Estimation of returning Atlantic salmon st oc k from rod exploitation r at e f or pr incipal salmon r ivers in England & Wales.

Abstract

F or eff ective fishery management, estimated stock sizes, along with their uncertainties, should be accurate, precise, and unbiased. Atlantic salmon Salmo salar stock assessment in England and Wales (and elsewhere across the Atlantic) estimate returning salmon stocks by applying a measure of rod exploitation rate (RER), derived from less abundant fishery-independent stock estimates, to abundant fishery-dependent data. Currently, RER estimates are generated for individual principal salmon rivers based on a v ailable local data and assumptions. We propose a single, consistent, transparent, and statistically robust method to estimate salmon stocks that transfers strength of information from "data-rich"rivers, i.e. those with fisheries-independent data, to "data-poor"rivers without such data. We proposed, fitted, simplified, and then validated a Beta-Binomial model of RER, including co v ariates representing angler and fish beha viours, riv er flo w, and random effects to control for nuisance effects. Our "best"model re v ealed co v ariate effects in line with our h ypotheses and generaliz ed to data not used to train it. We used this model to extrapolate stock estimates from 12 data-rich to 52 data-poor rivers, together with their uncertainties. The resulting river-specific salmon stock estimates were judged to be useful and can be used as key inputs to river-specific, national, and international salmon stock assessments.