COS 67-6 - Maximum likelihood method for analyzing mark-release-recapture data of Aedes aegypti using environmental and landscape data

Wednesday, August 14, 2019: 3:20 PM
L013, Kentucky International Convention Center
Tomás León1, Anthony J. Cornel2, Katherine K. Brisco2 and John M. Marshall1, (1)School of Public Health, University of California, Berkeley, Berkeley, CA, (2)Department of Entomology & Nematology, University of California, Davis, Parlier, CA
Background/Question/Methods

Mark-release-recapture (MRR) experiments for mosquitoes and other organisms are commonly conducted, but analytical methods for the data are limited and generally agnostic to the important role environmental factors play in modulating movement. In this study, MRR data of dusted Aedes aegypti mosquitoes from a 2015 release experiment in Clovis, California, were fit with a mechanistic movement model using a maximum likelihood approach for each of 28 trap locations and counts. Rather than simply looking at mean dispersal or distribution of flight distances, directionality and landscape were incorporated into the framework to better reflect the role of the environment in mosquito movement. Different Lévy walk conditions, timeframes, and environmental impedances were explored when finding the optimal movement model that fit the MRR data. Available wind data was incorporated to see if it improved the model’s ability to predict the trap data.

Results/Conclusions

Simulations for yellow-dusted, pink-dusted, and combined mosquito MRR data for 28 traps were fit to one-state and two-state Lévy walk movement patterns by finding the maximum likelihood for matching on each trap’s catch counts based on the Poisson distribution. Two-state movement patterns provided marginal improvement over one-state movement patterns, and the best fit overall was a correlated random walk (r = 0.98, maximum step size = 1 m) with 1.4 h of flight time. Incorporating wind, the best movement model factored in 0.1% of wind influence on mosquito movement and 1.6 h of flight time. The data analysis framework introduced in this study can be used to analyze any MRR dataset with satellite imagery available for the experiment and prior constraint information on organism movement patterns. Results provide a more nuanced interpretation of MRR data for describing mosquito dispersal and highlight the importance of incorporating wind and other environmental factors that affect movement. This interdisciplinary work encompasses ecology, entomology, engineering, and public health, with an aim to better control mosquito-borne disease transmission.