Stock Assessment Models Research Articles

A novel approach for directly incorporating disease into fish stock assessment: a case study with seroprevalence data

When estimating mortality from disease with fish population models, common disease surveillance data such as infection prevalence are not always informative, especially for fast-acting diseases that may go unobserved in infrequently sampled populations. In these cases, seroprevalence — the proportion of fish with measurable antibody levels in their blood — may be more informative. In cases of life-long immunity, seroprevalence data require less frequent sampling intervals than infection prevalence data and can reflect the cumulative exposure history of fish. We simulation tested the usefulness of seroprevalence data in an age-structured fish stock assessment model using viral hemorrhagic septicemia virus (VHSV) in Pacific herring (Clupea pallasii) as a case study. We developed a novel epidemiological model to simulate population dynamics and seroprevalence data and fitted to these data in an integrated catch-at-age model with equations that estimate age- and time-varying mortality from disease. We found that simulated seroprevalence data can provide accurate estimates of infection history and disease-associated mortality. Importantly, even models that misspecified nonstationary processes in background or disease-associated mortality, but included seroprevalence data, accurately estimated annual infection and population abundance.

Predicting year class strength for climate-stressed gadid stocks in the Gulf of Alaska

Climate change makes fish stocks more vulnerable to recruitment failure, and early detection of these events is important for an effective management response. Here, we evaluate the value of larval and juvenile surveys, and a thermal spawning habitat index, for predicting recruitment in two economically important gadids, walleye pollock (Gadus chalcogrammus) and Pacific cod (G. macrocephalus), in the Gulf of Alaska. These stocks have been exposed to rapid human-induced ocean warming since 2014, which has apparently contributed to anomalies in age structure, size at age, and other population variables (for pollock) and stock collapse (for cod). We found that warming results in recruitment that falls short of predictions from historical spawner-recruit relationships for both stocks, highlighting that climate change makes recruitment expectations based on historical experience less reliable. However, we also found that recruitment could be successfully predicted with surveys of early-life stages. Using Bayesian regression, we found that juvenile trawl survey data for pollock predicts recruitment to age-1 (as estimated by a stock assessment model), while prediction from larval surveys was less successful. Beach seine estimates of juvenile abundance also predicted pollock recruitment, a surprising result for a species that is typically sampled in offshore habitats. The spawning habitat index and beach seine survey both predicted cod recruitment to age-3 as estimated by the stock assessment model. We did not find a predictive relationship between cod larval abundance and recruitment. However, residuals from the larval model showed low-frequency variability, suggesting nonstationarity (time-dependence) in the predictive relationship. Dynamic Factor Analysis (DFA) models summarizing information across multiple data sets showed reasonable predictive value for both species (Bayesian R2 ≈ 0.4 for log recruitment), and they also allowed recruitment prediction for years with missing observations in some data sets. We conclude that surveying multiple early life stages may be the most useful approach for predicting gadid recruitment.

Global crustacean stock assessment modelling: Reconciling available data and complexity

AbstractDespite their growing socio‐economic importance globally, relatively little is understood about how crustacean stocks are assessed, which has potential to compromise fishery sustainability, especially under heavy exploitation and environmental changes. To inform stock assessment model application for emergent fisheries, we evaluated model use for crustacean stocks available in the RAM Legacy Database (RAMLDB) and the evolution of model use for four case‐study fisheries, emphasizing the relationship between data availability and model complexity. Differences in model use between FAO fishing regions and crustacean species sub‐groups were identified. Only 60.9% of crustacean stocks in the RAMLDB identified the model used for assessment. For the remaining stocks, we collected ancillary data to fill the information gaps, amounting to 92.5% of crustacean stocks in RAMLDB. Of these, model complexity varied from count‐based to environmentally explicit statistical catch‐at‐length methods, but tended to be data intensive, likely due to biases towards regions with more developed fishery management programmes. Furthermore, regional comparisons indicated that crustaceans are only well‐assessed in a few geographical hotspots. The progression of model use over time was inconsistent between case‐study fisheries, being driven by myriad factors including data availability, confidence in biological processes and ecological considerations. Our findings can be used as a resource to help inform model choice for fisheries management. Towards the goal of seeking global best practices for crustacean stock assessments, future work should address knowledge gaps in regional stock assessment model use and conduct comparative studies to evaluate stock‐specific costs and benefits relating to model complexity.

