Graduation Date

Fall 2016

Document Type

Dissertation/Thesis

Program

Master of Science degree with a major in Natural Resources, option Wildlife

First Committee Member Name

Dr. William "Tim" Bean

First Committee Member Email

Tim.Bean@humboldt.edu

First Committee Member Affililation

HSU Faculty or Staff

Second Committee Member Name

Dr. Mark Colwell

Second Committee Member Email

mark.colwell@humboldt.edu

Second Committee Member Affililation

HSU Faculty or Staff

Third Committee Member Name

Dr. James Graham

Third Committee Member Email

James.Graham@humboldt.edu

Third Committee Member Affililation

HSU Faculty or Staff

Abstract

Aerial photographic surveys from manned aircraft are commonly used to estimate the size of bird breeding colonies but are rarely used to evaluate reproductive success. Recent technological advances have spurred interest in the use of unmanned aircraft systems (UAS) for monitoring wildlife. The ability to repeatedly sample and collect imagery at fine-scale spatial and temporal resolutions while minimizing disturbance and safety risks make UAS particularly appealing for monitoring colonial nesting waterbirds. In addition, advances in photogrammetric and GIS software have allowed for more streamlined data processing and analysis. Using UAS imagery collected at Anaho Island National Wildlife Refuge during the peak of the nesting bird season, I evaluated the utility of UAS for monitoring and informing the reproductive biology of breeding American white pelicans (Pelecanus erythrorhynchos). By using a multitemporal nearest neighbor analysis for fine-scale change detection, I developed a novel, automated method to differentiate nesting from non-nesting individuals. All UAS images collected were of sufficient pixel resolution to differentiate adult pelicans from chicks, surrounding landscape features, and other species nesting on the island. No visual signs of disturbance due to the UAS were recorded. Pelican counts derived from UAS imagery were significantly higher than counts made from the ground at observation stations on the island. Analysis of multitemporal images provided more accurate classifications of nesting birds than did monotemporal images, on the condition that multitemporal images aligned with less than 0.5 m error. Nest classifications using multitemporal imagery were not significantly different when conducted across a 24 hour period compared to a 2 hour period. This technology shows promise for greatly enhancing the quality of colony monitoring data for large colonies and a species that is highly sensitive to disturbance.

Citation Style

Journal of Wildlife Management

Share

 
COinS