Efficacy of local neem extracts for sustainable malaria vector control in an African village

Background Larval control of malaria vectors has been historically successful in reducing malaria transmission, but largely fell out of favour with the introduction of synthetic insecticides and bed nets. However, an integrated approach to malaria control, including larval control methods, continues to be the best chance for success, in view of insecticide resistance, the behavioural adaptation of the vectors to changing environments and the difficulties of reaching the poorest populations most at risk,. Laboratory studies investigating the effects of neem seed (Azadirachta indica) extracts on Anopheles larvae have shown high rates of larval mortality and reductions in adult longevity, as well as low potential for resistance development. Methods This paper describes a method whereby seeds of the neem tree can be used to reduce adult Anopheles gambiae s.l. abundance in a way that is low cost and can be implemented by residents of rural villages in western Niger. The study was conducted in Banizoumbou village, western Niger. Neem seeds were collected from around the village. Dried seeds were ground into a coarse powder, which was then sprinkled onto known Anopheles larvae breeding habitats twice weekly during the rainy season 2007. Adult mosquitoes were captured on a weekly basis in the village and captures compared to those from 2005 and 2006 over the same period. Adult mosquitoes were also captured in a nearby village, Zindarou, as a control data set and compared to those from Banizoumbou. Results It was found that twice-weekly applications of the powder to known breeding habitats of Anopheles larvae in 2007 resulted in 49% fewer adult female Anopheles gambiae s.l. mosquitoes in Banizoumbou, compared with previous captures under similar environmental conditions and with similar habitat characteristics in 2005 and 2006. The productivity of the system in 2007 was found to be suppressed compared to the mean behaviour of 2005 and 2006 in Banizoumbou, whereas no change was found in Zindarou. Conclusion With a high abundance of neem plants in many villages in this area, the results of this study suggest that larval control using neem seed powder offers a sustainable additional tool for malaria vector control in the Sahel region of Niger.


Background
Malaria continues to place a large social and economic burden on African communities. Programs to control malaria transmission typically target the adult primary vectors, using techniques such as bed nets and indoor residual spraying that have a high impact on vectorial capacity. However, these methods are vulnerable to development of vector resistance to insecticides [1][2][3][4], vector behavioural adaptation, such as changing preferences for feeding and resting outdoors [5], and logistics and funding problems in reaching the poor, who are most at risk [6]. Historically, environmental management methods that targeted the larval stages of malaria vectors were effective in substantially reducing malaria transmission [7][8][9]. These methods fell out of favour with the widespread introduction of synthetic insecticides and bed nets, which reduce biting rates and are not dependent on such sitespecific knowledge as is required for larval control methods [10,11]. Integrated vector management programmes, employing a variety of tools including larval control, may provide the greatest chance for success in reducing malaria transmission rates [12]. Methods that target the larval stages of mosquitoes have the potential to be effective, low-cost and with low environmental impact [5,[13][14][15][16]. If modern-day larval control is to be a useful addition to the toolbox of malaria abatement methods, it will need to be both low-cost and sustainable.
Neem trees, like that shown in Figure 1, are widespread across the Sahel region of West Africa and are very adaptable and hardy plants [17]. The neem seed kernels contain insecticidal properties due to a combination of approximately 99 active compounds, the most potent of which is azadirachtin, present in the seeds at a concentration of about 5 mg/g of kernel [17]. Neem seed extracts provide a potential larval control method that could be complementary to other malaria abatement methods.
Extracts from neem seeds have documented effects on a variety of insects, which include repellence and anti-feeding, deterrence of egg-laying, inhibition of metamorphosis and disruption of growth and reproduction [17][18][19]. The extracts are also toxic to crustaceans, particularly aquatic crustaceans, some species of fish, such as gambusia and tilapia, as well as nematodes and snails [17,19] and for these reasons it is generally recommended that neem seed extracts not be used in complex aquatic ecosystems [19]. Although birds and bats are often observed eating the neem fruit in eastern Africa without ill effects, trials have shown that the seed kernel can be toxic to birds if consumed [17]. Neem seed extracts have also shown to be toxic to guinea pigs, rabbits and rats [19] and to produce ill effects in dogs, sheep, goats and calves [17] and thus domestic animals should be prevented from eating stored seeds. Tests have shown that neem seed extracts are non-toxic to beneficial species such as spiders, bees, crickets, many bugs and beetles, and are actually beneficial to earthworms [17].