Playing the detective: Using multispecies approaches to estimate natural mortality rates

Stock assessment models often assume natural mortality rates (M) are constant and attribute the variability in a stock’s overall mortality rate to variability in fishing mortality and recruitment. However, this assumption may not be valid if M varies due to both direct and indirect trophic interactions as well as environmental variability. This is particularly the case when there are substantial changes in the overlapping abundances of a key predator or prey species, or extreme environmental changes impacting an ecosystem. Hence multispecies models (MSMs) are used to explicitly capture variations in natural mortality rates and improve the ability to discriminate between mortality due to fishing versus natural mortality, plus quantify the influences of fishing on the broader ecosystem, consistent with an ecosystem approach to fishing (EAF). There are a growing number of approaches to modelling natural mortality rates in MSMs, with models developed for tactical applications requiring a higher level of rigour and consideration of uncertainty than broad strategic ecosystem models. We overview approaches to modelling natural mortality in Models of Intermediate Complexity for Ecosystem assessments (MICE) and use four case studies to highlight lessons learnt, applications and provide some guidelines going forward. We identify ten broad application categories for MSMs (incorporating explicit representation of components of M), to advance EAF. These include: (1) quantify variability in natural mortality by age and over time as inputs to single species stock assessment models, because of the recognition that predation mortality (M2) is a large and variable portion of total M; (2) inform ecosystem reference points; (3) use as operating models in Management Strategy Evaluation frameworks; (4) simulation test broader management levers; (5) optimise multispecies harvest strategies; (6) inform on management of bycatch and ‘choke’ species; (7) represent and forward project trajectories and model recovery scenarios for threatened and protected species; (8) correctly attribute sources of mortality to support management efforts; (9) pest management and (10) broader applications (e.g. habitat alterations). Using well-constructed MSMs to correctly specify M reduces bias in model parameters, reference points and projections and is increasingly important as ecosystems respond to a more variable and changing climate.

COVID-19 lockdowns reveal the resilience of Adriatic Sea fisheries to forced fishing effort reduction

The COVID-19 pandemic provides a major opportunity to study fishing effort dynamics and to assess the response of the industry to standard and remedial actions. Knowing a fishing fleet’s capacity to compensate for effort reduction (i.e., its resilience) allows differentiating governmental regulations by fleet, i.e., imposing stronger restrictions on the more resilient and weaker restrictions on the less resilient. In the present research, the response of the main fishing fleets of the Adriatic Sea to fishing hour reduction from 2015 to 2020 was measured. Fleet activity per gear type was inferred from monthly Automatic Identification System data. Pattern recognition techniques were applied to study the fishing effort trends and barycentres by gear. The beneficial effects of the lockdowns on Adriatic endangered, threatened and protected (ETP) species were also estimated. Finally, fleet effort series were examined through a stock assessment model to demonstrate that every Adriatic fishing fleet generally behaves like a stock subject to significant stress, which was particularly highlighted by the pandemic. Our findings lend support to the notion that the Adriatic fleets can be compared to predators with medium-high resilience and a generally strong impact on ETP species.

Walleye pollock (<i>Gadus chalcogrammus</i>, Pallas, 1814) fishery, size and age composition, state of stocks and catch prospects in the waters adjacent to Kamchatka peninsula and Northern Kuril Islands

The article provides data on the catch of the Northern Okhotsk, Eastern Kamchatka and Western Bering Sea walleye pollock stocks, fishery structure by the fishing gears in 2016–2019, size and age composition of the fish in the commercial trawl and Danish seine catches in 2010–2019. Data on the generation abundance and stock condition indices used in the stock assessment models are also demonstrated. Interannual dynamics of the total and spawning stock biomass, determinants of the dynamics and prospects of fishing are analyzed.

