Detecting Prescribed Fire, Haying and Grazing Events via Remote Sensing to Create Grassland Disturbance Landcovers for the Ring-necked Pheasant (Phasianus colchicus)

Authors

Megan A. Baldissara, University of Nebraska-LincolnFollow

First Advisor

Daniel R. Uden

Second Advisor

Andrew R. Little

Committee Members

Larkin Powell, John Carroll, Brian Wardlow, Zhenghong Tang

Date of this Version

12-2025

Document Type

Dissertation

Citation

A dissertation presented to the faculty of the Graduate College at the University of Nebraska in partial fulfillment of requirements for the degree of Doctor of Philosophy

Major: Natural Resource Sciences (Applied Ecology)

Under the supervision of Professors Daniel R. Uden and Andrew R. Little

Lincoln, Nebraska, December 2025

Abstract

This dissertation developed disturbance detection models to fulfill the need for remote sensing landcover products describing grassland structure. The use of landcover products derived from remote sensing is increasing over time in pheasant (Phasianus colchicus) research. Such landcover, however, does not provide relevant pheasant structural habitat information (Chapter 1). Pheasants require tall, high-density grassland for nesting, tall grassland with medium density for brood rearing, and tall grassland for wintering. Time since disturbance can serve as a proxy for structure, as it shapes vegetation by removing biomass and resetting succession. Disturbance is easier to detect than structure with current freely available remote sensing products. In the Great Plains, prescribed fire, haying, and grazing are significant sources of disturbance, so disturbance-specific detection models were developed and applied in northeast Nebraska from 2020 to 2023 to produce grassland disturbance landcover layers. Sentinel-2 remote sensing vegetation indices, weather data and management information were used to train Random Forest models to detect prescribed fire (Chapter 2), haying (Chapter 3), and grazing (Chapter 4). The models performed well, with the fire detection model having a 3% error and a false negative tendency, and the haying and grazing models having a 10% error and a false positive tendency. According to the models' predictions, in 2020, 6% of the study area grassland was disturbed, in 2021, 33%, in 2022, 93%, and in 2023, 45%. These disturbance layers enhance our understanding of wildlife habitat disturbance. This information helps management evaluate cumulative disturbance over time, allowing for predictions of grassland structure for the following years. Practices can then accordingly be adjusted to create a landscape mosaic of grasslands with varied structural types that support pheasants throughout their life cycle.

Advisors: Daniel R. Uden and Andrew R. Little