The entomopathogenic fungus Metarhizium anisopliae for mosquito control. Impact on the adult stage of the African malaria vector Anopheles gambiae and filariasis vector Culex quinquefasciatus

Insect-pathogenie fungi for mosquito control (Chapters 1-3)Malaria and lymphatic tilariasis impose serious human health burdens in the tropics. Up to 500 million cases of malaria are reported annually, resulting in an estimated 1.5-2.7million deaths, of which 90% occur in sub-Saharan Africa. Malaria is caused by protozoa of the genus Plasmodium and is transmitted through bites of mosquitoes belonging to the genus Anopheles. Lymphatic filariasis is caused by helminths, the most widespread species being Wuchereria bancrofti, and is transmitted through bites of mainly Culex quinquefasciatus and certain Anopheles species. Worldwide, approximately 146 million people are infected with the disease.Mosquito vector control is an important way to tight these diseases. In Africa, vector control is almost exclusively based on chemical insecticides, used predominantly to impregnate bed nets and for indoor residual spraying. Growing concerns about their negative impact on human health, on the environment, and about insecticide resistance are the reasons for increasing interests in vector control methods that are not based on chemicais, such as biological contro!.veral biological control agents are known to be effective against mosquitoes such as predatory tish (e.g. Gambusia ajfinis and Poecilia reticulata), nematodes (e.g. Romanomermis culicivorax), microsporidia (e.g. Nosema algerae), bacteria (e.g. Bacil/us thuringiensis israelensis and B. sphaericus), and insect- pathogenie fungi (e.g. Lagenidium, Coelomomyces and Culicinomyces).All of these, however, target the larval stages of mosquitoes. To date, there is no biological control agent for use against the adult stage of mosquitoes. However, reduction in survival of adult mosquitoes is considered to have a much higher impact on transmission than a reduction in the number of mosquito larvae. The objective of this PhD thesis was therefore to search for a biological control agent for adult mosquitoes, and to develop a method to use such an agent in integrated vector management (IVM) in Africa. The primary targets for this research were the major malaria vector Anopheles gambiae s.l., and, to alesser degree, the lymphatic tilariasis vector Culex quinquefasciatus. In Chapter 2 the most important insect-pathogenic fungi for (mostly) larval mosquito control are reviewed. Of these, the Hyphomycetes possess a characteristic that gives them a major advantage over other biocontrol agents to be used for killing adult mosquitoes: Unlike with bacteria, nematodes or microsporidia, the infectious propagules of these fungi do not need to be ingested. Instead, contact with the cuticle is enough for the infective propagules (conidia) to infect the mosquito. A conidium penetrates the insect cuticle by secreting cuticledegrading enzymes. Once inside, the fungus grows rapidly and secretes toxins, which kill the mosquito. Depending on temperature, fungal dosage, and susceptibility of the mosquito to the fungus, the process from inoculation to host death may take between approximately three and ten (or even more) days. After host death, and under favourable conditions of high humidity, the fungus will grow out of the cadaver and produce conidia asexually (sporulation).The strategy envisaged to infect and kill wild mosquitoes in sub-Saharan Africa is based both on the characteristic of Hyphomycetous fungi to infect insects through contact by penetrating the cuticle, and on the behavioural characteristic of An. gambiae mosquitoes to blood feed predominantly inside houses during the night, and remain indoors for at least several hours afterwards to rest and digest the blood mea!. If conidia are applied indoors on so-called 'mosquito resting targets' (see Chapter 9), mosquitoes are expected to acquire an infection ofthe fungus by resting on those targets.In Chapter 3, five different Hyphomycetous insect pathogenie fungi were tested on adult An. gambiae, including Beauveria bassiana, a Fusarium sp. and three isolates of Metarhizium anisopliae. Four of these fungi were isolated from insects in western Kenya, an area of endemie malaria. Isolate ICIPE30 of M. anisopliae proved to be highly virulent for the tested mosquito species, and it was decided to continue further studies with this isolate.The effect of the insect-pathogenie fungus Metarhizium anisopliae on African mosquito vectors (Chapters 4-7)As described in Chapter 4, M. anisopliae was tested both on An. gambiae as weil as on Cx. quinquefasciatus, and a standard contamination technique to infect adult mosquitoes was developed. Using this technique, the effect ofthe fungus on An. gambiae was quantified in more detail by a dose-response bioassay. This experiment showed that at a dose of 1.6 x 1010 conidia m-z, >83% were infected (i.e. mosquito cadavers with sporulating fungus), with a mean LT50 value of 5.6::1: 0.4 days. Later experiments (Chapters 6 and 8) showed that the fungus could be even more effective at that same dose, with infection levels up to 96.4%, and all mosquitoes dead by day 6, whereas uninfected female An. gambiae lived much longer with L T 50 values > 18 days.Apart from the principal effect of the fungus, causing mosquito death by direct contact with conidia, infection with M. anisopliae also caused at least two secondary effects (Chapter 5). One ofthose secondary effects is a reduction in feeding propensity. In one ofthe experiments of Chapter 5, individual female An. gambiae s.s. were offered a total of 8 blood meals. It was found that mosquitoes, inoculated with a moderately high dose of fungal conidia (1.6 x 109 conidia mOz), exhibited reduced appetite upon increasing effects of fungal growth. Of the fungus-infected females, the proportion of mosquitoes taking a second blood meal was reduced with 51 %. This