Why natural mortality is estimable, in theory if not in practice, in a data-rich stock assessment

The process equations of a stock assessment model that predict catch at age and survey CPUE at age can be recast as a single set of equations for the change in survey CPUE of a cohort from one year to the next. Given deterministic data, these equations can be solved for the model parameters by applying the EM algorithm, even when almost all of the age-specific survey catchabilities and natural mortality rates are free parameters. This exercise shows that the usual stock assessment data, if error-free, do contain the information necessary to estimate natural mortality, even age-specific rates, but only if the data contain adequate variation in fishing mortality rates. The EM calculations break down when applied to stochastic data; they are not an alternative to numerical model fitting for parameter estimation.

Error Distribution Model to Standardize LPUE, CPUE and Survey-Derived Catch Rates of Target and Non-Target Species

Indices of abundance are usually a key input parameter used for fitting a stock assessment model, as they provide abundance estimates representative of the fraction of the stock that is vulnerable to fishing. These indices can be estimated from catches derived from fishery-dependent sources, such as catch per unit effort (CPUE) and landings per unit effort (LPUE), or from scientific survey data (e.g., relative population number—RPN). However, fluctuations in many factors (e.g., vessel size, period, area, gear) may affect the catch rates, bringing the need to evaluate the appropriateness of the statistical models for the standardization process. In this research, we analyzed different generalized linear models to select the best technique to standardize catch rates of target and non-target species from fishery dependent (CPUE and LPUE) and independent (RPN) data. The examined error distribution models were gamma, lognormal, tweedie, and hurdle models. For hurdle, positive observations were analyzed assuming a lognormal (hurdle–lognormal) or gamma (hurdle–gamma) error distribution. Based on deviance table analyses and diagnostic checks, the hurdle–lognormal was the statistical model that best satisfied the underlying characteristics of the different data sets. Finally, catch rates (CPUE, LPUE and RPN) of the thornback ray Raja clavata, blackbelly rosefish Helicolenus dactylopterus, and common mora Mora moro from the NE Atlantic (Azores region) were standardized. The analyses confirmed the spatial and temporal nature of their distribution.

Effects of re-specifying the Northern Prawn Fishery bioeconomic model to include banana prawns

The Northern Prawn Fishery (NPF) is one of Australia’s most valuable federally managed fisheries. It has also been a focus for modelling in Australia, with over 40 years of development of bioeconomic, stock assessment and management strategy evaluation models that have been used to varying degrees to inform management. The current management model has been in operation for over 10 years and is used to set the total allowable effort (TAE) for the tiger prawn (Penaeus semisulcatus and P. esculentus) component of the fishery, historically undertaken after a mid-year closure. The success of management has resulted in increased stocks of tiger prawns, and a fishery before the mid-year closure during which fishing for banana prawns was historically the main activity. While the current management model includes the consequences of fishing for tiger prawns during the first season in terms of which weeks are available for fishing for tiger prawns when estimating the TAE, the lack of explicit inclusion of banana prawns in the bioeconomic modelling has resulted in concerns that the TAE may be biased, as effort is allocated in the model to tiger prawns that, in reality, should be applied to banana prawns. The current model was consequently extended to include an explicit banana prawn component, and a range of scenarios undertaken to explore the impact of this on the management advice. It was found that the current model overestimates the optimal level of effort to be applied to tiger prawns except during periods of very low banana prawn abundance. Banana prawns, which vary substantially inter-annually due largely to environmental drivers, were also found to be the key driver of the level of overall fishery profit, suggesting that greater attention to this component of the fishery may be necessary to achieve the objective of maximizing net economic returns from the NPF.

