| dc.description.abstract | Moose (Alces alces) aerial surveys provide the information needed for effective management only when they are accurate and precise. I aimed to identify gaps in and improve understanding of aerial moose surveys in North America by comparing survey techniques and exploring an application of thermal camera technology. Current aerial moose survey methods are compared in a jurisdictional review, including approaches to correcting visibility bias and a discussion of implementation of new technological advancements. Stratified random block (SRB) surveys are the most common, alongside distance sampling (DS) and other survey types. Thermal imagery, Geographic Information Systems (GIS), and Global Positioning Systems (GPS) have been implemented into survey designs to improve the accuracy of estimates or logistics of the survey.
Using a virtual population of moose derived from Newfoundland aerial moose survey observations, a simulation of SRB and two DS surveys is used to compare accuracy, precision, and effort of each survey type. DS survey transects are spaced 1000- and 5559-m apart, and both survey types are sampled at high (~3 moose/km2) and medium (~1 moose/km2) densities. Accuracy is used as the bias in the simulation and statistically significant differences in precision and effort occur for each survey type.
The final chapter focuses on the use of thermal and colour cameras for locating moose and explores a 22-km2 study site located at La Verendrye Provincial Park, Ontario. Collared moose travelling from Grand Portage Indian Reservation in Minnesota are visible in colour and thermal orthophotos. Moose thermal hotspots are computed from the thermal imagery, in a automated model. Hotspots were then examined manually. Two thermal hotspots over open water are misidentified as moose and 19 of 20 moose in the imagery were correctly identified by the automated search.
Management implications of the study are that DS is a viable alternative to SRB surveys and that thermal aerial imagery is limited to animal counts without correction. Under the simulated conditions, DS performs comparably to SRB surveys and is possibly less expensive, or at least less variable in cost. Manual identification of moose by viewing colour imagery is still required to complement a thermal-imagery system, and sexing moose is not achievable from orthophotos alone. | en_US |
