Accounting for post-release mortality in data-limited, length-based assessments for four demersal fish species with varying life histories

Abstract

Released (or discarded) fish from recreational and commercial line fisheries can experience significant levels of post-release mortality (PRM), affecting sustainability. Although PRM effects are often considered in data-rich (integrated model) assessments, this is uncommon in data-limited assessments. This study investigates impacts of PRM when applying data-limited, length-based assessment methods for four reef-dwelling demersal fish species in temperate Western Australia (Glaucosoma hebraicum, Chrysophrys auratus, Choerodon rubescens and Epinephelides armatus). These species differ considerably in their biology and susceptibility to PRM. For each species simulations were undertaken to (1) evaluate the reliability of a length-based catch curve (LBCC) method for estimating fishing mortality ($F$) and selectivity parameters, and (2) predict the impacts of different PRM levels on values of relative biomass, i.e. ratios of fished to unfished equilibrium spawning biomass ($B_{\mathrm{rel}}$), estimated using a length-based equilibrium analysis (LBEA). LBEA extends length-based per-recruit analysis to incorporate a stock-recruitment relationship as well as allow for PRM. On average, estimates of $F$ and selectivity parameters from LBCC for each species were unbiased when applied to simulated data. Precision decreased, however, when LBCC was fitted to data simulated with recruitment variation. For a given $F$, increased PRM levels reduced $B_{\mathrm{rel}}$ for all species, with the extent of impacts varying depending on species biology. LBCC was then applied to two actual datasets, firstly to lengths for all fish that had either been retained or released during historical recreational fishing trips (to estimate gear selectivity), and secondly to length data for retained fish from recent charter fishing (to estimate $F$ and selectivity of landings). The resultant estimates were then used in LBEA to estimate $B_{\mathrm{rel}}$ with and without accounting for PRM. Relative impacts of PRM varied among species depending on their biology and size limits. While PRM was predicted to have limited impact on $B_{\mathrm{rel}}$ for C. rubescens (as fish are not fully selected by the gear until they attain a relatively large size), accounting for PRM resulted in estimates of $B_{\mathrm{rel}}$ (at their current estimated $F$ levels) being reduced by 31 % for G. hebraicum, 36 % for C. auratus and 23 % for E. armatus. It is recommended that, where evidence of PRM exists for a species, this be accounted for in data-limited assessments.