A history and evaluation of catch‐only stock assessment models

AbstractUnderstanding the status of fish stocks is a critical step in ensuring the ecological and economic sustainability of marine ecosystems. However, at least half of global catch and a vast majority of global fisheries lack formal stock assessments, largely due to a lack of sufficient data. Catch data, loosely referring to any catch records be it inclusive of discards or not, are the only type of fishery data available across a wide range of fisheries at a global scale. This has given rise to a long list of so‐called “catch‐only” models, intended to estimate aspects of stock status based primarily on characteristics of a fishery's catch history. In this paper, we review the history, performance and potential of “catch‐only” models to estimate stock biomass status. While individual catch‐only models often report good performance, repeated efforts to examine the performance of these models have consistently found them to be imprecise and biased when applied to new data‐limited fisheries. We demonstrate that a large reason for this is the simple lack of information on stock status contained in the shape of a catch history alone. Off‐the‐shelf use of catch‐only models can lead to poor and biased estimates of stock status, potentially hindering efforts at effective management.

Identifying persistent biomass areas: The case study of the common sole in the northern Iberian waters

The common sole (Solea solea) is a species of flatfish of great commercial relevance in the northern Atlantic Iberian waters. Although many studies investigated the early life-stage of this species, little is known on the adult phases and the size structure of the population. Here we provide an analysis of the spatiotemporal distribution of adult’s common sole occurrence, conditional-to-presence biomass and median length in northern Atlantic Iberian waters using a Bayesian approach and three different spatiotemporal model structures that characterize the fundamental distributional patterns of fish as either persistent, opportunistic or progressive. Results highlight that the distribution of common sole adults in the area has been persistent over the years and that is partially affect by the bathymetry. Three main abundance hotspots were identified coinciding with spatial closure areas restricted to the fishing activity. Additionally, results show that small specimens tend to aggregate in shallower waters, while larger specimens inhabit deeper waters. Understanding the spatiotemporal distribution of the common sole population could improve our knowledge of this species and help to design more accurate biomass index that could be used to calibrate stock assessment models.

Age and growth estimates from three hard parts of the spotted catfish, Arius maculatus (Actinopterygii: Siluriformes: Ariidae), in Songkhla Lake, Thailand’s largest natural lake

The spotted catfish, Arius maculatus (Thunberg, 1792), is a euryhaline fish that is economically important in the Indo-West Pacific. Population dynamics studies and stock assessments of this species have focused on marine stocks, but not those from fresh water. In this study, the age and growth of A. maculatus were, therefore, investigated for the inland stock in Songkhla Lake, Thailand. A total of 213 individuals ranging between 35 and 238 mm TL were used. The length–weight relation indicated positive allometry of this population. Three hard parts (otolith, dorsal- and pectoral-fin spines) were used for aging. The marginal increment ratio confirmed that an annulus was deposited once a year in all three hard parts. All of the samples were aged between 0+ and 6+ years. Verification of age estimates from three readers showed that the otolith was the most suitable part for age estimation. Three growth models (von Bertalanffy, Gompertz, and logistic) were applied in the study. The von Bertalanffy model best described the growth of this fish in Songkhla Lake. The obtained asymptotic length was 290.87 mm TL and the relative growth rate parameter was 0.166 year–1. Our results will be applied as inputs for fish stock assessment models. The obtained growth parameters also can serve as a reference for A. maculatus stocks elsewhere.

Indexing starvation mortality to assess its role in the population regulation of Northern cod

Starvation is a ubiquitous process in nature as all animals depend on finite resources to survive. Limited food resources can lead to starvation-induced mortality and, depending on the scale of the limitation, population-level consequences may emerge. However, attempts to model starvation-induced mortality are rare as it is difficult to isolate the effects of specific sources of natural mortality. Using commonly collected data on body condition, we develop a statistical approach to index starvation-induced mortality by estimating the proportion of individuals experiencing severe emaciation in a fish population. This approach accounts for seasonal patterns in body condition, allowing data collected at any time of year to be included. Using Canada’s Northern cod (Gadus morhua) stock as a case study, we found a positive association between rates of natural mortality estimated by an integrated population model and the index of mortality derived from proportions of cod in poor condition. This index is also associated with the availability of two key prey species, capelin (Mallotus villosus) and Northern shrimp (Pandalus borealis). Taken together, our results indicate that starvation-induced mortality represents an important component of the natural mortality experienced by Northern cod, supporting the idea that insufficient food contributed to the collapse of the stock in the early 1990s and, subsequently, limited its recovery. These implications highlight the practical relevance of explicitly considering prey availability (e.g., through body condition or prey abundance indices) to improve stock assessment modeling and aid the development of actionable ecosystem-based fisheries management strategies.

