Date of Award

2017

Document Type

Open Access Thesis

Department

Art

Sub-Department

Earth and Environment Resources Management

First Advisor

John L. Ferry

Abstract

Exposure to pesticide residues continues to be a threat to both human and environmental health despite increased efforts in the agricultural industry to control end-product levels. Multiple government agencies routinely sample and screen common agricultural commodities (fruits and vegetables) for pesticide residues, albeit to do so they use different commodity sampling methods and satisfy different program objectives. Often, results of such screening programs are used in a supplementary fashion in human and environmental health studies, but rarely are the results studied against one another. Six years of archived data (narrowed down from 14) of two separate pesticide monitoring programs were isolated and matching quantitative data were compared against one another. Of particular interest are historical detections of various organochlorines, organophosphates, and organonitrates in common fruits and vegetables as well as detected concentrations of these compounds across both surveys. Historical outcomes were compared using linear regression models and t-tests of the matching detections to investigate trends in the pooled data over the various commodities sampled, chemical compounds detected, detection frequencies, and any effects potentially related to inherent characteristics of both the commodities or compounds. Nearly all t-tests indicated that mean detections of the surveys do not significantly differ at the 5% level; however, t-tests were more likely to detect significant differences as the number of observations grew. Roughly 25% of matrix-specific regression models could explain the variance of one survey’s outcome on another with r2 > 0.90, while nearly half of models had r2 > 0.50. Regressions of compound and matrix structural property effects against differences in survey outcomes were generally less reliable, but did show some trends in the models’ slopes. While not conclusive, the results lay a foundation for future concentrated research and demonstrate the need for increased data sharing and cooperation between all State and Federal agencies, as much of their annual data can be very useful beyond its intended scope when examined conjunctively.

Rights

© 2017, Justin Marshall

Download

Share

COinS