Assessing herbivore foraging behavior with GPS collars in a semiarid grassland

Area of Science:

• Animal Ecology
• Rangeland Management
• Remote Sensing Technology

Background:

• Global positioning system (GPS) technology advancements enable detailed tracking of free-ranging livestock distributions.
• Accurate assessment of livestock foraging patterns is crucial for understanding factors influencing their distribution in rangeland ecosystems.

Purpose of the Study:

• To develop and evaluate classification tree models for discriminating cattle grazing behavior using GPS and activity sensor data.
• To identify key activity sensor measurements predictive of foraging behavior.
• To compare the accuracy of binary versus multi-activity classification models and assess the robustness of models across multiple years.

Main Methods:

• Collected GPS and activity sensor data from collared cattle over four years (2008-2011) in eastern Colorado semiarid rangelands.
• Developed classification tree models to differentiate between grazing and non-grazing activities.
• Evaluated the predictive value of different sensor measurements, model types (binary vs. multi-activity), and temporal scopes (annual vs. multi-year).

Main Results:

• A binary classification tree model achieved an overall misclassification rate of 12.9%, correctly identifying 87.8% of grazing locations.
• Distance traveled and head-down sensor position were the most significant variables for predicting grazing activity.
• A single model incorporating data from all four years was more accurate than annual-specific models, indicating sample size is more critical than interannual variation.

Conclusions:

• GPS collars with activity sensors provide a viable method for classifying cattle foraging behavior in semiarid rangelands.
• Model accuracy is influenced by environmental and vegetation specifics, necessitating environment-specific calibrations for GPS collar data.
• Increased sample size across years enhances model robustness more than accounting for interannual variations in foraging behavior.