Although there are several commercial neem-based pesticides available, none are currently used in mosquito control programmes [20]. In recent years, there have been a number of studies conducted to investigate the particular effects of neem extracts on malaria-transmitting mosquitoes. Exposure of anopheline larvae to undiluted neem oil has resulted in 100% mortality within 12 hours [21]. When applied to artificial water bodies every two weeks over a period of three months, emulsified neem oil has been shown to have the same effect on larval mortality and adult density as commonly used synthetic insecticides [22]. A study using a neem oil formulation on third and fourth stage Anopheles gambiae s.s. larvae showed 50% inhibition of adult emergence at a concentration of 6 ppm [20]. A study using emulsified neem oil showed that within a three months period (five generations), anopheline larvae failed to develop resistance or change their susceptibility to the oil [22]. Research is ongoing into the potential for neem extracts to provide antimalarial treatments as well as prevention [23][24][25][26].
Resistance to neem-based compounds is more likely to develop using a refined larvicide based on a single active ingredient, such as azadirachtin, than if the whole seed is used with its multitude of compounds [20,27]. Because the efficacy of neem is targeted towards the larval stages, it does not have a "knock-down" effect on adult mosquitoes. It is, therefore, thought to be most effective if used to prevent adult mosquito populations from reaching large numbers and will not be as effective in reducing numbers of adult mosquitoes once populations are allowed to establish [19].
The most effective way to use neem is to apply seed extract to breeding sites when population numbers are low, during the dry season, in order to eradicate as many immature mosquitoes as possible and reduce the population available for breeding when conditions become more favourable. Once the rainy season commences, regular applications of seed extract should continue to prevent immature mosquitoes from emerging as adults. The efficacy of neem seed extracts has been shown to degrade under exposure to sunlight within seven days [18,28] and thus the toxicity is non-persistent in the environment. This provides environmental benefits but also means that regular applications of neem seed extracts would be required to maintain efficacy.
Neem trees are abundant in Banizoumbou village, the site of this study in western Niger. The fruits that fall from the trees are left to decompose on the ground where they fall or are dispersed by wind or animals. Hence, the use of neem seeds as an insecticide in this village does not represent the introduction of a new compound to this area. The difference is the location where azadirachtin will be concentrated, from bare ground beneath trees to the pools that provide mosquito habitat, and thus the shift in risk from exposure to neem in this new location. However, these breeding habitats are ephemeral, being temporarily formed during the rainy season in topographic low points, and do not represent complex aquatic ecosystems. The main risk would appear to be to the cattle in the village, if they were to drink from a pool where neem seeds were applied. To avoid this risk, neem seed powder was not applied to the one permanent pool in the village, which is used for cattle watering.
In a short field trial conducted in Mali, neem seed powder was applied on known Anopheles gambiae s.l. breeding sites on a single occasion at the end of the dry season, prior to the commencement of rains [29]. This was reported to lead to an 86% reduction in adult female mosquitoes in the trial village obtained during one subsequent indoor spray catch, while the control village saw sustained numbers of adult mosquitoes [29]. While this trial sounds promising, it was too limited to draw conclusions from and the detailed methodology and results remain unpublished. This study presents the first field trial of neem seed extracts produced locally and applied at the village scale over the length of a malaria transmission season.

