Addressing age measurement errors in fish growth estimation from length-stratified samples

Area of Science:

• Fisheries Science
• Quantitative Ecology
• Statistical Modeling

Background:

• Fish growth models are vital for fisheries stock assessments.
• Length-stratified age sampling (LSAS) is a common, cost-effective method for collecting fish length-at-age data.
• Age measurement errors (MEs) can introduce bias into growth estimations.

Purpose of the Study:

• To develop a methodology that simultaneously accounts for LSAS and age MEs in fish growth modeling.
• To provide accurate parameter estimates and measures of uncertainty for fish growth models.
• To validate the proposed methodology using simulations and real-world fisheries data.

Main Methods:

• Utilized empirical proportion likelihood for LSAS and structural errors-in-variables for age MEs.
• Employed a continuation ratio-logit model for age distribution and a discretization approach for computational efficiency.
• Developed model validation tools including uncertainty measures and standardized residuals.

Main Results:

• Neglecting age MEs leads to significant bias in fish growth estimation.
• The proposed methodology accurately estimates fish growth and standard errors, irrespective of age ME magnitude.
• Simulation studies and real data analysis confirmed the effectiveness of the new approach.

Conclusions:

• Accurate fish growth modeling requires accounting for both sampling design (LSAS) and data errors (MEs).
• The developed methodology offers a robust solution for estimating fish growth parameters with improved accuracy and reliability.
• The study provides practical tools and validated methods for fisheries stock assessment and management.