Fisheries Stock Assessment Models Research Articles

‘Slim pickings?’: Extreme large recruitment events may induce density-dependent reductions in growth for Alaska sablefish (Anoplopoma fimbria) with implications for stock assessment

Growth processes mediate survival and fecundity within fish populations, which are fundamental in regulating population dynamics. Therefore, accurate estimates of population scale and sustainable exploitation levels in contemporary fishery stock assessment models rely heavily on understanding and accurately characterizing growth processes. However, empirical studies that relate population-level changes in growth patterns to observable ecosystem and population conditions remain scarce. In particular, few studies directly consider the influence of intra-specific competition (i.e., cohort effects) on growth variability, and its associated implications for stock assessment estimates. Focusing on Alaska sablefish as a case study, we illustrate how multiple unprecedented large recruitment events since 2014 resulted in density-dependent declines in growth on the population-level scale, using a state-space growth model. Furthermore, we demonstrate how incorporating cohort-specific growth variability within the Alaska sablefish stock assessment model resulted in substantial differences in estimates of spawning biomass and recommended harvest levels. Overall, results from this study underscore the significance of cohort effects on growth processes and their implications for stock assessment models and associated harvest recommendations.

Bias in fishery stock assessment models.

Bias in fishery stock assessment models.

A tale of two species: Disaggregating mixed historical catches of two most common skates in the Northeast Pacific Ocean

Historical catch data represent a key source of information for fisheries stock assessment models. Historically, commercial fishery catch statistics were estimated from the portion of catch landed in ports. However, many species with relatively low economic value have not been recorded on a species basis but instead as a part of an aggregate category. Reconstructing component species catch from an aggregate category is a common challenge for fisheries stock assessment efforts around the world. Skates are one group of species with low economic value for which landed catch has not been commonly reported by species. In this paper, we present a novel approach to disaggregate the historical catch of the two most abundant skate species on the West Coast of the United States, longnose skate (Caliraja rhina) and big skate (Beringraja binoculata), landed within the aggregated skate category, in ports of Washington State. We used a combination of fishery-dependent and fishery-independent data sources to account for changes in the spatial extent of the fishery over time, and differences in the depth distribution of these two skate species. While developed to disentangle aggregate catch of longnose and big skates, the approach is not limited to skates on the West Coast of the United States, but can be adapted for any species which landings have been reported within an aggregated category elsewhere.

Lessons to be learned by comparing integrated fisheries stock assessment models (SAMs) with integrated population models (IPMs)

Integrated fisheries stock assessment models (SAMs) and integrated population models (IPMs) are used in biological and ecological systems to estimate abundance and demographic rates. The approaches are fundamentally very similar, but historically have been considered as separate endeavors, resulting in a loss of shared vision, practice and progress. We review the two approaches to identify similarities and differences, with a view to identifying key lessons that would benefit more generally the overarching topic of population ecology. We present a case study for each of SAM (snapper from the west coast of New Zealand) and IPM (woodchat shrikes from Germany) to highlight differences and similarities. The key differences between SAMs and IPMs appear to be the objectives and parameter estimates required to meet these objectives, the size and spatial scale of the populations, and the differing availability of various types of data. In addition, up to now, typical SAMs have been applied in aquatic habitats, while most IPMs stem from terrestrial habitats. SAMs generally aim to assess the level of sustainable exploitation of fish populations, so absolute abundance or biomass must be estimated, although some estimate only relative trends. Relative abundance is often sufficient to understand population dynamics and inform conservation actions, which is the main objective of IPMs. IPMs are often applied to small populations of conservation concern, where demographic uncertainty can be important, which is more conveniently implemented using Bayesian approaches. IPMs are typically applied at small to moderate spatial scales (1 to 104 km2), with the possibility of collecting detailed longitudinal individual data, whereas SAMs are typically applied to large, economically valuable fish stocks at very large spatial scales (104 to 106 km2) with limited possibility of collecting detailed individual data. There is a sense in which a SAM is more data- (or information-) hungry than an IPM because of its goal to estimate absolute biomass or abundance, and data at the individual level to inform demographic rates are more difficult to obtain in the (often marine) systems where most SAMs are applied. SAMs therefore require more 'tuning' or assumptions than IPMs, where the 'data speak for themselves', and consequently techniques such as data weighting and model evaluation are more nuanced for SAMs than for IPMs. SAMs would benefit from being fit to more disaggregated data to quantify spatial and individual variation and allow richer inference on demographic processes. IPMs would benefit from more attempts to estimate absolute abundance, for example by using unconditional models for capture-recapture data.

