Theses and Dissertations

ORCID

https://orcid.org/0000-0001-5254-2598

Advisor

Iglay, Raymond

Committee Member

Evans, Kristine O.

Committee Member

Samiappan, Sathish

Committee Member

Jones, Landon R.

Date of Degree

5-16-2025

Original embargo terms

Immediate Worldwide Access

Document Type

Dissertation - Open Access

Major

Forest Resources (Wildlife, Fisheries, and Aquaculture)

Degree Name

Doctor of Philosophy (Ph.D.)

College

College of Forest Resources

Department

Department of Wildlife, Fisheries and Aquaculture

Abstract

Accurate animal population estimates inform effective conservation and management strategies. Drones have become a common tool for animal monitoring. Typical drone flight planning involves image overlap to create an orthomosaic; however, this may lead to animal population estimate errors. To quantify this sampling error, I simulated a range of real-world monitoring scenarios for robust application in assessment of animal counts from drones using agent-based modeling. Scenarios investigated effects of 1) animal densities, 2) animal distributions, 3) animal movements, and 4) drone flight patterns. First, I investigated five animal densities (n = 4, 9, 25, 49, 100 animals/survey area) and three stationary animal distribution patterns (random, uniform, and clumped). Large variability was revealed, particularly when animals were clumped. Transect approaches when distributed across the 4.16 km2 survey area, consistently yielded more precise counts than lawnmower patterned or systematic point surveys when animals were clumped. Second, I investigated animal movement bias using one animal with a random, directional persistence, or biased walk at speeds of 2 - 10 m/s within a 0.22 km2 survey area. Drone flight pattern was the most influential factor for determining count accuracy and precision. A lawnmower survey pattern with 0% overlap produced the most accurate count of the solitary, moving animal on a landscape (average count = 1.1, SD = 0.6 individuals), regardless of the animal’s walk type and speed. Image overlap patterns were more likely to result in multiple counts when animals were moving even when accounting for image mosaicking. When animals are expected to be mobile, I recommend using a lawnmower pattern with 0% image overlap to minimize error and improve drone efficacy for animal surveys within more restrictive survey areas. Last, I added multiple animals with various movement patterns in the larger survey area (4.16 km2) to investigate valid subsampling approaches. I again found drone flight patterns to be the most influential factor for count accuracy, despite animal distributions and movements. However, animal distribution and walk type had a greater influence on count precision. My results emphasize the importance of considering survey design and ecological behaviors of target species in drone-based animal survey efforts.

Share

COinS