Skip to content

Mosquito Populations Linked Across Further Distances Than the Viruses They Carry

Closeup of a mosquito perched on a green leaf. The mosquito is mostly golden brown in color, with darker stripes on its abdomen and dark green eyes.

In an analysis of mosquito sampling across 20 years in Connecticut, mosquito populations—such as those of Culex pipiens, a primary vector of West Nile virus—were often correlated at sites 10 kilometers apart and sometimes as far as 40 kilometers apart. But the same data showed the presence of mosquito-borne viruses rarely correlated across distances more than 5 kilometers, complicating potential approaches to managing mosquitoes and the risk of vector-borne disease. (Photo by Ary Farajollahi,

By Joseph R. McMillan, Ph.D.

Joseph R. McMillan, Ph.D.

Joseph R. McMillan, Ph.D.

Effective mosquito control for preventing diseases like West Nile virus (WNV) relies on gathering information on the viral infection status of mosquitoes in numerous areas. In Connecticut, United States, the Connecticut Agricultural Experiment Station (CAES) runs such a viral surveillance program, which involves capturing and testing mosquitoes from 108 sites throughout the state. Local health departments, state agencies, and the public are notified throughout the mosquito biting season (June through October) of virus infections detected in collected mosquitoes. CAES then works with these programs to educate the public about mosquito-borne diseases as well as guide appropriate mosquito control and mosquito bite prevention measures.

One major difficulty of running such a vast surveillance program is communicating the risk of mosquito-borne diseases to the public. How might a resident who lives in Hartford (Connecticut’s capital) use WNV information obtained in New Haven, which is about 40 miles away? What if that resident instead lives in Hamden, which borders New Haven?

CAES continually works with its public health partners to rapidly and effectively communicate the risks of mosquito-borne diseases, and over the last five years CAES has developed a series of projects aiming to identify general and specific patterns of mosquito-borne disease risk and communicate those risks to the public. Most recently, as a postdoctoral researcher at CAES, I led a project using 20 years of surveillance data to measure metrics of population synchrony in mosquito and virus collection timeseries. Conducted with Luis Chaves, Ph.D., associate professor at Indiana University Bloomington and Philip Armstrong, Ph.D., CAES surveillance program director, our project findings were published as an open-access article in March in the Journal of Medical Entomology.

Synchrony is essentially a correlation measure between two or more timeseries from locations that accounts for the distance between locations or the timing of maxima and minima in the timeseries. Defining synchrony of mosquito and virus populations has many applied benefits, such as understanding animal dispersal and insect species invasions. Understanding mosquito and virus population synchrony can also help CAES scientists, health officials, and the public understand the degree of relation in both mosquito and virus activity between the sites in the surveillance program.

Using a mix of approaches, we examined multiple factors of community (e.g., collections of all mosquito species) and population (e.g., numbers of mosquitoes from a single species) synchrony with a particular focus on using climate, land cover, and distance information to predict variation in synchrony estimates.

The primary results of this work show that mosquito species collections were significantly correlated at distances from 5 to 40 kilometers (km) with a 10 km average maximum correlated distance across 19 examined species. Some important vector species collected in Connecticut, such as Culex pipiens (the main vector of WNV) and Culiseta melanura (the main vector of eastern equine encephalitis virus), were correlated up to 20 km.

These correlated collection distances are far larger than the political boundaries of Connecticut’s local cities, where mosquito control measures are managed at the discretion of local officials; if control measures are implemented, they are mostly restricted to the boundary of a city. These local programs are effective at reducing host seeking mosquito numbers; however, larger, coordinated mosquito control and bite prevention initiatives across neighboring cities in Connecticut may result in greater impacts on mosquito populations. While more research is needed to understand population level processes of synchrony, our correlated distance estimates could represent possible extents of treatment zones for population-level control of a target mosquito species.

Unlike the mosquito species examined, the viruses rarely displayed correlated collection patterns greater than 5 km. This could be because the conditional requirements for virus transmission involve interactions between mosquitoes and animal reservoir hosts—interactions that are inherently more complex to survey and often site-specific. A lack of synchrony among the viruses also indicates a high level of resilience in virus communities, suggesting that efforts to prevent mosquito-borne disease exposure in humans likely need to be applied intensely at much smaller spatial scales than mosquitoes.

One striking result from the predictive models of synchrony was that climate and landcover variables explained only a minor amount of variation in synchrony estimates. Instead, mosquito and virus species identity explained the most variance. This suggests that species-specific responses outweigh common impacts of localized weather events on population fluctuations across the 19 mosquito and seven arbovirus species studied in the publication.

CAES hopes that these results, in combination with other ongoing risk projection projects, can better equip Connecticut and its residents to respond to mosquito-borne disease threats. The approach employed in this investigation also has the potential to be transferred to other areas in the US and abroad.

Joseph R. McMillan, Ph.D., is an assistant professor in the Department of Biological Sciences at Texas Tech University in Lubbock, Texas. Email:

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.