Anyone who has swatted at a mosquito knows the small comfort of a repellent, but a new mathematical study asks what the insect does next. The answer, published in PLOS Computational Biology in December 2025, is that a turned-away mosquito rarely disappears; it searches for another meal, and for West Nile virus that meal is often a bird. In this way a private act of self-protection can ripple outward into the wider web of birds, mosquitoes and people that carries the disease. The researchers offer not only a warning but a path: a tool that quietly removes a small fraction of mosquitoes might ke
Repellents May Push Mosquitoes Toward Birds, Raising West Nile Transmission Risk, Model Finds
Repelled mosquitoes don't disappear; they look for another meal.
Mark
So the headline is that repellent could make West Nile worse?
Mimi
In the model, yes, under certain conditions. A repelled mosquito still needs blood, and birds are often the nearest alternative. The model puts the rise in bird transmission at up to 23 percent.
Luke
Up to. That's a ceiling from simulations, not a measured increase in any real place.
Mark
Why do birds matter so much?
Mimi
They're the reservoir. They can pass the virus back to mosquitoes. Humans and other mammals are dead ends, so a bite on a person doesn't carry it forward.
Luke
Though the dead-end assumption for all mammals is a simplification, and the authors say it likely gives an upper bound.
Mark
And the 2 percent figure?
Mimi
If a repellent killed 2 percent of the mosquitoes that meet it before they switch hosts, the diversion effect disappears in the model. It's a design target for manufacturers.
Luke
Nobody has shown that real products can hit it, or tested it outside the model. Also, the feeding data comes from palearctic Culex and the authors note it varies by setting.
Mark
Should people stop using repellent?
Mimi
The paper doesn't say that. It's about product design and how officials think about interventions.
Luke
Right, and nothing here measures human cases. The link to human risk is inferred through the chain of infected birds and mosquitoes.
Il Polso
- A repellent that keeps a mosquito off one person may simply send it toward a nearby bird, quietly feeding the virus's main reservoir.
- Modelling suggests bird transmission could climb by as much as 23 percent, a rise that would circle back to infect more mosquitoes and raise risk for people.
- The unseen diversion is now visible to researchers, who have built a model that tracks mosquitoes switching hosts after being repelled or disarmed.
- The team proposes a target: a product that kills about 2 percent of mosquitoes before they switch hosts could neutralise the harmful diversion effect.
- The findings remain a model rather than a field trial, but the openly shared code invites others to test and extend the framework to other diseases.
Anyone who has swatted at a mosquito knows the small comfort of a repellent, but a new mathematical study asks what the insect does next. The answer, published in PLOS Computational Biology in December 2025, is that a turned-away mosquito rarely disappears; it searches for another meal, and for West Nile virus that meal is often a bird. In this way a private act of self-protection can ripple outward into the wider web of birds, mosquitoes and people that carries the disease. The researchers offer not only a warning but a path: a tool that quietly removes a small fraction of mosquitoes might keep the protective act from becoming an unseen cost.
A team of researchers set out to answer a simple question with a complicated tool: when a person wearing repellent turns a mosquito away, where does it go? Their mathematical model, published on 18 December 2025 in PLOS Computational Biology by Emma Fairbanks, Matthew Baylis, Janet Daly and Michael Tildesley, suggests the insect rarely gives up. It moves on to another host, and in the case of West Nile virus, that host is often a bird.
The study focuses on Culex pipiens, the mosquito that carries the virus. Birds act as reservoirs, able to be infected and to pass the virus on, while humans and most other mammals are dead ends. Earlier models of disease risk tended to count insects and weigh climate, but rarely asked how an intervention aimed at one kind of host reshapes the others. The new framework fills that gap by following each mosquito through its encounters: it may feed, be killed, be disarmed, or search again.
The model draws on literature data, including blood-meal records from tens of thousands of Culex mosquitoes, and accounts for temperature across a wide range. Its central finding is sobering. Repellent use can push mosquitoes toward birds, raising transmission in the bird population by up to 23 percent, which in turn means more infected mosquitoes and greater risk to people.
The authors also sketch a remedy. If a repellent product killed roughly 2 percent of the mosquitoes it encountered before they found another host, the harmful diversion would disappear. They describe this as a target profile for future control tools. The work is a model rather than a field trial, and its results depend on the assumptions behind it, but the public code leaves room for others to test, refine and apply the approach to other mosquito-borne diseases.
A team of researchers has used a mathematical model to ask what happens to a mosquito when a person wearing repellent turns it away. Their answer, published December 18, 2025 in PLOS Computational Biology, is that the mosquito does not vanish. It looks for another meal, and in the case of West Nile virus, that meal is often a bird.
The paper comes from Emma Fairbanks, Matthew Baylis, Janet Daly and Michael Tildesley. It centres on Culex pipiens, the mosquito that carries West Nile virus. These mosquitoes feed on birds, humans and other mammals. Birds are the reservoir hosts: they can be infected and can pass the virus on to other mosquitoes. Humans and other mammals are dead-end hosts, which can be infected but cannot infect. Many models of disease risk count vectors and the effects of climate, the authors note, but do not capture how an intervention aimed at one host type changes the picture for the others.
The team built on an earlier model of relative vectorial capacity, which they had previously applied to viruses spread by Culicoides midges. That version considered only one susceptible host species. The new version adds reservoir and dead-end hosts and tracks what can happen when a mosquito meets a host: it may feed, be killed before feeding, be disarmed through prolonged inhibition of blood feeding, or give up and search for another host. A mosquito that does feed may still be killed after the blood meal. The model also draws on temperature, which affects how quickly mosquitoes complete feeding cycles, how long they live, and how long the virus takes to become infectious inside them.
The inputs came from a literature review. The incubation model used data from 2,145 mosquitoes. Feeding preferences came from a review of blood meals in 26,857 Culex mosquitoes, including 2,062 Cx. pipiens in palearctic regions. Of those, 68.3 percent had fed on birds, 14.1 percent on humans and 17.2 percent on non-human mammals. The authors assumed a 0.74 probability of transmission from mosquito to host for all hosts, and bird infectious periods of 7 and 10 days. They calculated results for temperatures from 10 to 32 degrees Celsius, and caution that real feeding patterns vary by setting. Because not every mammal is a host for the virus, they say their dead-end assumption likely represents an upper bound.
The central finding is that repellent used by people can push mosquitoes toward birds, raising transmission in the bird population by up to 23 percent. More infected birds mean more infected mosquitoes, which in turn raise the risk to humans.
The authors also identify a way around the problem. If a repellent product killed just 2 percent of the mosquitoes that encounter it before they find another host, the harmful diversion effect would be eliminated. They frame this as a target product profile, a set of specifications for developing control tools that avoid unintended consequences. They also compare the effects of repellency, killing before feeding, disarming and killing after feeding.
The work is a model, not a field trial, and its outputs depend on the assumptions and parameters fed into it. The authors say the framework is flexible and could be applied to other mosquito-borne diseases with complex host dynamics. The R code is publicly available, so others can test and extend it. The study was supported by the Horserace Betting Levy Board and a joint BBSRC/EEID grant, and the authors declare no competing interests.
Citazioni salienti
If repellents killed just 2% of the mosquitoes they encounter before they find alternative hosts, the harmful redirection would be eliminated (paraphrased).— Fairbanks, Baylis, Daly and Tildesley, PLOS Computational Biology
Vuoi la storia completa?
Leggi l'originale su PLOS (Public Library of Science) ↗
Science & Space