Malaria remains one of the highest priority insect transmitted diseases around the world, with Africa carrying the greatest burden. The Anopheles arabiensis mosquito is considered one of the major vectors of malaria in southern Africa with Anopheles funestus the next most important [¹].

Within South Africa, malaria affects five million people in the low and high risk areas of the country [²] with the threat concentrated in Mpumalanga, Limpopo and (north-eastern) KwaZulu Natal provinces [³]. According to the strategic plan for communicable diseases in South Africa, one of the main objectives is addressing malaria, to reduce the incidence of local transmission from 0.7 to 0.56 cases per thousand. Strategies employed include indoor spraying, definitive diagnosis of malaria cases and effective case management [⁴]. The malaria control strategy in South Africa has a two-pronged approach, targeting the malaria parasite with anti-malarial drugs and controlling the vector through the use of insecticides, targeting both larval and adult life stages [⁵].

Historically, the use of synthetic insecticides has been very effective in reducing malaria transmission. However, over time success has been hampered by the development of insecticide resistance in mosquitoes. Resistance to pyrethroids [⁶] and DDT [⁷] has been reported, and the potential for carbamate resistance has been detected in An. arabiensis, in northern Kwazulu-Natal [⁸].

The effective alternative insecticides that are currently available are up to six times more expensive than those (DDT and pyrethroids) used previously in regional malaria control programmes and accordingly are not cost effective for indoor residual spraying [⁹].

As most mosquito breeding sites are temporary habitats, the use of larvicides by malaria control programmes has been very limited. Larvicidal applications have however been favoured mainly in tourist destinations to limit mosquito populations. This form of vector control has been used as a complementary strategy to indoor residual spraying (for adult stages) and has been suggested as a method for eliminating over-wintering larval populations [¹⁰,¹¹].

Since insecticide resistance threatens to contribute towards the reintroduction of malaria in many parts of South Africa, efforts have focussed on finding an alternative form of mosquito control. For new interventions to be integrated into most malaria control programmes, they should necessarily be cost effective, practical and accessible.

The concept of screening plant extracts for larvicidal activity is not new. The insecticide pyrethroid was derived from flowers of the asteraceous Tanacetum cinerariifolium (= Chrysanthemum cinerariifolium, the Dalmatian chrysanthemum or pyrethrum daisy) [¹²]. Production of such constituents has been interpreted as a means by which plants protect themselves from harmful insect herbivores. The notion of using such plant extracts to control mosquito larvae has received widespread research attention, with plants from various countries tested for larvicidal activity [^13-16]. Such plant extracts have been investigated for their ability to kill a number of mosquito species known to be malaria vectors [¹⁴,^17-19].

The efficacy of different plant parts has been well established [¹⁶,²⁰,²¹]. The flower [²²], leaf and seed [²¹], fruits and roots [²³,²⁴] have all been investigated for their larvicidal properties, revealing variably useful leads for the development of practical interventions.

Although a small number of studies have investigated larvicidal activity of plant extracts against Anopheles stephensi[²⁵,²⁶] there are limited data regarding the testing of plant extracts against An. arabiensis. Zarroug et al.[²⁷] tested some Sudanese plant extracts against An. arabiensis larvae and obtained significant mortality in second and fourth instar larvae when using water extracts of Balanites aegyptiaca.

In the present study, 381 crude plant extracts were investigated for their larvicidal effect on An. arabiensis, and plants exhibiting promising results during screening were selected for further phytochemical investigation. The extracts consisted of plants found native or naturalized in southern Africa.

A total of 80 taxa from 42 families of plants native to or naturalized in southern Africa have been screened for activity against the 3^rd instar larvae of An. arabiensis. The third stage of larvae had been used to determine if the extracts had induced any growth inhibition or abnormalities in ecdysis to 4^th instar and pupation. Significant events had included failed or delayed transition relative to negative control. The results of primary screening have been presented alphabetically by family, genus and species and thereafter in descending order of mortality over a period of seven days (Cited in Additional file 1: Table S1). In order to assess the biological activity of each extract, all samples were subjected to stringent criteria to identify plants that exhibited larvicidal properties. Any plant with activity of more than 80% mortality was considered significant for further investigation [³²]. However, it must be borne in mind that assays were conducted primarily using the crude plant extract.