Addressing complex fleet structure in fishery stock assessment models: Accounting for a rapidly developing pot fishery for Alaska sablefish (Anoplopoma fimbria)

Fisheries management operates under uncertainty, often driven by the dynamic nature of marine ecosystems and associated fisheries. Stock assessment models, which form the scientific basis of decision-making in fisheries management, strive for realistic representations of biological and fishery processes. However, data limitations and knowledge gaps necessitate simplifying assumptions for representing these complex bio-socioeconomic systems, which can increase uncertainty in the assessment process. Addressing time-varying fishery dynamics (i.e., due to regulatory changes or alterations in harvester behavior) is a common and particularly challenging problem for stock assessment models. Time-varying fishery selectivity is widely utilized to address changes in fishery dynamics but may not be adequate when regulatory changes substantially alter gear usage and associated assessment fleet structures. We explore the implications of accounting for, or ignoring, complex temporal changes in fleet structure and selectivity within stock assessment models by utilizing a recent and high-value case study, Alaska sablefish (Anoplopoma fimbria). Our findings demonstrate that the treatment of fleet structure (i.e., adding fleet complexity to account for gear transitions) did not greatly influence estimates of spawning biomass trajectories. However, associated selectivity assumptions had substantial impacts on sustainable harvest recommendations. We recommend that the treatment of fleet structure and associated selectivity assumptions should incorporate a priori considerations and subject-matter expertise of fishery and biological dynamics to ensure pragmatic and appropriate model parameterizations. Moreover, we advocate for multi-fleet models as a useful diagnostic tool for validating model estimates from single-fleet assessments when uncertainty in fleet dynamics exist.

Inclusion of ageing error and growth variability using a bootstrap estimation of age composition and conditional age-at-length input sample size for fisheries stock assessment models

Statistical catch-at-age assessment models used for fisheries management integrate multiple sources of information that are statistically weighted in a joint likelihood framework; the relative statistical weighting between these sources of information is an important, yet often subjective aspect of stock assessment. Input sample size (ISS) is a quantity that is used to statistically weight composition data in these types of models. Both design-based bootstrap and model-based estimators have been proposed, however, these methods to determine ISS do not explicitly account for uncertainty from ageing error and growth variability that are inherent to expanded age composition and conditional age-at-length data. In this study, we evaluate the impact of including ageing error and growth variability within bootstrap methods that estimate age composition and conditional age-at-length ISS. We find that for all the stocks evaluated the ISS determined from bootstrap methods decreased as these addition sources of uncertainty were included. The decrease in ISS was species type specific, but generally decreased up to 40% when ageing error was introduced, up to 50% when growth variability was included, and up to 60% when both sources of uncertainty were included. These results indicate that there is more variability within age composition or conditional age-at-length data than would be accounted for with ISS estimates that do not include these sources of uncertainty. The method and results provided here allow for assessment scientists to statistically weight age composition and conditional age-at-length with ISS that takes into account ageing error and growth variability from either fishery-independent or fishery-dependent sources. Including these sources of uncertainty improves bootstrap estimates of ISS to capture all the sources of variability in age composition and conditional age-at-length and will subsequently improve stock assessment model quality.

