The lifespan of wild mosquitoes determines the prevalence of malaria
The length of time that adult mosquitoes live determines how easily malaria can spread through a human population. The longer a female mosquito lives (only the females bite), the greater the chance of them biting an infected human host, and becoming infected with the parasite themselves. Within the mosquito the parasite undergoes a complex cycle of life stages, where it transforms itself, and migrates from the mosquito gut to the salivary glands. The whole process takes around 10 days, and is called the extrinsic incubation period. Only once it has arrived in the salivary glands can the parasite be passed to a human, the next time the mosquito blood feeds. If mosquitoes live longer, it is more likely they will survive the extrinsic incubation period, and go on to bite an uninfected human; passing on the disease.
Our current understanding of how long mosquitoes live in the wild is poor
Whilst it is important to know how long adult mosquitoes live in the wild, currently this aspect of mosquito ecology is poorly understood. The primary method for estimating mosquito lifetimes are mark-release-recapture experiments. In these experiments, mosquitoes are either reared in a lab, or captured from the wild. They are then marked, typically using a fluorescent dust, and released into the wild. By monitoring the number of marked mosquitoes that are recaptured over time this allows us to estimate the combined rate at which mosquitoes die, and migrate away from the study area.
The above figure shows example results from two hypothetical mark-release-recapture experiments, where 1,000 marked mosquitoes were released (on day 0). The mosquitoes were recaptured on even days following release, for two different mosquito populations: one with a long lifespan and/or a slow migration rate (blue dots); another with a relatively short life expectancy and/or a faster rate of migration out of the study area (orange).
These experiments aren’t perfect, and outside factors can affect the number of mosquitoes that we actually recapture on a given day. Because of these random factors that we can’t control, we use statistical techniques to account for them, and estimate the average mortality/dispersion of the mosquitoes. The lines in the figure show the statistical estimates of the average number of marked mosquitoes remaining in the study area (those who neither die, nor immigrate) over time, for each population.
It is worth noting that mark-release-recapture experiments, as I have described them here, cannot differentiate between a mosquito that is not recaptured due to death, or because it has moved out of the study area. This means that the estimated ‘lifetime’ we calculate are lower bounds on mosquito longevity.
Mark-release-recapture experiments are expensive, and only produce rough estimates of mosquito longevity
Mark-release-recapture experiments are expensive, both in terms of the time and effort taken to try to recapture mosquitoes, as well as the financial costs. The cost of these experiments is high because first large numbers mosquitoes need to be reared/collected and marked, and second because substantial recapture efforts are required over a relatively long period (typically a couple of weeks or more) in order to stand a chance of producing accurate estimates.
An insight into the costs of these experiments can be gained by a personal anecdote. When I first came to work on malaria, I was astonished that one of the most effective methods for capturing mosquitoes is so-called ‘human landing catches’. This method is effective because female mosquitoes are strongly attracted to humans in order to blood feed. The ‘collectors’ typically work in in pairs, and whilst a mosquito attempts to blood feed on one of the pair, the other uses a suction device to collect the mosquito.
The use of human landing catches often comes under scrutiny due to ethical concerns regarding the safety of the collectors. However, a 2013 study of 152 collectors in Western Kenya found that their incidence of malaria was 96.6% lower than in an equivalent sample of non-collectors, due to the malaria drugs that were being taken by the workers (Ginmig et al., 2013). This suggests that, so long as adequate malarial drug provision is made available to workers, the risks involved for the collectors are minimal.
Even for well-designed, and well-funded mark-release-recapture experiments, the recapture success rate is extremely low: a 2014 review paper found that the median recapture success percentage for one genus of mosquitoes was only 1% (Guerra et al., 2014). The practical implication of these low rates is that in order to attain an accurate estimate of mosquito lifetime, a large sample of mosquitoes (typically numbering in the thousands) must initially be marked. Even if the numbers of mosquitoes released are high, weather conditions, or simple bad luck, can result in relatively few recaptures, with a large uncertainty in resultant estimates.
Finally, it is worth noting that in in order to estimate average mosquito lifetime, rather than just the average time a mosquito remains in the study area, it is in principle necessary to undertake recapture efforts across a range of spatial locations. This spatial information allows researchers to estimate the rate at which mosquitoes disperse. By accounting for the dispersal rate, this allows researchers to produce valid estimates of mosquito lifespan. However, this extra effort further inflates the cost of these studies. This extra cost means that the majority of mark-release-recapture studies do not collect spatial recapture information.
Pooling results from many existing mark-release-recapture experiments provides accurate estimates of mosquito lifetime at the species level, at no extra cost
Fortunately, a recent review paper introduced a database of over 300 previously-published mark-release-recapture experiments (Guerra et al., 2014). I have been the primary researcher (alongside Ace North, and Charles Godfray) on a paper which pools this data across these different studies, and uses this combined data to estimate mosquito longevity. We have used a method known as a hierarchical Bayesian model to pool this information in an intelligent way, and estimate mosquito lifetimes. An added benefit of the data, and method used, is that it allows estimation of longevity at the species level – a resolution which is rarely possible in individual studies.
The paper is currently being finalised and submitted to a journal. Will post back when we have our final results!