Relative magnitude of cohort, age, and year effects on size at age of exploited marine fishes

• Individual growth is a basic demographic process, and growth rates of fishes vary over time.
• We use a database of weight at age to estimate growth variation for 91 fish stocks.
• Results indicate that 2/3 of variation in average weight at age can be explained using year, cohort, and age effects.
• Half of variability in average weight at age for each stock can be explained using a single factor.
• Results can be used to guide inclusion or smoothing of weight-at age data in assessment models.

Variation in individual growth rates contributes to changes over time in compensatory population growth and surplus production for marine fishes. However, there is little evidence regarding the prevalence and magnitude of time-varying growth for exploited marine fishes in general, whether it is best approximated using changes in length-at-age or weight-at-length parameters, or how it can be represented parsimoniously. We therefore use a database of average weight in each year and age for 91 marine fish stocks from 25 species, and fit models with random variation in length and weight parameters by year, age, or cohort (birth-year). Results show that year effects are more parsimonious than age or cohort effects and that variation in length and weight parameters provide roughly similar fit to average weight-at-age data, although length parameters show a greater magnitude of variability than weight parameters. Finally, the saturated model can explain nearly 2/3 of total variability, while a single time-varying factor can explain nearly 1/2 of variability in weight-at-age data. We conclude that time-varying growth can often be estimated parsimoniously using a single time-varying factor, either internally or prior to including ‘empirical’ weight at age in population dynamics models.