Ivermectin is a macrocyclic lactone extract from the bacteria Streptomyces avarnitalis that acts at invertebrate glutamate-gated chloride channels 1 inhibiting neurotransmission and thereby affecting the nervous system and muscle function in parasites. This drug has proven to be effective for treating mammals against a broad range of nematodes and ectoparasites 2 and is used globally in animal husbandry practices. Ivermectin has been found safe for treatment of humans 3 and is currently being used with great success in mass drug administration (MDA) campaigns against onchocerciasis and lymphatic filariasis 4 5 . In addition, numerous reports have demonstrated the efficacy of ivermectin in significantly reducing the longevity of several arthropod species such as triatomine bugs, ticks, sandflies, and scabies 3 .

In 1983 the effects of ivermectin on the tsetse fly was first documented by Distelmans et al. 6 who demonstrated high mortality in Glossina tachinoides flies that were fed on piglets treated with ivermectin. Other investigations have shown that the treatment of host animals with the active ingredient can achieve some level of tsetse population control 7 , although the extent of this control was not further investigated in this study. Similar treatment effects were observed in mosquitoes following the MDA of ivermectin to humans: An. gambiae could be killed within 6 days after the administration of a standard dose of ivermectin (150 to 200 μg/kg body weight), indicating a mosquito population reducing effect that may have the potential to reduce malaria transmission 4 . However, this lethal effect was only apparent for one week, therefore this treatment modality alone will not serve as an effective intervention strategy 2 5 .

Numerous studies have demonstrated the efficacy of ivermectin as a larvicide 8 9 and when fed to vertebrates have proven to be effective in reducing longevity to an array of mosquito species such as Culex quinquefasciatus 8 9 10 , An. gambiae 5 11 , An. arabiensis 11 , An. stephensi 9 12 , Ae. aegypti 9 10 , Ae. albopictus 10 , An. sacharova 12 , An. farauti 13 , and An. quadrimaculatus 14 . Reducing the longevity of medically important mosquitoes is a favourable side effect of medical treatments with ivermectin. In addition to this, it was hypothesized that ivermectin could be used for female elimination in mass-rearing facilities that produce male mosquitoes for use in incompatible insect technique (IIT) 15 16 17 18 , and the sterile insect technique (SIT) 19 where only-male releases are permitted.

The SIT is an environmental friendly vector control tactic that becomes more effective under inversely density dependent conditions and is usually integrated with other control tactics on an area-wide basis. The target insect is reared in large numbers in specialised factories, the male sex sterilized with ionizing radiation and sequentially released in the target area in numbers that allow them to outcompete their wild counterparts for mates. Mating of a sterile male with a virgin wild female results in no offspring and the reduced replacement rate will eventually suppress or even eradicate a target population 19 20 .

Spiking bloodmeals with an insecticide for the purpose of mosquito sex separation is not a new idea. This method was implemented in an An. albimanus control program with an SIT component in El Salvador, whereby >95% of females were eliminated by adding malathion to bloodmeals 21 . Although not all females could be eliminated and increased male mortality was a negative side effect, this method was applied for lack of a better alternative until it was replaced by a genetic sexing strain (GSS) based on propoxur resistance 21 in the males.

A number of other substances have also shown to reduce mosquito longevity; however, for the purpose of female elimination in a mass-rearing setting for the SIT, the fatal effects of these compounds need to be quick and reliable, resulting in 100% female kill, while having no detrimental effects on male mating efficiency following their release. We therefore tested substances in addition to ivermectin and malathion that may be suitable, and have been previously reported as having lethal effects in mosquitoes or other insects: spinosad, a known insecticide with acute oral toxicity in honeybees 22 and has been tested for attractive toxic sugar baits (ATSB) for mosquito control 23 ; boric acid which was also used effectively for ATSB against An. gambiae and An. arabiensis in Mali 24 ; dieldrin which has been used in previous studies by Yamada et al. 25 and has shown to have high contact toxicity in An. arabiensis, and finally household detergent to assess if it had any lethal effects when ingested by adult mosquitoes (as opposed to its effects when applied to larval stages, 26 ).

