Good Practices for estimating and using length-at-age in integrated stock assessments

Estimating growth (increase in size with age) is an integral component of fish population assessment. The use of integrated assessment models combined with the influence of misfitting size composition data on results have led to renewed interest in how growth is modeled in the assessment process. The types of data available to describe the growth process control how the length-at-age relationship will be estimated. Many factors contribute to the complexity of estimating length-at-age, including multiple sources of biological variability and difficulties in getting representative samples. The growth process in the population dynamics model is linked to all other processes and data but most directly influences the assessment model through 1) converting numbers into weight and vice versa, 2) productivity, and 3) modifying fits of size composition data. In some cases, an assessment may be insensitive to moderate levels of misspecification of the growth process, and therefore, relatively simple treatments may be adequate. However, in many cases, especially those where the fit of size composition is influential in estimating scale, a more thorough treatment of the growth process is needed. A complete treatment of growth will estimate the most important forms of biological variability, including individual, sex-specific, temporal, and spatial variability. Several types of sampling bias, including selectivity, length-stratified sam-pling, and spatial and measurement error, will likely also need to be addressed. When sufficient data are available, assessment authors should consider estimating the growth process as part of the integrated assessment model or consider empirical approaches for situations with high biological variability and sampling bias.

Automated summary

Estimating growth is important for fish population assessment. Integrated assessment models and the influence of misfitting size composition data have renewed interest in how growth is modeled. The available data types control how the length-at-age relationship is estimated. Estimating length-at-age is complex due to multiple sources of biological variability and difficulties in obtaining representative samples.