Simulated likelihood methods for complex double-platform line transect surveys

Area of Science:

• Marine Biology
• Ecological Survey Methods
• Statistical Ecology

Background:

• Conventional line transect methods for estimating animal abundance can be inaccurate.
• Limitations include imperfect detection on the track line, discrete observations, positional errors, and varying detectability.
• These issues necessitate alternative analytical frameworks for robust survey data.

Purpose of the Study:

• To develop a hazard probability framework for independent observer surveys.
• To address limitations of traditional line transect methods in challenging survey conditions.
• To provide a more accurate estimation of effective search width.

Main Methods:

• Developed a hazard probability framework for independent observer surveys.
• Established data likelihood under an assumption of no measurement errors.
• Modified likelihood using simulation-derived functions to account for measurement errors and complexities (simulated likelihood).

Main Results:

• The simulated likelihood method provides a robust framework for complex survey data.
• Successfully applied the methodology to a double-platform survey of minke whales.
• Demonstrated improved estimation of effective search width in challenging scenarios.

Conclusions:

• The hazard probability framework and simulated likelihood offer a superior alternative to conventional methods.
• This approach enhances the reliability of abundance estimates in ecological surveys with imperfect detection and measurement error.
• The method is broadly applicable to various wildlife survey designs facing similar challenges.