Date of Award
1-1-2022
Publication Type
Thesis
Degree Name
M.Sc.
Department
Great Lakes Institute for Environmental Research
Keywords
Aedes, Best practices, Centroids, Ecological niche model, Mosquito, Species distribution model
Supervisor
H.J. MacIsaac
Supervisor
J.J.H. Ciborowski
Rights
info:eu-repo/semantics/openAccess
Abstract
Understanding species distribution is fundamental to ecology and biogeography, with important implications for species management. This is especially true for species of public health concern—such as mosquitoes—which can impart both economic and health impacts. Species distribution models (SDMs) are powerful tools to accomplish this goal. However, obtaining reliable conclusions, those which are accurate and applicable for policy development, from SDMs is not straightforward owing to the many required considerations such as objective with respect to limitations of the available data. The goal of this thesis was to determine recommended methods to address known and unknown mosquito SDM limitations. First, I identified limitations of SDMs across the literature with respect to best-practice standards. I found that mosquito SDMs exhibited a very high proportion of unacceptable practices. Specifically, SDMs require greater attention to temperature and precipitation thresholds within ecologically relevant scales. Secondly, I quantified the ability of SDMs to determine reliable conclusions from imprecise occurrences, occurrences represented as an administrative/geopolitical boundary centroid, of a virtual species designed to reflect mosquitos of public health importance. I found that imprecise occurrences had a strong negative effect on the reliability of conclusions. Through careful consideration of SDM model building allowed for appropriate conclusions. General boosted regression models, mean or weighted mean ensembles with balanced training and removal of low contributing predictors, consistently provided reliable conclusions from imprecise occurrences. This thesis provides methodologically recommendations to aid in SDM model development, which should translate into better quality predictions of mosquito and possibly other species distributions and management.
Recommended Citation
Barker, Justin, "Mosquito species distribution models: limitations and best practices" (2022). Electronic Theses and Dissertations. 8735.
https://scholar.uwindsor.ca/etd/8735