Presenter(s)
Lauren Beck
Abstract
Moose (Alces alces) give birth during a short seasonal window and exhibit distinct changes in movement behavior, including long-distance travel followed by a period of localized movement while caring for newborn calves. Previous research has shown that these patterns can be identified using GPS collar data; however, it remains unclear whether movement data alone can reliably predict calving events. This study evaluated whether movement-based metrics derived from processed GPS collar data from the 2013 calving season could be used to develop a predictive model for calving. Movement data were compiled into a dataset of daily movement features and event-level summaries, and candidate events were identified based on sustained reductions in movement (localization), event duration, and a composite event score. Model predictions were evaluated against confirmed 2013 calving records using precision and recall. Results showed that movement data contains meaningful signal, with models achieving moderate recall (up to 0.679), but relatively low precision (0.45–0.58), indicating frequent false positive detections. Incorporating a biologically informed temporal constraint improved precision but reduced recall, highlighting a tradeoff between sensitivity and accuracy. These findings suggest that while movement patterns are useful for identifying general calving behavior, they are insufficient as a standalone predictor due to behavioral overlap and potential GPS collar error. Additional ecological context may be necessary to improve predictive performance.
College
College of Science & Engineering
Department
Mathematics & Statistics
Campus
Winona
First Advisor/Mentor
Christopher Malone
Location
Kryzsko Great River Ballroom, Winona, Minnesota; United States
Start Date
4-23-2026 9:00 AM
End Date
4-23-2026 10:00 AM
Presentation Type
Poster Session
Format of Presentation or Performance
In-Person
Session
1a=9am-10am
Poster Number
5
Evaluating Movement-Based Detection of Moose Calving Using GPS Collar Data
Kryzsko Great River Ballroom, Winona, Minnesota; United States
Moose (Alces alces) give birth during a short seasonal window and exhibit distinct changes in movement behavior, including long-distance travel followed by a period of localized movement while caring for newborn calves. Previous research has shown that these patterns can be identified using GPS collar data; however, it remains unclear whether movement data alone can reliably predict calving events. This study evaluated whether movement-based metrics derived from processed GPS collar data from the 2013 calving season could be used to develop a predictive model for calving. Movement data were compiled into a dataset of daily movement features and event-level summaries, and candidate events were identified based on sustained reductions in movement (localization), event duration, and a composite event score. Model predictions were evaluated against confirmed 2013 calving records using precision and recall. Results showed that movement data contains meaningful signal, with models achieving moderate recall (up to 0.679), but relatively low precision (0.45–0.58), indicating frequent false positive detections. Incorporating a biologically informed temporal constraint improved precision but reduced recall, highlighting a tradeoff between sensitivity and accuracy. These findings suggest that while movement patterns are useful for identifying general calving behavior, they are insufficient as a standalone predictor due to behavioral overlap and potential GPS collar error. Additional ecological context may be necessary to improve predictive performance.
Comments
Beck, Lauren A