The development and spread of insecticide resistance in mosquito vectors is thought to be a major threat for malaria control and elimination programmes worldwide. Resistance to pyrethroid insecticides has been identified in populations of malaria vectors across Africa [1, 2] and, given the central role of this class of insecticides in insecticide-based vector control (the only one approved by the WHO to be used in treated bed nets and the most-used in indoor residual sprays [3]) the increasing prevalence of resistance is regarded with concern, as it may undermine malaria control and elimination activities [4, 5].

Continuous monitoring of insecticide susceptibility in malaria vector populations informs the choice of chemicals to be used in an area and allows for the management of insecticide resistance [6]. However, insecticide toxicity does not only depend on the active ingredient. The efficacy of a chemical against its target is also a function of the formulation, the biology of the insect, and the environment in which these interact [7]. Thus, it is difficult to predict how an insecticide susceptibility test in a laboratory or insectary, where insecticide dose, mosquito physiological status (e.g., age, blood feeding, larval nutrition) and climate are controlled [8], translates to the efficacy of an insecticide in the field [9].

Environmental temperature in particular has been shown to influence the outcome of insecticide exposure; temperature differences expected to occur naturally under field conditions can lead to notable variations in chemical efficacy [10]. The importance of such changes in effectiveness to the control of malaria vectors has not been widely considered, though temperature-dependent sensitivity to insecticides has been demonstrated in Anopheles gambiae (pyrethroid permethrin: [11], pyrrole chlorfenapyr: [12]) and Anopheles stephensi with varying levels of resistance (organochlorine DDT and organophosphate diazinon: [13], permethrin: [11], permethrin and organophosphate malathion: [14]).

Here, the effects of temperature on the expression of insecticide resistance in Anopheles arabiensis and Anopheles funestus were examined by exposing susceptible and resistant strains of these species to the pyrethroid deltamethrin or the carbamate bendiocarb at 18, 25 or 30°C. The ability of the pyrethroid synergist piperonyl butoxide (PBO) to restore susceptibility in pyrethroid-resistant mosquitoes was also evaluated at these temperatures. This is the first investigation into the effects of temperature on insecticide susceptibility in these major vectors of southern Africa, and the first look at the contribution of environmental temperature to the efficacy of PBO in resistant mosquitoes.

Here it was demonstrated for the first time that: (1) temperature impacted insecticide toxicity in An. arabiensis and An. funestus; (2) temperature affected the toxicity of deltamethrin and bendiocarb differently; and, (3) the synergist PBO fully restored pyrethroid susceptibility independent of temperature. These chemicals are of interest due to their utility in current (i.e., pyrethroids [3, 4]), and future (i.e., synergist PBO, incorporated into the next generation of pyrethroid-LLINs [25]) vector control interventions.

Temperature has long been known to be a critical factor underlying insecticide toxicity [7, 9, 11], and has now been demonstrated to impact the efficacy of public health insecticides against a number of malaria vector species (An. gambiae and An. stephensi [11, 12, 14]; An. arabiensis and An. funestus, this paper). The relationship between insecticide-induced toxicity and temperature is described in terms of a temperature coefficient (TC). If toxicity increases as temperature increases, the chemical has a positive TC. In other insects, DDT and pyrethroid insecticides often have negative TCs, i.e., toxicity decreases with increasing temperature, so the chemical is most effective at lower temperatures [7]. Here, the effect of temperature on deltamethrin toxicity depended on mosquito strain and its temperature coefficient was not always consistently positive or negative. Against both FUMOZ and SENN-DDT females (two different species and selection backgrounds, but similar levels of deltamethrin tolerance/resistance) this pyrethroid was more lethal at temperatures both lower and higher than the standard insectary temperature. Hodjati and Curtis also observed a non-linear (bi-modal) change in pyrethroid toxicity with temperature in An. stephensi exposed to permethrin [11]. They discussed the potential interaction of chemical toxicity or nerve sensitivity and mosquito behaviour over different temperatures. Although behaviour could not be observed in these experiments, the influence of temperature on mosquito activity levels is likely to be especially important in the case of pyrethroid insecticides, which are characterized by their effects on mosquito behaviour: irritancy and knockdown [7]. As the L1014F kdr mutation is present only in the SENN-DDT strain [21], a difference in neural sensitivity may underpin some of the variation seen between the strains.

