Culicoides species and livestock overlay analysis: a habitat suitability framework for Culicoides insignis, stellifer, and venustus and potential Bluetongue virus presence using environmental and meteorological variables to enhance trap detection
Date
2022
Authors
Kessinger, Peter James, author
Magzamen, Sheryl, advisor
Bosco-Lauth, Angela, advisor
Schaeffer, Joshua, committee member
Mayo, Christie, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Culicoides spp. midges are blood feeding insects capable of transmitting a variety of pathogens. Of particular concern are Bluetongue virus and Epizootic Hemorrhagic Disease virus. Bluetongue virus is extremely dangerous for ruminants, infecting mainly sheep and cattle, and is a growing concern for areas like the United States. There is little known about the range and habitat preference for Culicoides midges, especially in the United States. Our study focuses on predicting habitat suitability for three species of concern: Culicoides insignis, Culicoides. stellifer, and Culicoides. venustus. Each of these species are linked to the spread or potential spread of Bluetongue virus. We obtained data from the Southeastern Cooperative Wildlife Disease Study that included the presence and absence data from midge traps for each of the species of interest from 2008-2020. We combined these data with meteorological data and environmental data to generate a habitat suitability model. The maps were then used to predict the probability of midge species presence in that area and create an overlay analysis for each species of midge and livestock of interest: goats, sheep, and cattle. For the statistical analysis, we used both generalized linear models with binomial regression and random forest models to predict potential midge habitat suitability. We then used the AUC scores to determine model fit using both training and test datasets. Our results indicated that environmental and meteorological variables of significance vary between the species of interest. Most variables were significant for the species of interest, with the most common exception being wind direction. The generalized linear models performed better than the random forest model overall, with C. insignis, C. stellifer, and C. venustus having AUC scores of 0.86, 0.70, and 0.71, for generalized linear models respectively. Overall, prediction models were successful in visualizing and predicting midge presence on the provided environmental and meteorological variables. However, further sampling should be conducted, and variables reassessed for suitability.
Description
Rights Access
Subject
Culicoides
vector
Bluetongue
virus
epidemiology