Mosquito Modeling: A Promising Tool for New Markets

University of Florida researchers have developed mosquito prediction models that will help PMPs develop targeted customer service and mosquito control strategies.

Thinkstock.com

Is there any chance to predict future markets for the pest management industry? Can we predict the change in the number of requested pest control services for the next season? The answers are “yes” — but only if there are available georeferenced data on both mosquito records and climate.

Generally, there are two strategies to predict mosquito management markets. The first strategy depends on requests for mosquito services, which is called “passive surveillance.” The other is called “active surveillance,” which depends on mosquito records from customers’ structures. This is achieved by setting up mosquito traps or inspecting for mosquito larvae.

Each of these strategies is powerful but they are time-consuming — unless they are integrated with modeling software that can predict mosquito markets in un-surveyed areas. Too, the software needs to be able to present prediction maps in an easy-to-understand way. This could be conducted through an interdisciplinary approach utilizing geographic information systems (GIS), projected climate data, and some customer calls and/or mosquito records. Recent advances in GIS and remote sensing (RS) made the use of many geographic data handy in different fields (such as transportation, infrastructure, urban planning and weather prediction).

The reliability of the prediction models depends on the accuracy of the data used in building the models. There are a tremendous number of modeling techniques. Some of these models profile distribution patterns using the records data of mosquito prevalence or customer calls, regardless external predicting variables. Whether you are using mosquito or customer call records, both will give you a depiction about areas that may encounter increased requests for pest management service.

Other models utilize only climate and environmental data — as mosquito predictors — to project hot spots. This gives deep insights on the response of mosquito distribution patterns to their predicting variables. These types of models require either laboratory or field data for both mosquitoes and climate. Laboratory models simply measure the influence of climate predictors on the biological aspects of mosquito development under laboratory conditions, such as development times, survival, etc. Field models require field-collected data on mosquito occurrence/density and their predicting climate variables. Usually, the availability of data determines the type of prediction model.

Click Image to View Gallery

A potential distribution model requires a deep understanding of all factors that predict mosquitoes. For instance, temperature is a critical climate variable for distribution, development and survival of mosquitoes. The increase/decrease in temperature may expand/delimit the geographic distribution of mosquito species.

Other environmental variables may have a significant influence on mosquito distribution such as urbanization and vegetation. Urbanization may show a substantial effect on distribution of urban mosquitoes; meanwhile, vegetation has a significant influence on floodwater and agriculture-related mosquito species. Selection of relevant predicting variables is a crucial step that affects model reliability.

PREDICTING VECTORS. Recently researchers at the University of Florida tried to model the distribution of mosquito vectors of West Nile virus (WNV), Culex quinquefasciatus; and both mosquito vectors of Zika virus (ZIKV): Aedes aegypti and Ae. albopictus. (These are the same vectors of dengue [DV] and Chikungunya viruses [CHIKV]). Our model was specific for Florida and Texas, due to the recent records of locally transmitted ZIKV cases in both states. Accordingly, we used all available field-collected mosquito records, as well as projected climate and environmental data, in both states for the current season. A total of 33 biophysical variables data were included for both study areas in order to predict the hot spot areas.

In our model, the mosquito density data was coupled with human population census per block in order to estimate the mosquito-human contact (MHC) ratio as an indicator for biting rates in these residential blocks. From the standpoint that the flight range of Culex mosquitoes (~5 km) is much more than Aedes (a few feet), we extracted physical data (climate and environment) within the flight range of each mosquito vector. Additionally, to link the estimated biting rate with disease transmission, we assigned the generated risk maps at the minimum infection rate for arboviruses in mosquito pools (0.8). These threshold means highlight areas with risk probabilities 0.8, which reflects the probability of WNV transmission in these areas.

Generally, the distribution of mosquito-human contact ratios varied spatially for each species and both states. The response of each mosquito vector to their corresponding biophysical variables differs from place to place according to distribution of their hosts(s), suitable climate and environmental factors. For Cx. quinquefasciatus, the biology and ecology of this mosquito vector is most likely outdoors in terms of feeding and resting. This vector is known to have elasticity in their feeding preference to a wide range of hosts representing birds, animals and human. Therefore, suitable climate variables, in terms of annual average temperature and annual rainfall, seemed to be the best predicting variables that determine the range of this vector in Florida rather than in Texas.

Both Ae. aegypti and Ae. albopictus are urban mosquitoes closely associated with urbanized settings. Their indoor feeding and resting behavior make them substantially correlated with areas inhabited by humans. Although some breed outdoors in tires, leaf axils and tree holes, they depend on a human blood meal source to breed. Accordingly, the prediction power of climate variables on the distribution of both Aedes mosquito vectors was shared by urbanization and socio-economic status.

These models will help companies develop targeted customer service, mosquito surveillance and control strategies. However, valid mosquito modeling is limited based on the availability of accurate georeferenced mosquito records, recent human population census and climate. These limitations might hinder the usefulness of prediction tools in modeling mosquito distribution. In conclusion, utilization of GIS/RS tools and recent climate models in predicting mosquito distribution can be a great platform through which we can estimate needs for mosquito service by the pest management industry.

The author is a post-doctoral associate at the Entomology and Nematology Department, University of Florida, Gainesville, Fla.

April 2017
Explore the April 2017 Issue

Check out more from this issue and find your next story to read.