Assessment Model
Data Sources
A sampling program commenced in 1995 to obtain biological information, in particular size and age composition, to inform assessments for Banded Morwong. Sampling was conducted annually in 1996, 1997 and between 2001–2005, then every second year from 2007 onwards (i.e., 2007, 2009, 2011, 2013, 2015, 2017, 2019, 2021). In this sampling program, fish are collected during the spawning closure by commercial fishers working under permit and contracted to the Institute for Marine and Antarctic Studies. Sampling sites and general fishing practices (including the use of standard commercial ‘Banded Morwong nets’) have been standardised as much as possible. Approximately 400 fish were collected in each sampling year. For each fish collected, FL (to the nearest 1 mm) and weight (to the nearest 1 g) were measured, and the pair of sagittal otoliths (hereafter otoliths) were removed, cleaned, and stored dry in plastic vials. Gonads were dissected, weighed (to the nearest 0.1 g), sexed and staged macroscopically according to West (1990). As sampling was not conducted in 2022, only the results of the sampling program up to 2021 are included here.
Commercial catch and effort data are collected through compulsory Tasmanian Commercial Catch, Effort and Disposal Returns. The catch and effort logbooks have been amended several times (1995, 1999, 2007, 2010, 2013 and 2019) in an effort to report at finer spatial scales and provide greater operational detail. While the offshore fishing blocks are still at the 30 nm (1/2 degree) spatial resolution, the logbooks introduced in 2010 have redefined the scale of the coastal blocks.
Data Analysis
Biological Data Analysis
Length and age frequency plots were developed for each sampling year. Sex and sampling-year specific patterns in growth of Banded Morwong were modelled using a modified version of the Schnute and Richards (1990) growth function fitted by nonlinear least-squares regression of FL on age. A sample of six recently settled juveniles collected from Bicheno in 1996, estimated to be around 6 months old, were used to anchor the growth functions for all years and sexes.
Generalised linear mixed-effect models (GLMMs) were used to model the length at maturity of female Banded Morwong. Maturity state (immature or mature) was treated as a binomial response variable with logit link function and modelled as a function of FL. Area (i.e., TAC Areas 1, 2 and 3) was modelled as a random effect term in all models to eliminate potential bias or pseudoreplication resulting from the non-independence of samples collected within the same area. Due to low numbers of immature females in the samples, sampling years were combined as follows: 1996–1997, 2001, 2002–2003, 2004–2005, 2011 & 2013, 2015 & 2017 & 2019, and 2021. Samples collected in 2007 and 2009 were excluded due to small numbers of immature females.
Catch and Effort
For the purposes of this assessment, catch, effort and catch per unit effort (CPUE) analyses are restricted to commercial data provided for the period March 1995 to February 2022. The assessment year for Banded Morwong is based on the quota year (1st March to the last day of February the following year) rather than the financial year (July to June) as for other scalefish species. The current Banded Morwong assessment includes data up to, and including the 2021/22 quota year (which ended 28 February 2022).
Under the quota management system, commercial catches prior to 2016/17 were reported as numbers of fish rather than weight. These numbers were converted to weight based on a conversion ratio of 1.3 kg per fish. In addition, and particularly prior to the introduction of the quota management system, fish were landed in a variety of forms, including gilled and gutted, trunked, and filleted. In these instances, the equivalent whole weight was estimated by applying a standard conversion factor (conversion factors to whole weights are 2.50 for fillet; 1.50 for trunk; and 1.18 for gilled and gutted.).
Two measures of effort have been examined in this and previous assessments: (i) days fished (i.e., number of days on which a catch of Banded Morwong was reported); and (ii) quantities of gear/time fished using the method (effort gear units). For gillnets (the main fishing method for Banded Morwong), effort gear units are measured in 100 m net hours.
Catch returns for which effort information was incomplete or unrealistically high or low (either due to data entry error or misinterpretation of information requirements by fishers) were flagged and excluded when calculating effort levels based on gear units or catch rates based on catch per unit of gear. No fishing records for 2021/22 needed to be excluded in this manner.