Data-Limited Stock Assessment for Fish Species Devoid of Catch Statistics: Case Studies for Pampus argenteus and Setipinna taty in the Bohai and Yellow Seas

For many fish stocks, such as Pampus argenteus and Setipinna taty in China, size composition data are more accessible than catch data. Varied results can arise when different length-based stock assessment models are applied to these data, and fishery managers often need to reconcile conflicting estimates of population status. Superensemble modeling, a relatively recent innovation in fish stock assessments commonly used in other fields, may provide an effective solution to resolving uncertainties among the results from multiple length-based models. To verify potential for this approach to improve estimates of population status, we applied ensemble modeling to fit simulated data of P. argenteus and S. taty in the Bohai and Yellow Seas using predictions from a length-based integrated mixed effects (LIME) and length-based spawning potential ratio (LB-SPR) models as covariables in a superensemble model developed in this study. All simulation modeling of P. argenteus and S. taty in the Bohai and Yellow Seas was conducted using the operating model in the R package LIME. Initially, the LIME and LB-SPR performances were tested separately under three scenarios of fishing mortality and recruitment variability (“equilibrium scenario,” “endogenous scenario,” and “one-way base scenario”). Then, estimates of spawning potential ratio (SPR) were combined with the superensemble models (a linear model, a support vector machines, a random forest and a boosted regression tree). We trained our superensemble models with 80% of the simulated data and tested them with the remaining 20%. Our results showed that superensemble modeling substantially improved the estimates of SPR, with support vector machines performing the best at estimating population status: precision improved by 12.7% for S. taty and 8% for P. argenteus on average (namely, median absolute proportional error decreased by 0.127 and 0.08 on average) compared to the individual models. This finding has important implications for fisheries management in the context of species for which catch data are unavailable. Applying the size composition survey data, the results from support vector machines superensemble model suggested that neither S. taty nor P. argenteus in the Bohai Sea in 2019 are overfished, but the stock status of P. argenteus warrants vigilant monitoring.

Preface: Developing the next generation of stock assessment software

• Preface for Special Issue on the next generation of general stock assessment models. • Overview of recommendations for model development. • Stock assessment best practice requires better and more general software.

Spatial and Temporal Variation in Annual versus Biennial Reproductive Cycles in Eastern Bering Sea Snow Crab Chionoecetes opilio

AbstractThe stock assessment model currently used for the management of the fishery for snow crab Chionoecetes opilio in the eastern Bering Sea assumes an annual reproductive cycle. However, temperatures below 1°C can extend the snow crab reproductive cycle to 2 years, which could depress the total reproductive potential of the stock. Our objective was to determine the prevalence of snow crab in the eastern Bering Sea on a biennial reproductive cycle, relative to latitudinal and thermal gradients, and evaluate whether available bottom temperature data could be used to predict the proportions of biennial spawners. Mature female snow crab were collected on the annual National Marine Fisheries Service eastern Bering Sea bottom trawl survey during 2014–2018. Bimodality in the frequency distribution of oocyte area suggests that oocyte area can likely be used to distinguish annual and year‐1 biennial spawners from year‐2 spawners. In 2014, the highest rates of biennial reproduction were found in northern, colder areas in the eastern Bering Sea. In 2015–2018, very low rates of biennial reproduction coincided with warm temperatures across the eastern Bering Sea. This work could help in the estimation of female reproductive potential and allow the development of female‐based reference points for the eastern Bering Sea snow crab fishery.