The following experiments were designed to test the various toxicants in bloodmeals to determine whether they can eliminate An. arabiensis females quickly and reliably. In addition, further investigations were conducted to determine the mating efficiency of the remaining males in terms of insemination rates of new, virgin females made available to these males. Based on these results the most efficient toxicant was then used on a larger scale (n = 1000 insects) including irradiation of the insects as pupae to assess the feasibility of this female elimination method in the context of the SIT.

Mosquito colony rearing: An. arabiensis originating from Dongola, Northern State, Sudan, were reared in the insectary as described by Yamada et al. 25 .

The An. arabiensis females readily took a blood meal from day 1, which is not always observed in other mosquito species. Those that fed on the ivermectin-spiked blood were killed reliably and efficiently within 12 hours. No instance of females surviving a sub-lethal dose was observed. At these concentrations, even non-fully engorged females were killed. The challenge is to motivate as many of the females as possible as soon as possible to take a blood meal. Therefore, improving the blood-feeding method may directly increase female killing, for instance removing the sugar source several hours before offering the blood-meal, using fresh blood as opposed to thawed frozen blood, timing the blood-feeding to preferred times of day (for example dusk/dawn, or during the night), prolonging blood-feeding for several hours, or feeding multiple times per day, keeping in mind not all of these suggestions are practical or cost effective in a mass-rearing setting. Some of the treatment cages contained adult females that did not take a blood meal at all (approximately 1 in 500, i.e. 0.002%). These were few, and mostly unusually small in size, and as they refuse to blood feed, (even when offered arm-feeding by a volunteer), they would not pose a threat as a disease transmitter. However, it has been reported for other species, that virgin females sometimes do not blood feed initially, but then may later take a blood meal once inseminated. It was found by Charlwood et al. 29 that in the wild, the size of female An. gambiae is important but not completely responsible for them feeding as virgins. Most of the smaller females survived to mate, but did not seem to survive for very long thereafter 29 . Whether An. arabiensis females that refuse bloodmeals initially go on to mate and subsequently seek a blood meal is not known, and would require further research. Furthermore, if females were to survive bloodmeals that perhaps have a sublethal concentration of ivermectin (due to possible miscalculations while preparing the blood meals), it has been reported that ivermectin in the mosquito disrupts parasite development, thereby reducing the female’s disease transmitting potential 5 .

For sterile male releases in the context of the SIT, the complete elimination of females from a mosquito population at the adult stage as is the case here needs to be quick. The longer the females are present in the cages, the higher the chances of their mating sterile males, which are unable to replenish their sperm stock, making them less efficient in inseminating wild females post-release and therefore compromising the SIT’s effectiveness. However, there have been some studies suggesting that irradiated mosquitoes require some days to recover before regaining mating activity 30 , making the few days needed for 100% female kill not necessarily detrimental to male mating efficiency post release. The implications of keeping the adult males for several days before releasing them have not yet been fully investigated. Allowing adult males to sugar feed, mature sexually, and recover from handling and irradiation before mass releases may indeed improve their competitiveness compared to releasing them as pupae, or immediately after emergence.

The observation of some mortality in both males and females in control cages were not unexpected as overcrowding cages and regular blood-feeding has been shown to cause elevated mortality rates in insectary rearing. Male mortality was low in treatment cages, but survival may still be improved by adding resting sites and additional sugar sources.

Male survival is equally a priority for the SIT. Therefore, the toxicants to be chosen for female killing should ideally induce high oral- and low contact toxicity. In a mass-rearing facility, large amounts of the toxicant may be required, therefore the substance should be relatively inexpensive, readily available, and most importantly, safe for humans, other vertebrates and the environment. Some of the substances tested in this experiment did not meet these requirements, were less effective than ivermectin and should therefore not be considered for mosquito sexing purposes on a large scale. Both ivermectin and spinosad are good candidates as they can be considered a natural product that are used safely in organic agriculture 31 . They have a good environmental profile, and both substances have low vertebrate toxicity, making them relatively safe for human handling.