Reductions in sampling effort for fishery-independent age and length composition: balancing sampling efficiency, data uncertainty, and workforce health

Unavoidable survey effort reduction has become a reality that must be accounted for in fisheries stock assessment. In addition, negative consequences to survey staff health due to repetitive motion injuries are becoming increasingly costly for managing agencies. In this study, we evaluated the outcomes of reductions in age and length data used in fisheries stock assessment models. The main goal was to determine whether sampling can be reduced to a level that does not excessively increase data uncertainty, yet provides a reduction in repetitive motions that can cause injury to survey staff. We found that reducing length sampling to a maximum of 100–150 fish sampled per haul (either sex-specific or combined sex) provides length composition data for which the uncertainty is not appreciably increased, and it has minimal effect on the uncertainty in age composition data that is subsequently expanded from this subsampled length frequency data. The method employed here, and the results presented, can aid management agencies to balance the magnitude of data collection and subsequent consequences to fisheries stock assessment models.

Process and sampling variance within fisheries stock assessment models: estimability, likelihood choice, and the consequences of incorrect specification

Abstract Increasingly, mixed-effect fishery stock assessment models are being developed where deviations about functional forms of different processes are modelled as random effects and the extent of variance is estimated internal to the model. Concurrently, sampling variance parameters associated with likelihoods for fitting composition data within fisheries assessments are more often being estimated internal to the model as well. We examine the performance of stock assessment models when multiple process variance and sampling variance terms are simultaneously estimated within assessment models. We specifically examine how assessment performance is affected by the choice of composition likelihood, the degree of overdispersion in composition data, overparameterization, and modelling variation on the wrong process. In doing so, we build a simulation containing overdispersion and correlations in composition data, directional variation in catchability and/or selectivity, and estimation models which include random effects and composition likelihoods with theoretically estimable variances. Results suggest that with standard data available in fisheries assessments, process variance parameters associated with some commonly employed methods and sampling variance parameters can be simultaneously estimated internal to an assessment, and performance greatly improves with increased composition data. Our results also suggest little downside to overparameterization of selectivity and catchability when the true process is not time-varying, which largely agrees with previous research. However, when a process is truly time-varying and the assessment models time-variation on a different process, namely when selectivity is time-varying and instead natural mortality is modelled as potentially time-varying, we find a risk of severe increases in bias and decreases in confidence interval coverage for assessed quantities. This bias and decrease in coverage could, however, be partially mitigated by also modelling time-variation on the correct process.

An age‐ and length‐structured statistical catch‐at‐length model for hard‐to‐age fisheries stocks

AbstractIt is challenging in fisheries stock assessment to estimate cohort dynamics from length‐based data for hard‐to‐age stocks, and existing approaches, for example, age‐structured catch‐at‐length models (ACL) are unable to account for length‐dependent processes within each cohort. Fisheries‐dependent data are usually considered the default input to stock assessment models. However, with widespread recognition of the uncertainty of fisheries‐dependent data and the increasing availability of high‐quality survey data, a new situation emerges in some fisheries where a stock assessment model based only on survey data can provide good estimation of population dynamics. We develop an age‐ and length‐structured statistical catch‐at‐length model (ALSCL) to estimate age‐based dynamics from survey catch‐at‐length data. This approach also provides a good basis to then integrate fisheries‐dependent data in the model. ALSCL can explicitly include length‐dependent mortality and growth within each cohort by simultaneously tracking the three‐dimensional dynamics across time, age, and length. We first use simulations of yellowtail flounder (Limanda ferruginea, Pleuronectidae) and bigeye tuna (Thunnus Obesus, Scombridae) to demonstrate that ALSCL outperforms ACL by providing more accurate estimates of age‐based population dynamics when length‐dependent processes are important. Next, we apply ALSCL to estimate the cohort dynamics of female yellowtail flounder on the Grand Bank off Newfoundland using survey catch‐at‐length, weight‐at‐length, and maturity‐at‐length data. We consider ALSCL as a hybrid between ACL and length‐structured stock assessment models that keeps the advantages of both, and its ability to simultaneously track age and length dynamics is an important step toward the next‐generation of fisheries stock assessment models.

