A generalized model for longitudinal short- and long-term mortality data for commercial fishery discards and recreational fishery catch-and-releases

Abstract

Conservation concerns and new management policies such as the implementation of ecosystem-based approaches to fisheries management are motivating an increasing need for estimates of mortality associated with commercial fishery discards and released fish from recreational fisheries. Traditional containment studies and emerging techniques using electronic tags on fish released to the wild are producing longitudinal mortality-time data from which discard or release mortalities can be estimated, but where there may also be a need to account analytically for other sources of mortality. In this study, we present theoretical and empirical arguments for a parametric mixture-distribution model for discard mortality data. We show, analytically and using case studies for Atlantic cod (Gadus morhua), American plaice (Hippoglossoides platessoides), and winter skate (Leucoraja ocellata), how this model can easily be generalized to incorporate different characteristics of discard mortality data such as distinct capture, post-release and natural mortalities, and delayed mortality onset. In simulations over a range of conditions, the model provided reliable parameter estimates for cases involving both discard and natural mortality. These results support this modelling approach, indicating that it is well suited for data from studies in which fish are released to their natural environment. The model was found to be less reliable in simulations when there was a delay in discard mortality onset, though such an effect appears only in a minority of existing discard mortality studies. Overall, the model provides a flexible framework in which to analyse discard mortality data and to produce reliable scientific advice on discard mortality rates and possibilities for mitigation.

Estimating and mitigating the discard mortality of Atlantic cod (Gadus morhua) in the Gulf of Maine recreational rod-and-reel fishery

“…Since ongoing, passive acoustic telemetry was used to obtain the (event) times at which cod tagged with acoustic transmitters either died or were last observed alive (i.e. longitudinal survival data), survival data were evaluated using a specifically adapted parametric survival analysis modelling approach described by Benoît et al (2015).…”

“…predation) needed to be accounted for to avoid inflating the estimates of fishing-related mortality. We used the model of Benoît et al (2015) since it is particularly well suited for this context. At its core, the model is founded on a basic mixture distribution of released individuals that have been adversely affected by the capture-and-release process and will die as a result, and unaffected released individuals that will otherwise survive (see Equation (3), Benoît et al, 2015).…”

“…We used the model of Benoît et al (2015) since it is particularly well suited for this context. At its core, the model is founded on a basic mixture distribution of released individuals that have been adversely affected by the capture-and-release process and will die as a result, and unaffected released individuals that will otherwise survive (see Equation (3), Benoît et al, 2015). This model was further generalized by the authors to include distinct capture-handling and post-release mortality rates, as well as a natural mortality rate that applies to all released fish.…”

“…where t is the probability of surviving capture and handling, p controls the probability that an individual was adversely affected by the fishing event, a and g are, respectively, the scale and shape parameters of an underlying Weibull distribution that determines the mortality patterns over time for the adversely affected individuals, and M is the instantaneous rate of natural mortality (details in Benoît et al, 2015). From Equation (1), it is clear that at t ¼ 0,…”

“…Models were fit using maximum likelihood (see Benoît et al, 2015 for details) and then compared using Akaike's information criterion corrected for small sample sizes (AICc; Burnham and Anderson, 2002). Survival functions predicted by each model were also plotted along with the non-parametric survival functions estimated with the KM estimator to gauge model fit.…”