CMSY 및 BSS 모델을 활용한 한국의 고등어 자원평가 연구

This study attempted to assess the stock status of chub mackerel(Scomber japonicus), which is one of commercially important fish species in South Korea. Recent studies on stock assessment of chub mackerel have been done, but they have several limitations such as setting up prior information and distributions for major variables. In this study, stock assessment models, Monte Carlo method (CMSY) and Bayesian state-space (BSS) method, that can be analyzed even in limited data and based on prior information and distributions with relatively clear evidence were used. To evaluate the stock status of chub mackerel, annual changes in the amount of biomass and the ratio of the amount of biomass in the last year to BSUBMSY/SUB (the biomass which can produce maximum sustainable yield) were estimated. Results showed that the status of chub mackerel was assessed to be

Prediction–variance relation in a state-space fish stock assessment model

AbstractThe state-space assessment model (SAM) is extended by allowing a functional relationship between observation variance and the corresponding prediction. An estimated relationship between observation variance and predicted value for each individual observation allows the model to assign smaller (or larger) variance to predicted larger log-observations. This relation is different from the usual assumption of constant variance of log-observations within age groups. The prediction–variance link is implemented and compared to the usual constant variance assumption for the official assessments of North East Arctic cod and haddock. For both of these stocks, the prediction–variance link is found to give a significant improvement.

Incorporating intra-annual variability in fisheries abundance data to better capture population dynamics

To reduce the risk of overexploitation and the ensuing conservation and socio-economic consequences, fisheries management relies on receiving accurate scientific advice from stock assessments. Biomass dynamics models used in stock assessment tend to rely primarily on indices of abundance and commercial landings data. Standard practice for calculating the indices used in these models typically involves taking averages of survey tow data over large, diverse spatial domains. There is a lot of variability in the choice of methodologies used to propagate index uncertainty into the assessment model, many of which require specifying it through expert knowledge or prior distributions. Here we propose an alternative approach that treats each individual survey tow as an independent estimate of the true underlying biomass in the stock assessment model itself. This reduces information loss and propagates uncertainties into the model directly. A simulation study demonstrates that this approach accurately captures underlying population dynamics and reliably estimates variance parameters. We further demonstrate its utility with data from the Inshore Scallop Fishery of south-west Nova Scotia. Results show significant improvements in parameter estimation over previous models while providing similar predictions of biomass with less uncertainty. This reduced uncertainty can improve the resulting scientific advice and lead to improved decision-making by fisheries managers.

Improved Accuracy for Automated Counting of a Fish in Baited Underwater Videos for Stock Assessment

The ongoing need to sustainably manage fishery resources can benefit from fishery-independent monitoring of fish stocks. Camera systems, particularly baited remote underwater video system (BRUVS), are a widely used and repeatable method for monitoring relative abundance, required for building stock assessment models. The potential for BRUVS-based monitoring is restricted, however, by the substantial costs of manual data extraction from videos. Computer vision, in particular deep learning (DL) models, are increasingly being used to automatically detect and count fish at low abundances in videos. One of the advantages of BRUVS is that bait attractants help to reliably detect species in relatively short deployments (e.g., 1 h). The high abundances of fish attracted to BRUVS, however, make computer vision more difficult, because fish often obscure other fish. We build upon existing DL methods for identifying and counting a target fisheries species across a wide range of fish abundances. Using BRUVS imagery targeting a recovering fishery species, Australasian snapper (Chrysophrys auratus), we tested combinations of three further mathematical steps likely to generate accurate, efficient automation: (1) varying confidence thresholds (CTs), (2) on/off use of sequential non-maximum suppression (Seq-NMS), and (3) statistical correction equations. Output from the DL model was more accurate at low abundances of snapper than at higher abundances (>15 fish per frame) where the model over-predicted counts by as much as 50%. The procedure providing the most accurate counts across all fish abundances, with counts either correct or within 1–2 of manual counts (R2 = 88%), used Seq-NMS, a 45% CT, and a cubic polynomial corrective equation. The optimised modelling provides an automated procedure offering an effective and efficient method for accurately identifying and counting snapper in the BRUV footage on which it was tested. Additional evaluation will be required to test and refine the procedure so that automated counts of snapper are accurate in the survey region over time, and to determine the applicability to other regions within the distributional range of this species. For monitoring stocks of fishery species more generally, the specific equations will differ but the procedure demonstrated here could help to increase the usefulness of BRUVS.