Methods
This study was undertaken in Banizoumbou village, located in western Niger, approximately 60 km northeast of Niamey ( Figure 2) and home to approximately 1,000 people. The area around Banizoumbou has a typical Sahelian semi-arid landscape, gently sloping topography and savannah vegetation. Neem trees are abundant in the village, with approximately 85 trees within a 500 m radius, but the fruits are not utilised by the residents. This density of neem trees within the village was observed to be typical of villages in the area. The fruiting season is roughly June to August annually. During a rainy season that extends from May to early October and peaks in August, many ephemeral pools form within and around the village in topographic low points, which is a typical feature of the hydrology in this region [30]. These pools do not form complex aquatic ecosystems and were not observed to be utilized by the residents. However, they do provide an ideal breeding habitat for Anopheles gambiae s.l.mosquitoes, the major local malaria vector. These pools were the targeted areas for neem seed powder applications. There is only one permanent pool of surface water in the village; it is used primarily for cattle watering and is not used by the people.
Environmental variables, including precipitation, temperature, relative humidity and wind speed and direction, and mosquito abundance have been measured in Banizoumbou since June 2005. Two years' monitoring of environmental conditions and vector dynamics, during 2005 and 2006, enabled a targeted strategy to be developed for the neem seed powder trial in the third year of observations in 2007. The ephemeral pools that form within and adjacent to Banizoumbou were monitored during 2005 and 2006 to determine which pools became habitat for anopheline larvae and it was these pools that were targeted in 2007 for the neem seed powder applications.
Neem seed powder was prepared and applied following suggestions by Schmutterer [17] and a published laboratory trial [31]. Residents of Banizoumbou were asked to collect neem seeds on five occasions throughout the rainy season, beginning in early July when the first ephemeral pools began to appear. The residents were familiar with the seeds and where to find them and were able to easily and quickly collect an adequate supply. Due to the abundance of neem trees within the village, seeds could be gathered mostly from fallen fruit and trees did not have to be stripped of unripe fruit. The fleshy pulp was removed from the outside of the seed casing and seeds were stored either as the bare seed kernel or with the white protective casing around the kernel left intact. Seeds were spread out on grass mats inside a mud brick house to dry for approximately 5-7 days before use.
On the morning of an application day, seeds were crushed into a coarse powder using a mortar and pestle. The mortar and pestle was identical to those used by women in the village for grinding millet and was purchased from a local market in Niamey, to avoid appropriating a mortar and pestle currently used for food preparation. The grinding was carried out by a female resident of the village, using the same technique as is employed for grinding millet. This methodology required only minimal tools -a grass mat for drying seeds, a mortar and pestle for grinding the dried seeds and a bucket for carrying around the powder -that can typically be found within the village. The methodology was designed to be implementable by the village residents and to be low cost.
The first application of neem seed powder occurred on 9 th July 2007. At that time, only one ephemeral pool was present in the village. After this initial application, the pool dried out and there was no rain for several days. The next application occurred on 20 th July 2007, after rain had created some pools in the village, and thereafter applications continued twice weekly until early October 2007, when all pools dried out completely following cessation of rains. This application frequency was chosen to ensure continued efficacy of the powder, due to the short active lifetime of azadirachtin [18,28].
On each application day, powder was applied to all ephemeral pools in and immediately surrounding the village that were known to be breeding sites for An. gambiae s.l. (Figure 2). These ephemeral pools were not observed to establish complex aquatic systems and are not used by the village residents. The powder was carried around in a bucket and liberally sprinkled over the surface of a known breeding pool. The average rate of powder application was approximately 10 g/m 2 of pool surface area, with particular attention paid to pool edges where larvae were observed to congregate. No powder was applied to the one permanent pond in the village because of its use by cattle for drinking water, to avoid any toxicity risk. Applications of the neem seed powder in this study were carried out by the authors to ensure consistency of application rates and locations.
