Sampling populations of mosquitoes by means of light trapping provides crucial information on human risk of mosquito-borne diseases during times of epidemics. The interpretation of mosquito light trap data for predicting disease risk is problematic in that results are species dependent and have an unknown relationship to absolute population density for any species. Integration of remotely sensed data and light trap counts using a multi-disciplinary approach permitted the estimation of population densities within the 256 acre Bronx Zoo. Basic entomological data on host preferences and flight ranges were combined with spatial statistics, artificial neural networks and geographic information systems. Estimates of distributions and densities of An. Punctipennis, Ades vexans, Oc. trivittatus, Culex spp. were used to direct control measures. Although the level of autocorrelation as quantified by Moran’s I was found to vary with species, the combination of both light trap data and remotely sensed vegetation through kriging and a multi-layer feed-forward artificial neural network resulted in marked improvement in the identification of clustering properties of these species. Cluster was found to be independent of previous estimates of flight range distances.
Species 1: Diptera Culicidae Ochlerotatus trivittatus
Species 2: Diptera Culicidae Anopheles punctipennis
Species 3: Diptera Culicidae Aedes vexans
Keywords: spatial statistics, artificial neural networks
Back to Ten-Minute Papers, Section B. Physiology, Biochemistry, Toxicology, and Molecular Biology
Back to Ten-Minute Papers, Section B. Physiology, Biochemistry, Toxicology, and Molecular Biology
Back to The 2002 ESA Annual Meeting and Exhibition