Catch Per Unit Effort (CPUE)
In the Banded Morwong Fishery, the amount of gear set and the fishing duration is recorded by fishers, however these data have not been reported consistently over time and among fishers. Accordingly, for the purposes of this assessment, CPUE is calculated using days fished as a measure of effort. Previous work has shown that this is highly correlated with CPUE derived from using gear and soak time where that data is reliably available (IMAS unpublished data). Following Ziegler et al. (2007b), CPUE for Banded Morwong was standardised in order to remove the influence of confounding effects such as area, depth, season and operator on relative trends in abundance. CPUE was standardised using general linear models (GLMs). Standardisation of CPUE was conducted for an annual time scale, and at four spatial scales (whole of TAC area, and individual north-east, east and south-east areas within the TAC area). Data were restricted to skippers who had reported catches for at least two years.
The Generalised Linear Models (GLMs) were fitted to different combinations of factors for which information were available, namely skipper, vessel, fishing block, depth zone fished (0–10 m, 11–20 m, 21–30 m, > 30 m), bimonthly period and reported seal interactions (presence or absence). A bimonthly period rather than month was included as a temporal factor to ensure there were sufficient records in each period to give reliable results. Due to the annual spawning season closure in March and April, five bimonthly periods were available for each year. Interaction terms between some of the factors were also considered, but these were limited to combinations for which sensible interpretations could be ascribed.
Standardised CPUE were fitted to natural log-transformed (Ln) catch rate data (assuming a lognormal distribution), using a normal distribution family with an identity link. All GLMs were fitted using a forward approach by manual stepwise addition of each factor starting with the timestep. The optimal model was chosen by minimisation of the Akaike’s Information Criterion (AIC; Tasmanian Banded Morwong Fishery Assessment 2021/22 Page 10, Burnham and Anderson 1998). Adding new data from 2021/22 resulted in the same model selection as in the 2020/21 assessment.
Stock Modelling
Model Structure
The 2021/22 assessment was conducted using the Banded Morwong population model implemented in CASAL v 2.30 (Bull et al. 2012). The CASAL framework is widely used for fisheries assessments, including several New Zealand fish stocks. The implementation of the Banded Morwong model in CASAL is mathematically equivalent to the previous model used to assess Banded Morwong stocks developed by Ziegler et al. (2007b). The model for the 2021/22 assessment was updated with catch and CPUE information, and a minor adjustment to the treatment of growth information. The model is split into two areas, inshore and offshore, and separates processes into two time-steps per year to reflect a 10-month fishing period and a 2-month no-fishing period during the estimated spawning period. The model was run for the entire TAC area instead of each of the three sub-areas individually. This is consistent with recent assessments of Banded Morwong, which have stressed that a greater emphasis should be placed on the whole quota region for TAC setting purposes.
Because the distribution of Banded Morwong extends beyond the depth of the fishery, there is the potential for a component of the TAC area ‘stock’ to be located in a depth refuge (safe from fishing mortality). The model accounted for this by specifying a fished portion of the stock inshore and an unfished portion offshore. In recent assessment years, the model has consistently estimated that approximately 50% of the adult stock resides in the depth refuge. An alternative, single area, version of the model was also run that does not assume 50% of the adult stock is exempt from fishing pressure. This model configuration estimates spawning stock biomass (SSB) based on the more conservative assumption that individuals escape any risk of fishing mortality solely based on their size (through fishing selectivity and upper size limits), rather than the area in which they reside. Other than this difference, the single area model is identical to the standard, two area model.
Model Inputs
Biological components: Sex-specific lengths-at-ages 1–16 were modelled using the Schnute and Richards (1990) growth function across all fishery areas during the years when biological sampling was conducted (1990–1998 (=’1990’ in model), 1999–2001 (=’1999’ in model), 2002–2003 (=’2002’ in model), 2004–2006 (=’2004’ in model), 2007, 2009, 2011, 2013, 2015, 2017, 2019, 2021). Early sampling years were aggregated based on similar growth trends and to ensure a large enough sample size following Ziegler et al. (2007b). Interpolated growth was assumed for non-biological sampling years within the time series and extrapolated from the most recent sampling year for projected years. Sex-specific patterns in age-at-maturity were modelled using a logistic function fit to gonad stage. Natural mortality was assumed to be constant across sex, age, and time at M = 0.05.
