Copy this text
A Modeling Framework to Meta-Analyse Discard Survival Experiments – Nephrops Norvegicus from European Demersal Trawl Fisheries, a Case Study
Discard survival in commercial fisheries is key in fisheries management policies but challenging to estimate. Pooling information from multiple studies can improve the estimation of survival rate at regional scale and the identification of key drivers. Here, we present a meta-regression (MR) framework that considers differences in experimental design, quality and context specificity between individual studies to produce reliable inferences.First, studies are filtered through a systematic critical review to exclude uncertain or biased results. Discard survival rates are then corrected to limit estimation method bias, and associated uncertainty is included as a weighting. The MR is finally applied under a hierarchical mixed-effects framework to account for the nested structure of the data and correct for experimentally induced mortality bias.We illustrate how the MR can address methodological and analytical limitations in discard survival studies using Norway lobster (Nephrops norvegicus) discarded from European demersal trawl fisheries. While some effects were already identified from single studies, such as the temperature change, the MR highlighted other effects not perceptible at a regional scale, varying more at the haul level, such as tow duration, but also at the level of the individual animal experience, such as carapace length, and potentially physiology.This flexible framework has applicability to other species or contexts. The case study provided insights to make recommendations for future survival studies to improve the predictive potential of this type of MR, such as the importance of following standardized protocols and analyses, and to report data at the finest resolution.
Keyword(s)
Discard mortality, Effect size, Meta-regression, Mixed model, Bottom Trawl, Nephrops norvegicus
Full Text
File | Pages | Size | Access | |
---|---|---|---|---|
Preprint | 31 | 322 Ko |