The mating frequency of the first batch of non-irradiated males from the treatment environment was acceptable considering that there was a 1:1 male:female ratio on the first day post-emergence. This can be explained by the males not having reached full sexual maturation at this point, resulting in minimal to no mating activities during the first day. On the second day, the male:female ratio was closer to 10:1 with only few females available for mating. By day 3, there were no more females available. When these males were added to virgin females again at a 1:1 ratio, they were able to mate as efficiently as the virgin control males in the 3 day mating period. However, it is important to note that fertile An. arabiensis males are able to replenish their sperm stock for a period of at least 6 days 32 . Nevertheless, fertile males from control cages were far less efficient in mating the virgin females after spending 4 days together with females during the experiment although they had a 3 day mating period in which sperm production and mating could take place.

Post irradiation, there was an indication that the females did not take a blood meal as readily on the first 2 days as the non-irradiated females in the previous experiment. This could be due to the need for a recovery phase after the irradiation process 30 . The irradiation did not, however, increase mortality in control females during the first 4 days post-emergence. Although by day 3, most of the females (96.1%) in the treatment groups had been killed, there was less blood-feeding and thus less mortality among females during days 1 and 2. This allowed females to remain available for mating, which in theory may decrease male mating potential after this period considering that fully sterile males are no longer able to produce new sperm and can mate only a few times 32 . Partially sterile males may show similar characteristics in terms of mating abilities and limited sperm production, but this requires further investigation.

As initially expected, the mating efficiency of the partially sterile males from the control cages achieved lower insemination rates in the new virgin females as they had the opportunity to mate with the surviving females for the duration of the 4 blood feeding days. Nevertheless, the males were able to mate with 30% of the virgin females, suggesting that the peak mating activity may be delayed by the effects of irradiation, or partially sterile males are able to mate more often than initially expected. More surprisingly, treatment of males and virgin sterile males showed no difference in their mating capabilities in terms of the proportion of new virgin females inseminated. Although there were still females present in the treatment cages on days 1 and 2, a delay in mating activity until after these days could again explain that there was little to no mating in the treatment cages. Control virgin fertile males were able to inseminate around 60% of the virgin females in the 3 nights. The 10% decrease in the insemination rates seen in both the treatment and virgin sterile males may be attributed to their partial sterility only, with little or no impact on mating potential from the female elimination process.

Ivermectin is an acceptable substance in terms of environmental and health safety, and is very efficient in killing female An. arabiensis following oral exposure via spiked bloodmeals. It is possible to eliminate all females quickly, allowing the remaining males to sustain their mating potential following release, even after irradiation at 70 Gy. Further tests are needed to ensure these capabilities following a fully sterilising dose of 120 Gy 27 33 . Alternatively, the approach we describe here followed by male irradiation at the adult stage is an option that may improve both blood-feeding behaviour in females, and male fitness post release, as it has been reported that irradiation at later (adult) stages incurs less somatic damage thus improving competitiveness in mosquitoes 27 . This method requires further evaluations on the mating potential and competitiveness of males remaining after blood spiking that will be irradiated versus wild males. All in all, the method of female elimination by spiking blood meals discussed here has room for improvement, but has the potential for use in SIT, IIT, or other techniques requiring simple and effective female elimination.

Although the method described here will likely be practical as a temporary solution and for lack of better alternatives, it does, however, have limitations. Keeping females until a blood feeding age in a mass-rearing facility will increase the costs significantly compared to rearing only males until adulthood in terms of feeding, as well as space and labour requirements. In addition, the reliability of the method will not remain at a constant 100% as it is dependent on behavioural attributes, which can never be controlled completely, and may change over time or in response to other external factors. The efficiency and reliability of a good genetic sexing strain will be difficult to obtain, but this method is effective, simple, and has potential for application against many other mosquito species that currently lack alternative sexing methods.

The authors declare that they have no competing interests.

HY conceived of and designed the study, carried out the experiments and drafted the manuscript. SMS assisted in the development of the experiment protocols, the rearing and supply of the mosquitoes, and the dissection of the female mosquitoes. MJBV and DDC contributed substantially in the development of the manuscript. JLRG oversaw the project as group leader. All authors read and approved the final version of the manuscript.