Details of catchR, an R package to estimate the age and length composition of fishery catches, with an application to 3Pn4RS Atlantic cod

Abstract

Estimating the age and length composition of catches is common practice in stock assessment, as such information on population structure improves our understanding of stock productivity and the impacts of fishing on it. Since aging fish is costly, labour intensive and requires experienced staff, age composition is often inferred from stratified subsampling of age at length and the length composition of the original sample. Although the underlying principle is relatively straightforward, gaps and inconsistencies in the sampling process have led to multiple proposed algorithms in the literature. Importantly, these algorithms are often applied annually in a manner that can be ad hoc and undocumented. Replicating estimates of catches at age from source data can therefore be difficult and can impact the credibility of assessments. For this work, we employed the commonly used forward age-length key coupled to a hierarchical stratified procedure based algorithm for filling gaps in sampling. The R package catchR resulted from this work, and proposes a new, fully transparent and automated framework to estimate catch-at-age. Using the NAFO 3Pn4RS Atlantic cod (Gadus morhua) stock as an example, this research document provides a description of the main catchR functions, get.samples and get.caa, used to calculate catch-at-age. For the 3Pn4RS Atlantic cod stock, the automated algorithm, as currently defined, exposes the shortcomings in the estimation of catch- and weight-at-age, in part because the result of each year is now fully comparable and because no subjective adjustments can be applied to correct for example a poor sampling coverage for any given combination of year, month, NAFO unit area and gear.