Ecological inference using data from accelerometers needs careful protocols.

Area of Science:

• Behavioral ecology
• Biologging technology
• Animal movement analysis

Background:

• Accelerometers in animal-attached tags are vital for studying behavior and estimating energy expenditure.
• Data repositories require understanding sensor error sources like type and placement for ecological inference.
• Standardized protocols for error assessment and minimization are needed.

Purpose of the Study:

• To assess the accuracy of tri-axial accelerometers and their impact on dynamic body acceleration (DBA) measurements.
• To investigate how tag type and placement influence acceleration signals in birds.
• To highlight challenges in data interpretation due to confounding factors and propose calibration methods.

Main Methods:

• Laboratory trials with human participants to test accelerometer accuracy and DBA measurement errors.
• Field experiments on pigeons and kittiwakes to evaluate tag type and placement effects on acceleration signals.
• A case study on red-tailed tropicbirds using different tag generations and attachment procedures across seasons.

Main Results:

• Individual accelerometer axes required correction, leading to up to 5% DBA differences in humans.
• Tag placement significantly affected DBA: 9% variation in pigeons, 13% in kittiwakes.
• Seasonal variations in tropicbirds showed DBA differences up to 25%, indicating confounding factors.

Conclusions:

• Accelerometer accuracy, tag placement, and attachment critically impact signal amplitude and biological interpretation.
• A simple field calibration method is proposed for accelerometers before deployment.
• Researchers should assess accuracy in existing data and be cautious of variable placement causing artifactual trends.