Semester of Graduation

Spring 2025

Degree

Master of Science (MS)

Department

Renewable Natural Resources

Document Type

Thesis

Abstract

Aerial surveys play a crucial role in monitoring North America's waterfowl populations and habitat conditions, and consistency across geographies and years allows biologists to track trends through time. The goal of this project was to redesign the northeastern portion of the Louisiana wintering waterfowl survey to align with statistically robust transect surveys conducted in neighboring states within the Mississippi Alluvial Valley. To identify the optimal sampling method, we evaluated three transect weighting schemes: stratified random (unweighted), stratified by watershed, and stratified by expert opinion. Approximately 10% of the total sampling area was randomly selected for each survey period, adhering to all three schemes. Aerial flights, conducted by a Louisiana Department of Wildlife and Fisheries (LDWF) biologist, an LDWF contracted pilot, and myself, followed LDWF's established protocol for survey dates, speed, data recording, and flight altitude. Observers recorded waterfowl out to 250 m on both sides of the aircraft. We used expansion factors to extrapolate total waterfowl abundance from approximately 10% of the habitat we sampled. We calculated the coefficient of variation (CV) for each sampling scheme to identify the most precise method for conducting transect surveys. We used bootstrapping to determine how the CVs would change if fewer transected lines were flown, such that the Louisiana Department of Wildlife and Fisheries can make optimal cost-benefit decisions about the extent of wintering waterfowl surveys in the future. Our data analysis revealed that CVs did not meet our precision goal of ≤20% across all sampling schemes, with Random 10% scheme CVs ranging from 17.80% to 29.40%, Watershed-based scheme CVs ranging from 20.72% to 87.08%, and Expert Opinion-based CVs ranging from 21.66% and 83.28%. Such high coefficients of variation indicate that waterfowl are patchily distributed in very dense concentrations throughout the Louisiana portion of the Mississippi Alluvial Valley. However, by dropping the first November survey where water is scarce and birds are clustered, 10% Random sampling can achieve a CV of ≤25%, and this is the sampling scheme we recommend in future surveys.

Date

4-22-2025

Committee Chair

Ringelman, Kevin M.

Download

Share

COinS