Measuring, manipulating and exploiting behaviours of adult mosquitoes to optimise malaria vector control impact

Abstract

Residual malaria transmission can persist despite high coverage with effective long-lasting insecticidal nets (LLINs) and/or indoor residual spraying (IRS), because many vector mosquitoes evade them by feeding on animals, feeding outdoors, resting outdoors or rapidly exiting from houses after entering them. However, many of these behaviours that render vectors resilient to control with IRS and LLINs also make them vulnerable to some emerging new alternative interventions. Furthermore, vector control measures targeting preferred behaviours of mosquitoes often force them to express previously rare alternative behaviours, which can then be targeted with these complementary new interventions. For example, deployment of LLINs against vectors that historically fed predominantly indoors on humans typically results in persisting transmission by residual populations that survive by feeding outdoors on humans and animals, where they may then be targeted with vapour-phase insecticides and veterinary insecticides, respectively. So while the ability of mosquitoes to express alternative behaviours limits the impact of LLINs and IRS, it also creates measurable and unprecedented opportunities for deploying complementary additional approaches that would otherwise be ineffective. Now that more diverse vector control methods are finally becoming available, well-established entomological field techniques for surveying adult mosquito behaviours should be fully exploited by national malaria control programmes, to rationally and adaptively map out new opportunities for their effective deployment.

Conflict of interest statement

Competing interests: None declared.

Figures

Figure 1

An illustration of how high coverage with bed nets can enhance the impact of a second domestic vector control measure with insecticides, such as IRS, by forcing mosquitoes to visit far more houses than they normally would. (A) A schematic representation of how reducing the availability of human blood (Z) with 80% human usage (Uh=0.8) of bed nets (N) can double the number of encounters (E) with humans required by Anopheles arabiensis to obtain a blood meal, relative to baseline conditions with no nets (0). (B) Estimated coverage of the mosquito population (CM) with exposure to insecticide delivered through IRS, at varying levels of house coverage (Ch). Mosquito population coverage is expressed as the proportion of mosquitoes exposed to insecticide per feeding cycle and calculated by expressing equation 8 of a previously published model using the same notation as the model of A. arabiensis early-exit behaviour, assuming that 90% of all attacks on humans would occur indoors in the absence of any protection measure (πh,i,0=0.9). IRS, indoor residual spraying.

Figure 2

A schematic illustration of how sequential layers of vector control interventions against particular fractions of blood-feeding mosquitoes can create measurable opportunities for complementary approaches to achieve increasingly dramatic impacts on vector survival and residual transmission. This illustration is based on the well-characterised example of Anopheles arabiensis in southern Tanzania, as described in the section entitled Manipulating vector behaviours to create new intervention opportunities. We provide a simple online interactive graphical model (https://andysouth.shinyapps.io/coverage1/) allowing the reader to investigate the implications of combining interventions targeting different behaviour patterns under different baseline scenarios of proportional feeding indoor and on humans. The source code (in the statistical language R) is also provided so that the reader can run offline (https://github.com/AndySouth/coverage). LLIN, long-lasting insecticidal net.

Figure 3

Examples of extremely heterogeneous behavioural outcomes, which arise from behavioural plasticity of malaria vector mosquitoes and their human victims, and occur across the full range of spatial scales that are relevant to vector control intervention selection. (A) Specimens of blood-fed, indoor-resting Anopheles arabiensis sampled from 12 different locations within a single village in northern Tanzania yielded estimates for the proportion of blood meals obtained from humans, which are distributed across the full possible range of values. (B) The estimated fraction of A. arabiensis which rest indoors after feeding (reported originally as the estimated usage rate for indoor resting sites per feeding cycle52) varies across a range of more than 300-fold in 21 distinct villages surveyed all across Africa. (C) Variations of only 1–3 hours in the times at which people go indoors for the evening and leave the house in the morning, among 9458 occupants of houses with well-screened windows and ventilation points in a single African city, result in derived estimates for the proportion of remaining residual transmission exposure that occurs outdoors (assuming that a 90% protective effect of the screening is accounted for as previously described9) which are widely distributed across most of the full range of possible values.

Figure 4

Examples of how field-surveyed metrics of adult mosquito behaviours can be predictive of vector control impact. The first two panels illustrate how preintervention measurements for indicators of outdoor resting (A) and early morning biting (B) were predictive of the impact of indoor residual spraying with Propoxur on Anopheles gambiae sensu lato during the Garki Project in northern Nigeria in the early 1970s, while the last panel illustrates how known preferences of mosquito species for feeding on humans are predictive of the impact of window screening in contemporary Dar es Salaam, coastal Tanzania (Chaki et al, Unpublished).