Understanding narwhal diving behaviour using Hidden Markov Models with dependent state distributions and long range...

Area of Science:

• Marine Biology
• Animal Behavior
• Quantitative Ecology

Background:

• Narwhal (Monodon monoceros) diving behavior remains largely uncharacterized.
• Previous analyses of whale diving patterns often assumed independence between dive variables.

Purpose of the Study:

• To describe narwhal diving behavior using Hidden Markov Models (HMMs).
• To investigate the impact of dependence between dive variables on behavioral conclusions.
• To identify diurnal patterns in narwhal diving.

Main Methods:

• Utilized a rich dataset of 8,609 narwhal dives from East Greenland over 83 days.
• Applied Hidden Markov Models (HMMs) with 2, 3, or 4 states, incorporating dependent and independent log-normal and gamma distributions.
• Incorporated covariates and periodic B-splines to analyze dive depth, duration, post-dive surface time, and diurnal patterns.

Main Results:

• The study fitted HMMs to a three-dimensional response vector of dive characteristics.
• Accounting for dependence between state distributions significantly impacted conclusions about narwhal diving behavior.
• Inferred distinct diurnal patterns in narwhal diving behavior.

Conclusions:

• Advanced HMMs provide a more accurate representation of narwhal diving behavior than models assuming independence.
• Understanding dive behavior dependence is crucial for accurate ecological interpretations.
• Narwhals exhibit significant diurnal variations in their diving patterns.