Consistent with preceding assessments, the fishing selectivity-at-length was assumed to follow a knife-edge curve, selecting fish of both sexes between lengths 360 and 460mm with 100% probability. This selectivity function is a rough approximation to the true selectivity, reflecting the keyhole size limit currently in place for the commercial fishery. The survey selectivity-at-length was specified to follow a dome-shaped curve that reflects every fish retained using a net identical to the commercial net (Murphy and Lyle, 1999). The model converts these selectivity-at-length specifications to sex-specific selectivity-at-age using the length-at-age data.
Fishery (harvest) components: Annual total catches within the TAC area for the period 1990/91–2021/22 and standardised CPUE (described above) for the period 1995/96–2021/22 were used in the model. Catchability was estimated in the model from the relationship between observed CPUE and exploitable biomass.
Recruitment: Recruitment was modelled to occur at the start of each year and was assumed to be equal between males and females and occur uniformly along the entire coastline of the modelled area. All recruitment in the model was assumed to occur to inshore populations, consistent with observations of juvenile Banded Morwong in inshore shallow waters and a gradual outward migration with increasing size (Leum and Choat 1980; McCormick 1989b). Recruitment year class strength (YCS) was estimated in the model from the survey age composition data. The mean and variability from YCS estimates during the period 1986-2017 were resampled by the model to project SSB into the future.
Model Outputs
The model was used to estimate SSB trajectories within the TAC area from the start of the fishery to present, and into the future assuming a total allowable catch of 31 t at a unit value of 26 kg / quota unit with the exception of assumed catch for the ongoing 2022/23 season. The TAC for the 2022/23 season is set at 35.8 t to account for a portion of the TAC under-catch in 2020/21 and 2021/22. Model fits to the CPUE and age composition indices were evaluated by visual inspection and examination of residuals.
Model Review
An external review of the assessment model supporting our evaluation of stock status was performed in May 2022 by a stock assessment expert at the Centre for Environment Fisheries and Aquaculture Science (CEFAS) in the UK. The purpose of the review was to identify key areas of model improvement that could facilitate more accurate and precise estimates of biomass and acquire advice on how best to implement the suggested improvements. The review was requested to ensure the assessment model is in line with current best-practices used by assessment scientists internationally. The review was based on the 2020/21 configuration of the model, which included the changes implemented from our internal model review described in previous assessment reports (see reports from 2019/20 and 2020/21).
The major suggestions from the external review included acquiring more data to justify the two-area model and migration rates between areas, reviewing the source and treatment of fishing selectivity specified in the model, and investigating the impact of applying data weighting to all data sources. Additional suggestions included reviewing and updating the assumptions around the stock recruitment relationship and natural mortality for Banded Morwong. The external review thereby confirmed areas of improvement already identified by the internal review. Chiefly among them is the assumption of an offshore region with a migration rate that allows for a large proportion (~50%) of the population to be protected from fishing pressure. Considering this, the one area model will continue to be reported to stakeholders alongside the traditional two area model to adhere to the precautionary approach until research on population biomass and exchange between shallow and deeper reef habitats has been completed. The suggestions from the external review will be further addressed in 2022/23 when additional biological survey data will also become available.
| Year | Type | Alteration |
|---|---|---|
| 20/21 | Data Inputs | Inclusion of catch and standardised CPUE data for 2020/21. Inclusion of sex-specific proportion-at-age biological data for 2021. Inclusion of sex-specific survey composition data for 2021. |
| Model parameters | Update the cv on CPUE time series using Francis (2011) method Apply annual proportion of growth according to time step assumptions: 10 months in first time step, 2 months in second time step. |
Stock Status Definitions
In order to assess the Banded Morwong Fishery in a manner consistent with the national approach (and other jurisdictions) we have adopted the national stock status categories used in the Status of Australian Fish Stocks (SAFS) reporting (Table 7). These categories define the assessed state of the stock in terms of recruitment impairment, which is often treated as a limit reference point. Depleted stocks are not generally collapsed but they do have reduced productivity. The scheme used here does not attempt to assess the fishery against any target outcomes.
Performance Indicators and Reference Points
The determination of stock status is based on the consideration of model outputs and the commercial catch and effort data, which are assessed by calculating fishery performance indicators and comparing them with the limit reference point (SSB of 30% of initial SSB must be exceeded in five years (i.e., 2025 for the current assessment) with a 90% probability).
Other measures are also taken into consideration in the determination of stock status such as changes in biological characteristics of the stock, indicators of stock stress and significant external factors related to fishing activity.