Interestingly, the ability of the synergist PBO to restore pyrethroid susceptibility was not affected by temperature. At the temperatures tested, all mosquitoes exposed to PBO and then deltamethrin died. This is the first examination of the interaction between PBO and temperature in mosquitoes, and is especially important given that the next generation of long-lasting insecticidal nets (LLINs) that have received interim approval from the WHO to integrate PBO as a pyrethroid resistance countermeasure [25, 26]. Although it is unclear how the addition of PBO might drive increases in the intensity of resistance in mosquito populations, these experiments indicate that PBO could be effective in restoring susceptibility to deltamethrin independent of climatic conditions (in comparison to other novel vector interventions [12]). One possibility is that the addition of PBO as a selection pressure may shift the mechanism of pyrethroid resistance from P450-based to an alternative mechanism as, although P450s are the most common metabolic detoxification mechanism, it is not the only one [27]. Another important observation was that PBO alone for 1 h killed mosquitoes at 30 °C. The increased toxicity of PBO at high temperatures may be a manifestation of an increased reaction rate of irreversible inhibition at the higher temperature; given time, PBO will shift from being a synergist (defined in part by its lack of insecticidal activity on its own) to being toxic [28].

It is important to keep in mind that the outcomes presented here relate to the standard WHO susceptibility tests [8]. However, the observed drastic change in susceptibility over the three temperatures (that are reasonable in African transmission settings [22]), along with the increasing need to integrate non-pyrethroids into control programmes, reinforces the need for further investigation into the influence of temperature on insecticide toxicity. As demonstrated here (and elsewhere [10, 14]), it is important to consider local environmental conditions when monitoring insecticide resistance, as they have the potential to alter the outcomes and thus to affect conclusions and actions needed [9]. As such this work contributes to the ongoing debate on the limitations of the WHO tube assay [29, 30]. For example, SENN would be classified as resistant to deltamethrin, but exposures at cooler temperatures kill enough mosquitoes for SENN to be classified as susceptible. In addition, FUMOZ and FUMOZ-R are resistant to bendiocarb at the standard test temperature. However, when exposed to warmer temperatures mortality increases to 95 and 92%. Although both would still be classified as ‘resistant’, as mortality is below the WHO threshold of 98%, these differences (approx. onefold increase in mortality) may have a significant epidemiological impact. As the current and future chemical arsenal is limited [31] and resistance to multiple insecticide classes is now common [3], an adequate qualification of resistance is critical to maximize the number of available effective tools in the vector control toolbox.

Perhaps more importantly, this phenomenon (observed temperature-toxicity effects) may apply to actual chemical vector control interventions (LLINs, IRS or other chemical-based interventions, such as durable wall liners). Tools may be more or less effective under certain conditions, given that there are strong effects of temperature on toxicity. This was recently highlighted for chlorfenapyr, a pyrrole insecticide being evaluated for inclusion on LLINs for pyrethroid-resistance management [32]. Chlorfenapyr displayed a strongly positive temperature coefficient between 21 and 29 °C against the susceptible KISUMU strain of An. gambiae: while 82–100% of mosquitoes were killed at 27 °C, exposure at 22 °C killed 12–45% [12]. Although it is unclear what this difference means in terms of loss of disease control, it is clear that the importance of temperature in determining the efficacy of tools (and of other well-known factors, such as mosquito age, blood-feeding status, available dose, circadian rhythm [see 9]) adds to the complexity of the ongoing debate on the impact of insecticide resistance on intervention efficacy and transmission intensity [1, 33–36].

Given their utility, insecticide-based vector control tools will continue to play a crucial role in malaria control and elimination strategies [37]. As novel active ingredients for public health insecticides are expected to be delivered no earlier than 2020 [1], and the second generation of non-pyrethroid, multi-insecticide-treated nets will only be available for widescale deployment in several years’ time, the current chemical arsenal needs to be deployed to its maximum potential, which may require us to think beyond laboratory insecticide resistance results. Performing efficacy tests with actual vector control products and local (wild-caught) vectors under real, field conditions (which would include exposures during the appropriate season and relevant time of day) would limit the impact of confounding factors (such as temperature, shown in this paper) and yield more accurate entomological intelligence for evidence-based decision-making.