Mature neem trees are reported to produce approximately 20 kg of fruit per year, of which the seed kernel accounts for 10% of the weight [31]. Therefore, the 85 neem trees in Banizoumbou are estimated to produce a total of approximately 170 kg of seed kernel per year, during a fruiting season that coincides with the rainy season and thus presence of ephemeral breeding habitats. At an application rate of 10 g/m 2 of pool surface area, with twice weekly applications for about 12 weeks, the trees in Banizoumbou could cover a total surface area of about 700 m 2 per application. The ephemeral breeding pools in Banizoumbou ranged in size from about 4 m 2 to about 200 m 2 and the quantity of seeds available was sufficient to adequately cover these pools.
The targeted pools were monitored twice weekly in 2007. Observations were made of pool presence, as an indicator of habitat availability. Adult mosquito populations were monitored using CDC miniature light traps deployed at six locations within the village ( Figure 2)  Statistical testing was performed to determine if the relationship between rainfall and anopheline mosquito captures was significantly different in the intervention and non-intervention years. The analysis was performed for both Banizoumbou and Zindarou. It was considered that cumulative rainfall over each season was more appropriate for comparison than hourly or weekly rainfall, because of the non-linear way that breeding pool formation and persistence depends on the rainfall history. At the time of each mosquito capture event (CDC miniature light trap deployment), taken to be midday on the day that the deployed traps were collected, the cumulative rainfall in each year up to that time was calculated. This cumulative rainfall (given in mm) represents the integral of the hourly rainfall from June of each year until the mosquito capture event. Cumulative anopheline mosquito captures were calculated in the same way for each year and in each village. The data points were binned to allow calculation of confidence intervals for these means. 95% confidence limits were calculated for both cumulative rainfall and cumulative mosquito captures with a t-distribution to test the difference of sample means of unknown variance.
Prior to undertaking the field study, preliminary laboratory testing was performed on An. gambiae s.l. larvae obtained from a pool in Niamey. This testing confirmed the efficacy of local neem seeds against anopheline larvae.

Environmental variables
Environmental variables of rainfall, air temperature and relative humidity were recorded during the non-intervention and intervention years to determine if ambient con-ditions could have contributed to changes in observed adult mosquito captures.   Relative humidity values were low in the early part of the season (37-47% in May to June), rose to high levels during the later part of the season when rain events were regular (73-76% in late-July to early-September) and then decreased again after the cessation of rains in November (24-31%).

Breeding pool availability
Observations of pool persistence were recorded for the ephemeral breeding pools during the non-intervention and intervention years to determine if changes to breeding habitat availability could have contributed to changes in observed adult mosquito captures.  intervention years as an indication of any general environmental effects occurring in Banizoumbou that might affect mosquito populations, given that culicine and anopheline mosquitoes share the same ambient environment (air temperature and humidity, rainfall etc) but not the same breeding sites in Banizoumbou.  Figures 8 and 9 show the relationship between cumulative rainfall and cumulative An. gambiae s.l. captures for the intervention and non-intervention years in Banizoumbou and Zindarou respectively. The figures depict paired data points of cumulative mosquito captures and cumulative rainfall, which have been ranked in order of increasing cumulative rainfall for each of years 2005, 2006 and 2007 in each village and then divided into four bins of equal sample size. Figures 8 and 9 show the means and 95% confidence limits for each bin, with the data points left on

Discussion
The relationship between rainfall and An. gambiae s.l. abundance in Banizoumbou is highly non-linear, as shown in Figure 8. There are many factors that could potentially influence the populations of An. gambiae s.l. mosquitoes in Banizoumbou besides the application of neem seed powder to breeding habitats. The collection of data related to ambient environment, breeding pool availability and culicine mosquitoes was undertaken to determine if these other factors could have affected anopheline mosquito abundance.
