Using hidden Markov models to deal with availability bias on line transect surveys

Area of Science:

• Wildlife population estimation
• Ecological survey methodology
• Statistical ecology

Background:

• Line transect surveys are widely used for estimating animal abundance.
• Availability bias, where animals are missed because they are not detectable when encountered, is a significant challenge.
• Existing methods often struggle with complex availability processes or require instantaneous surveys.

Purpose of the Study:

• To develop novel estimators for line transect surveys that address stochastic unavailability.
• To provide a parametric method for correcting availability bias using hidden Markov models.
• To assess the performance and flexibility of these new estimators compared to existing methods.

Main Methods:

• Formulation of the detection process as a hidden Markov model with a binary state-dependent observation model.
• Incorporation of both perpendicular and forward distances into the observation model.
• Application of estimators to aerial and shipboard whale surveys and evaluation via simulation.

Main Results:

• The developed estimators are more general and flexible than existing parametric models.
• The hidden Markov model approach effectively handles availability bias even without direct observation of availability events.
• Methods relying on availability correction factors or assuming temporal independence can be highly biased.

Conclusions:

• The new hidden Markov model-based estimators offer a robust solution for availability bias in line transect surveys.
• These methods provide more accurate population estimates, especially for non-instantaneous surveys with temporal dependence in availability.
• The approach is broadly applicable to various wildlife survey scenarios.