This repository provides data, methods, scripts and results used to estimate the status of the yellowtail scad fishery in New South Wales (Australia). A complete description of this work is available from Fisheries Management and Ecology
The data consist of measurements of age from a sample of commercial catches, weights from individual fish in specific intervals of age, total catch and purse seine catch and effort.
Total catches in this fishery have been relatively stable, varying between 340 and 670 tonnes per year. The proportion caught using a purse seine varied between 56 and 86% of the total (excluding the unusual first year). The number of days the purse seine fleet fished between 1996/97 and 2017/18 has declined. In the last 8 years, purse seine catches varied in the directions as effort.
Several models describing how mortality rates varied by age-group through time (depending on fishing effort, catchability and age-specific gear selectivity) were developed. All models assume a constant rate of mortality due to natural causes which was estimated. These mortality models have been expressed using hazard functions and converted into probabilities of dying at age from fishing using a method developed in the field of statistics called survival analysis. These age-specific probabilities of dying from fishing were combined with age data sampled from commercial catches into a likelihood function. A range of hypotheses regarding gear selectivity and natural mortality expressed into the mortality models were compared with Akaike Information Criteria (AIC).
So far, three mortality models have been estimated from the data. All models estimate a constant natural mortality (M), catchability (q) and proportions selected by the gear at age between 0 and 6 years old
- Model 1 estimates 8 parameters (M, q and 6 selectivities for age-groups 0--1 to 5--6 years old to have remained constant throughout the entire time series. Gear selectivity was fixed to 1 for older age-groups.
- Model 2 estimated M, q and 2 parameters for a logistic gear selectivity (refer to this article for details). Estimates of selectivity at age are provided here.
- Model 3 estimates 5 parameters assuming a double-logist gear selectivity.
Model 1 is most supported by the data according to AIC. This diagnostic plot shows discrepancies between model 1 and age data: this model does not provide a great fit to the data. There is more variability in the observation of number at age than the model can account for. Profiles of its negative log-likelihood function are smooth around the minimum for each parameter and allow to visualise parameter estimates uncertainties.
Results of applying hazard functions to yellowtail scad data to estimate mortality rates affecting this stock are not as satisfying as in the case of sea garfish. One of the problem with the results presented here is that catchability is estimated to be extremely low, with great uncertainties. It is not clear whether this results from a problem with this analysis, from the data, the method or both. Or if it is really the case that fishing mortality on this stock is extremely low. An alternative analysis using catch curves suggests also that fishing mortality is very low [REF?].\
The model best supported by the data suggests that fishing mortality is very low, well below estimates of natural mortality. Given those estimates, one would conclude that current level of exploitation are not unsustainable.