The results show that environmental variables of rainfall, temperature and relative humidity were comparable between the non-intervention and intervention years. The stability of culicine mosquito populations, during a time when the anopheline mosquito populations were significantly altered, indicates that there was an impact on the mosquito life cycle that only affected the anopheline species. It was observed in each of the study years that, in Banizoumbou, culicine mosquitoes tend to breed in different habitat locations than An. gambiae s.l. and would not, therefore. have been affected by the neem applications. However, all mosquitoes share the same ambient environment, in terms of temperature, humidity, wind speed and prevailing direction, and populations of human inhabitants. Although the behaviour and tolerance to dryness are different for Ae. aegypti, Culex sp. and An. gambiae s.l., the observed population stability of the culicine mosquitoes over the three years shows that there was no major climatic effect on mosquito populations in 2007. Thus, the different behaviour of An. gambiae s.l. in 2007 can be attributed to the fact that anopheline mosquitoes were affected in their breeding habitat. Given that anopheline breeding habitat characteristics were similar between 2006 and 2007, it is suggested that the observed difference in An. gambiae s.l. abundance in 2007 compared with 2006 is due to the addition of neem seed powder applications.
A comparison of the powder's efficacy with previous studies is difficult as they have been conducted under laboratory or highly controlled field conditions, generally using concentrated neem extracts. The method presented here used the entire seed and the powder was produced using minimal tools. Field effects such as wind dispersal, dissolution and mechanical mixing from birds and carts would have reduced the impact of the applied powder. It is, therefore, considered that the neem seed powder performed favourably under true field conditions in this study.
A previous laboratory study has recorded approximately 25% reduction in longevity in adults that were exposed to a neem oil formulation at a concentration of 4 ppm as larvae [20]. This effect is important as a reduction in average adult daily survival rate is crucial for lowering a vector's disease transmission potential. Although adult longevity was not measured in this study, it is possible that adult longevity was also affected by the applications of neem seed powder and contributed to the observed reductions in An. gambiae s.l. abundance in 2007.
Accurate, quantitative data on An. gambiae s.l. larval presence in the breeding pools were not collected with sufficient sampling density in space and time to provide quantitative measures of larval abundance. However, it is known that neem seed extracts affect mosquito larvae primarily by inhibiting metamorphosis and suppressing adult emergence [17]. It is conceivable that a breeding habitat where neem has been applied could exhibit a comparable larval abundance to an unaffected breeding habitat. However, the neem-affected habitat would not be expected to produce as many adult mosquitoes as the unaffected habitat. Hence monitoring of larval abundance may not capture the impact of neem seed powder on mosquito populations. For these reasons, it is suggested that monitoring of adult populations is more appropriate in this study for assessing the effectiveness of neem in a field setting than larval abundance.
The techniques used in this study for seed preparation and application could easily be taught to residents and carried out in other villages. The main obstacle to this technique being implemented by residents in other locations is the ability to identify Anopheles breeding habitats and thus to appropriately target the applications. This would be particularly important if many pools were present in a village and neem seeds were not sufficient to cover every surface water body. In those cases especially, targeting of powder only to pools that were known to be breeding habitats would be important for efficient use of time and resources. As part of this study, residents of Banizoumbou were educated about the mosquito life cycle, the connection between malaria illness and the ephemeral breeding pools, and the reasons for applying powder to these pools. This kind of education, as well as some training in habitat identification, would be necessary to implement this technique in other locations.
The most significant cost of this method is the labour and time required for collection of seeds, preparation of the powder and application to pools. It is estimated that in Banizoumbou, these tasks would require three days per week of labour for one person throughout the transmission season, which lasts about 16 weeks in this region. Using estimates of local daily labour wages, it is anticipated that this intervention would cost about US$200 per year, or roughly US$0.20 per person per year for each Banizoumbou resident. The results of this study suggest that neem seeds could provide an appropriate, sustainable larvicide for the malaria vector An. gambiae s.l. in the Sahel region of Niger and adjacent areas having similar environmental characteristics and vector dynamics. A larger-scale study is recommended to test the efficacy of this method in other, similar villages in the region. A multi-year trial is also recommended to test for any long-term residual effects of using the powder. Although this method will not replace other forms of malaria abatement in Africa, it is suggested that neem seed powder could be a useful additional tool in the fight against this infection.