Spatial and temporal variability in somatic growth in fisheries stock assessment models: evaluating the consequences of misspecification

AbstractDistinct types of fish species experience variation in somatic growth rates over their life span; however, growth has historically been assumed to be invariant across time and space in integrated analysis-based stock assessment. A few previous studies have reported biased and imprecise assessment model outcomes when variability in somatic growth was ignored. In this study, we used a simulation-estimation framework to expand previous analyses and to examine the consequences of ignoring or incorporating spatial and temporal (year- and cohort-specific) variability in somatic growth in stock assessment models. The study included three life history types: small pelagic (e.g. sardine), gadids (e.g. cod), and long-lived (e.g. rockfish). In general, ignoring any type of variability in somatic growth led to biased and imprecise estimates of stock spawning biomass and management quantities. Unequal distribution of fishing mortality across space had large impacts on the performance of estimation models as well. Conversely, accounting for somatic growth variability, either by including an environmental index, estimating annual deviates, or implementing a spatially explicit model, produced unbiased and precise results. This study shows that somatic growth variability might produce large effects in stock assessments when ignored and provides pertinent information for stock assessment best practice guidelines.

The longer the better? Trade‐offs in fisheries stock assessment in dynamic ecosystems

AbstractEnvironmental change and anthropogenic activity, alone or in combination, may cause vital rates of fish populations to exhibit low‐frequency, large‐magnitude variation leading to non‐stationary population processes in inherently dynamic ecosystems. It remains unclear why, when and, so, whether, non‐stationary population processes should be taken into account in fisheries stock assessment models. Here, we clarify the necessity and conditions for including non‐stationary population processes in stock assessment models. Specifically, the convention of treating population vital rates with constant parameters might be unreliable under regime shifts characterized by low‐frequency, large‐magnitude variation in drivers of fish population dynamics. We hypothesized and simulated effects of a U‐shaped trade‐off between the length of time‐series data and model estimation error for fish populations exhibiting non‐stationary dynamics. In the case of low‐frequency, large‐magnitude variation, when measurement errors were small, the effects of non‐stationary population processes were stronger than otherwise. Including time‐varying parameters of population vital rates in the assessment model resolved this dilemma as anticipated, albeit at the cost of increased model complexity, suggesting that accounting for non‐stationarity with time‐varying parameters alone may not be sufficient to improve stock assessments and other approaches should also be considered.

Implementing two-dimensional autocorrelation in either survival or natural mortality improves a state-space assessment model for Southern New England-Mid Atlantic yellowtail flounder

Survival is an important population process in fisheries stock assessment models and is typically treated as deterministic. Recently developed state-space assessment models can estimate stochastic deviations in survival, which represent variability in some ambiguous combination of natural mortality (M), fishing mortality (F), and migration. These survival deviations are generally treated as independent by age and year, despite our understanding that many population processes can be autocorrelated and that not accounting for autocorrelation can result in notable bias. We address these concerns, as well as the strong retrospective pattern found in the last assessment of Southern New England yellowtail flounder (Limanda ferruginea), by incorporating two-dimensional (2D, age and year) first-order autocorrelation in survival and M. We found that deviations were autocorrelated among both years (0.53 ± 0.09, 0.63 ± 0.16) and ages (0.33 ± 0.12, 0.40 ± 0.16) when estimated for survival or M, respectively. Models with 2D autocorrelation on survival or M fit the data better and had reduced retrospective pattern than models without autocorrelation. The best fit model included 2D autocorrelated deviations in survival as well as independent deviations in M and altered estimates of spawning stock biomass by 18 % and F by 21 % in model years. In short-term projections with F = 0, including 2D autocorrelation in survival or M reduced spawning stock biomass by 48 %. We conclude that incorporating 2D autocorrelated variation in survival or M could improve the assessment of Southern New England yellowtail flounder in terms of model fit and consistency of biomass projections.

Improving the robustness of fisheries stock assessment models to outliers in input data

Outliers caused by atypical observation error often occur in fishery data. These outliers have an adverse effect on the parameter estimation for fishery stock assessment models. We tested a robust distribution for identifying and removing outliers from fishery data. We conducted a simulation study in which a surplus production model was used to mimic fishery population dynamics and outliers caused by atypical observation error were imposed in the biomass index data. The method performed well by effectively identifying the real outliers and avoiding defining other data points as outliers. By removing the detected outliers and fitting the model with the remaining data points, the accuracy of the parameter estimation was improved. We discussed the precautions of applying this method and its potential applicability in other fishery stock assessment models.

