esaba and Musilongo, elevation 1380–1480 m), and two sites (Eshibinga and Katsombero) were located in malaria epidemic-prone highland (elevation 1500–1650 m) [3]. Within each region, one site was randomly assigned as an intervention site and the other as a no-intervention control site by coin toss. A baseline adult mosquito density survey was conducted in all sites using the pyrethrum spray catch (PSC) method in 60 randomly selected houses in January 2011. All larval habitats within the study sites were sampled for larval distribution and abundance at baseline. All larval habitats within the intervention sites were treated with FourStar™ briquets 180-days formulation. Briquets were placed with respect to larval habitat sizes, according to manufacturer’s recommendations for each briquet. No habitats in the control sites were treated by the larvicide. Larvicide was applied in February 2011 during the dry season, and evaluation of the intervention was conducted from March to June 2011. The primary endpoint was adult mosquito abundance. Adult mosquito population abundance was monitored weekly in 20 randomly selected houses within each site using the PSC method. Because of residual mosquito killing and repellent effect of pyrethrum used in PSC collection, the houses were rotated on weekly basis, and thus in each site 60 houses were sampled. To minimize the migration effects of mosquitoes from neighbouring larval habitats outside the 2 × 2 km² intervention area, all 60 houses selected for evaluation were located within 500 m from the edge of the study area.

To test the hypothesis that the long-lasting microbial larvicides can reduce the abundance of adult vectors inside houses and outdoor, a randomized, paired-cluster design was undertaken in 2012 with six 2 × 2 km² sites selected in Katsombero located in the highland area of western Kenya where breeding habitats were generally clustered around streams and valley bottom [23, 24] (Fig. 3). The six selected sites were randomized into intervention and control sites. Larval and adult mosquito populations were monitored in these sites at baseline in January 2012. Sixty houses were randomly selected in the sites and indoor and outdoor biting mosquito density was estimated using CDC light traps. In the intervention sites breeding habitats were then treated with the Bs/Bti briquet formulation whilst habitats in the control sites received no treatment. The briquets were applied in all larval breeding sites at the middle of the dry season (February) and just before the rainy season (March) to reduce the dilution effect of rainfall. Mosquito abundance indoor and outdoor was monitored using the CDC light traps. Each week, 20 of the 60 houses selected were sampled in rotation. CDC light traps were hanged indoors and outdoors in five houses per night for four nights per week. Because our objective was to compare the extent of indoor and outdoor biting in areas treated and not treated with the long-lasting larvicides, the vector abundance monitoring method with the use of CDC light traps was appropriate.

In the experiments to test the efficacy and duration of efficacy of the briquets in mosquito control under semi-natural environment, t test was done to determine the statistical differences in the mean number of pupae emerging from the treated and untreated microcosms. For experiments to determine the efficacy of the briquets in pupal productivity control in natural habitats, analysis of variance (ANOVA) with repeated measures was done to determine the statistical differences between the intervention and control habitats. For the studies to determine the effect of the briquet on mosquito densities inside and outside houses, densities of mosquito per house per night was calculated in the intervention and control sites, and ANOVA with repeated measures was used to determine the statistical significance. Numbers of An. gambiae s.l. and An. funestus mosquitoes were pooled due to very low An. funestus abundance.

In the semi-natural microcosm experiment, no pupae emerged from any treated microcosms during the first 3 months, whilst a pupation rate of 89.4% was found in the untreated control microcosms. In months 4, 5 and 6, pupation rate was 2.0, 5.7 and 13.0% in the treated microcosms, respectively, whilst that of the untreated microcosms were 92.7, 89.3 and 89.7%, respectively (t test, P < 0.0001 for all comparisons) (Fig. 4). This result suggests that the FourStar™ briquets 180-days formulation reduced pupal productivity of An. gambiae mosquitoes in semi-natural microcosms for up to 6 months by at least 85.5%.

During the study period the average monthly rainfall was 139.2 mm in the highland site and 17.1 mm in the lowland sites. Pre-intervention rainfall was 150.7 and 17.2 mm in the highland and lowland, respectively, whilst post intervention average monthly rainfall was 147.4 and 16.7 mm (Fig. 5a, b).

In the highland site, pupal productivity between intervention and control habitats was similar during the 5-month pre-intervention period (4.20 vs 4.24 pupae per dip, F_1,8 = 0.01, P > 0.05; Fig. 5a). In the first 2 months after the larviciding application, no pupae were detected from any of the treated habitats in the highland areas whereas the average productivity of control habitats was 4.8 pupae per dip (Fig. 5a). From months 3 to 5, the average pupae density in the intervention habitats was 0.73 whilst that in the control habitats was 3.9, leading to a relative reduction of 87.4% in pupal productivity (F_1,8 = 110.79, P < 0.0001).

Similar to findings in the highland sites, the pupae productivity in the habitats in the lowland site in the intervention and control arm were similar (4.60 vs 4.36 pupae per dip, F_1,8 = 1.07, P > 0.05). Pupal productivity was 0 during the first three months after microbial larvicide application, and 4.6 in the control habitats. In months 4 and 5, pupal productivity was 0.55 in the intervention habitats whereas 4.2 pupae were found in the control habitats, a reduction of 95.4% (F_1,8 = 276.5, P < 0.0001; Fig. 5b).

Prior to larviciding, anopheline adult mosquito density collected indoors by PSC was comparable for each of the three areas. Generally, the impact of larviciding on adult mosquitoes was observed two to three weeks after larviciding. The application of the FourStar™ briquets formulation caused a reduction of 60–85% in the density of adult malaria vectors in the next 14 weeks in the lowland (F_1,22 = 9.64, P < 0.005; Fig. 6a). Larviciding impact on mosquito density started to decay 9 weeks after larviciding application. In the highland fringe region, application of FourStar™ briquets reduced adult mosquito density by 65–80% in the following 14 weeks (F_1,22 = 15.53, P < 0.001), and the effect of larvicide started to decay 10 weeks after larviciding (Fig. 6b). In the highland, larvicide exhibited the longest period of effectiveness, a reduction of 76–90% in adult density was found 14 weeks after larviciding (F_1,22 = 76.92, P < 0.0001; Fig. 6c). A total of 1812 mosquitoes were collected for this experiment, comprising of 88.9% (1610) An. gambiae s.l., 4.9% (89) An. funestus and 6.2% (113) Culex spp.

Pre-intervention vector density within the intervention sites was 0.28 female mosquitoes per house per night compared to 0.21 in the control sites. During the first 3 months post-application of the briquets, the density of adult malaria vectors inside residential houses within the intervention sites was decreased by 76–82% compared to the untreated control sites (0.09 vs 0.33; F_1,32 = 131.87, P < 0.0001; Fig. 7a). The density of malaria vectors caught outdoors was reduced by 67–75% (0.08 vs 0.18; F_1,32 = 76.03, P < 0.0001; Fig. 7b). Thus, the slow-release microbial larviciding significantly reduced indoor- and outdoor-biting mosquitoes for at least 3 months. A total of 3325 mosquitoes were sampled for this study, comprising of 84.8% (2821) An. gambiae s.l., 5.4% (179) An. funestus with the rest being Culex spp. 9.7% (325).