The results of preliminary screening of 381 plant extracts against the third instar larvae revealed substantial variation in toxicity, differing with plant species, organ, and extraction type. Whilst a substantial number of plants showed moderate to low toxic effects, almost 11% (24 extracts) of the total samples possessed good larvicidal properties relative to the positive control. In an earlier study, Maharaj et al.[³⁰] screened crude extracts of many of the same plant species here reported on, for adulticidal properties. Very limited adulticidal activities were observed, indicating a much greater susceptibility of larvae to the same extracts. The toxicity of the highly active positive control sample had exhibited 100% mortality within 24 hours of exposure, whilst both negative control tests indicated insignificant toxicity of solvents used.

Of the 42 extracts exhibiting high activity in this study, a total of 24 samples demonstrated excellent results (100% mortality during the seven day exposure period) (Table 1). In a separate study, the crude extracts used in this experiment were screened and five samples with repellent properties identified during preliminary assays [²⁹]. Although none of the potential repellents had maintained a repellent effect over a prolonged period, one of five samples, Litogyne gariepina had also been identified as a potential larvicide in the current study, suggesting a dual effect in different environments.

Results indicated that organic extracts exhibit a greater effect than aqueous extracts, a trend also noted in the previous repellency study [²⁹]. In another similar insecticidal experiment, two of the highly active families screened during this experiment, Rutaceae and Zingiberaceae, were reported to yield promising botanical larvicides [³³]. It was also observed in this study that almost half of the prioritized extracts induced mortality within 48 hours, suggesting a very high toxic effect at the crude stage.

Based on the mortality obtained from a large number of crude extracts, the biological activity of each plant was assessed. The most promising candidates represented by four taxa and four extracts were subjected to bioassay-guided fractionation. A total of 36 semi pure fractions were tested in a manner similar to that of the crude extracts (Table 2). A loss of activity was observed in two of the four extracts, relative to the preliminary screening of crude samples. Aloe greatheadii var. davyana (Asphodelaceae) showed 100% mortality for only one fraction, whilst four of the five fractions of Toddalia asiatica (Rutaceae) induced 100% mortality within 96 hours of exposure, making it the favoured extract for further investigation of active compounds. The Rutaceace have earlier been shown to induce insecticidal effects against mosquitoes [³⁴,³⁵]. Murraya koenigii possesses mosquitocidal properties through effects of hormone regulation with subsequent disruption of instar development of Anopheles stephensi, Culex quinquefasciatus and Aedes aegypti[³⁴]. Tiwari et al.[³⁵] found that the essential oil obtained from seeds of Zanthoxylum armatum (Rutaceae) displayed promising larvicidal activity when tested against An. stephensi, Cx quinquefasciatus and Ae. aegypti larvae.

Toddalia asiatica has gained popularity amongst traditional health practitioners for treating numerous ailments. Amongst the documented ethnomedicinal uses, the fruit of this plant is known to have been popularly applied in treating malaria, particularly in East Africa [³⁶]. The current study has revealed that the leaves of T. asiatica contain compounds toxic against the larvae of An. arabiensis. The current findings compare well with a study conducted in India, which has revealed that T. asiatica fruits contains anti-larval compounds active against the 3^rd instar stages of the two mosquito vectors, Ae. aegypti and Cx quinquefasciatus[³⁷].

Although organophosphates, such as temephos, have shown to be very effective in this study, they are not favoured due to their high toxicity to humans, low stability and high cost implications [³⁸]. Bioactive constituents of plants hold potential to be employed as larvicides useful in controlling mosquito vectors [³⁹]. The present study has successfully identified a plant with superior larvicidal activity for both the crude and semi pure fractions thus indicating a potential for use in the control of the malaria vector, An. arabiensis. Results obtained during this study have initiated ongoing investigations into the isolation of purified active components of T. asiatica, with a view to developing an environmentally acceptable tool of value in integrated vector control.

There have been no competing interests.

RM was involved in designing of the study and supervising trials. RG conducted the experiments and was involved in the interpretation of the results. NRC was involved in rationally selecting suitable plant candidates for investigation. Extracts were prepared by VM and PP. NB and PF coordinated and provided scientific inputs into the entire study. All authors read and approved the final manuscript.

We gratefully acknowledge the Department of Science and Technology of South Africa for the Innovation Fund Grant (Project 31313 and TM1002FP). Furthermore, we wish to thank Mr Ashokoomar Saikoolal for his technical assistance in conducting the larvicidal screening.