was further reduced to 35.3% for the 4th blood meal. The other observed secondary effect was that infected females took smaller blood meals, resulting in fewer eggs per gonotrophic cycle.In order to achieve the highest possible impact on mosquito populations, it is desirabIe that other contamination pathways besides the primary mode of contamination are utilized to spread the fungus through the population, such as horizontal transmission. The results of experiments described in Chapter 6 showed that, under laboratory conditions, conidia can be transferred from an inoculated female to a 'clean' male during the process of mating, with mean male infection rates between 10.7::1: 302% and 33.3 ::I: 3.8%.Since the mosquito inoculation method described above is based on mosquitoes that rest on conidia-impregnated sheets, it is desirabIe that mosquitoes are not repelled by conidia. To test this, behavioural effects of female An. gambiae in close vicinity of the fungus were investigated (Chapter 7). The results showed that dry conidia have a significant repellent effect (p<0.05). However, when conidia were applied in a suspension of 8% adjuvant vegetable-oil formulation and impregnated on paper, this effect ceased (p=0.205). The results suggest that if the fungus is to be applied as a biological control agent against Afrotropical mosquitoes, conidia should be impregnated on e.g. cotton sheets in an oil-based formulation to avoid repellency effects.Practical approach to mosquito vector control in Africa using M. anisopliae (Chapters 8-10).From a practical and economic point of view, the interval between applications ofthe control agent should ideally be as long as possible, without the agent losing too much efficacy. In the case of commonly used chemical residual insecticides such as permethrin this is about 6 months. Laboratory experiments (Chapter 8) showed that M. anisopliae conidia impregnated on paper and on netting material remained virulent to An. gambiae up to one month after impregnation. Experiments on conidial shelf life under different conditions showed that conidia kept in 8% vegetable oil remained viabie up to at least I month. Conidia stored in 0.05% Tween 80 exhibited only slightly reduced viability after 3 months at 27° and after 6 months at 4°C. Dry conidia stored with silica gel retained viability for at least 6 months. The results suggest that, if applied in the field, re-impregnation should be carrPractical approach to mosquito vector control in Africa using M. anisopliae (Chapters 8-10).From a practical and economic point of view, the interval between applications ofthe control agent should ideally be as long as possible, without the agent losing too much efficacy. In the case of commonly used chemical residual insecticides such as permethrin this is about 6 months. Laboratory experiments (Chapter 8) showed that M. anisopliae conidia impregnated on paper and on netting material remained virulent to An. gambiae up to one month after impregnation. Experiments on conidial shelf life under different conditions showed that condidia kept in 8% vegetable oil remained viable up to at least 1 month. Condidia stored in 0.05% Tween 80 exhibited only slightly reduced viability for a least 6 months. The results suggest that, if applied in the field, re-impregnation should be carried out monthly, but dry conidia can be stored for at least 6 months under conditions of very low relative humidity. Chapter 9 of this thesis describes a field study of domestic application of M. anisopliae in houses in south east Tanzania, a region holoendemic for malaria and lymphatic filariasis. The fungus was applied on black cotton sheets, attached to ceilings as indoor mosquito resting targets. Indoor resting catches of mosquitoes were carried out daily and collected mosquitoes were kept alive in small containers as long as possible to determine survival. Almost 90% of all collected mosquitoes were An. gambiae s.l. (of which 94.7% were An. gambiae s.s. and 5.3% An. arabiensis). In total, 181 wild An. gambiae s.l. and 6 wild Cx. quinquefasciatus were infected with the fungus. Infected mosquitoes died significantly sooner than uninfected mosquitoes, with an average daily survival rate of 0.722 for infected female An. gambiae, against 0.869 for uninfected females. Calculated from the total number of An. gambiae s.l. and Cx. quinquefasciatus that were caught from the fungusimpregnated resting targets, respectively 33.6 and 10.0% had acquired fungal infection. Of the total number of 580 female An. gambiae collected from the houses containing fungusimpregnated sheets, 132 were infected, which is an effective coverage of22.8%. Ifthis same coverage level is assumed at village level, and, together with the reduced daily survival rate, is introduced into a malaria transmission model, the total number of infectious bites per person per year (Entomological Inoculation Rate; EIR) drops from 262 to 14 (Chapter 10). Although the field experiment was on a relatively small scale and of short duration, the predictions of the malaria transmission model strongly indicate that application of M. anisopliae, aimed at the adult stage of African mosquito vectors can have a high impact on transmission intensity. It is argued that large-scale application of this method, implemented as part of an integrated vector management (IVM) strategy including larval control using biological control agents, the use of repellent plants and of unimpregnated bednets, malaria can effectively be controlled without the use of chemical insecticides. This thesis may form a first step towards such a strategy. Further research is necessary, especially in 1) searching for a fungal isolate that has even higher virulence against the targeted mosquito species, 2) testing of non-target effects and safety of the most effective fungal strain for registration, 3) searching for the most optimal formulation and application method to increase infection percentages.