Examining the relationship between morphological variation and modeled broadband scattering responses of reef-associated fishes from the Southeast United States

Non-lethal methods are sought to provide indices or measure of absolute abundance to fit fishery stock assessment models, including those computed for reef fishes. Calibrated scientific echosounders provide the foundation for stock assessments in low-diversity ecosystems but are hampered by complicated target strength responses in high diversity systems like coral and rocky reefs. Newly available broadband echosounders present advantages over narrowband predecessors including increased spatial resolution, increased signal to noise and spectral resolution; however, a better understanding of the sound scattering properties is necessary for operational use in surveys of diverse reef fish assemblages. To gain insight into the feasibility of implementing broadband techniques, we first model the frequency dependent backscatter of ecologically and commercially important species; results will aid our understanding of the backscattering as well as provide a supporting dataset to develop algorithms for classification of species/functional groups. Computed-tomography (CT) scans were performed on 149 individuals across reef fish species from the northeastern Gulf of Mexico to generate three-dimensional swim bladder models for estimation of morphometrics to input into numerical scattering models. Principal component analysis of the swim bladder morphology indicated significant variation among species; however, closely related species (confamilials) were not different. We employed boundary element method modeling to examine the frequency-dependent backscatter responses across a range of fish orientations (+/- 45°). Comparisons of frequency-dependent backscatter revealed strongest similarities between closely related species and greatest differences between more distantly related species or species groups. Patterns in the spectra have the potential to be used to classify targets toward species (or species group) for estimates of density from marine ecosystem acoustic surveys. Given that broadband echosounders are not yet commonplace, and many surveys in the region utilize narrowband echosounders, we also developed narrowband target strength – length relationships for six of the reef fish species with model fits ranging widely (r2 = 0.05 to 0.93). The potential utility of broadband echosounders to enhance ecosystem studies, or to develop fisheries-independent indices of abundance for use in stock assessment is discussed.

Length–weight relationship and condition factor for 7 loliginid squid species in Mexican waters

The length–weight relationship and the condition factor are key parameters in stock assessment. Despite the fisheries potential loliginid squids have in Mexico, the group is scarcely known and no formal management has been implemented. A total of 1,105 squids captured on the Pacific and Atlantic coasts of Mexico were measured to estimate length–weight relations and condition factors. Except for Lolliguncula brevis and Lolliguncula argus, all species showed negative allometric growth. The 2 species with isometric growth showed a poor condition, according to Fulton’s condition factor, and the rest of the species were in good health, according to the relative condition factor. The type of growth exhibited by the species identified in this study could be associated with their swimming habits or otherwise be affected by the same variables that determine the type of growth of other squid found in Mexican waters. The parameters evaluated in this research are now available for incorporation in further fishery stock assessment models for loliginid squids.

High levels of site fidelity in movement patterns of Southern Rock Lobster (Jasus edwardsii) in Victoria, Australia

ABSTRACT Movement patterns of the Southern Rock Lobster Jasus edwardsii in Victoria, Australia were investigated from 8,533 tag-recapture events across a 20-year period (1992–2012). In total, 83% of lobsters were recaptured within 1 km of their tagging site and 93% within 5 km. While largely resident, elevated movements were observed within specific regions, with the overall direction of movement being from inshore to offshore areas. Movement was not impacted by lobster sex, size, or reproductive stage. The fishery for Southern Rock Lobster within Victoria is currently managed into two zones with separate total allowable commercial catches in each area. Given the high levels of site fidelity, our findings suggest that management of the resource at this spatial scale is appropriate and that consideration of significant movement between zones is not warranted in fishery stock assessment models.

Environmental indices for spanner crab (Ranina ranina) catch rates depend on regional oceanographic features

In the Australian spanner crab (Ranina ranina) fishery, management and industry are looking for improvements to the existing indicators of stock abundance. Prior research linked several oceanographic indices to the catchability of spanner crabs; however, it was unclear whether nearshore (e.g. river-runoff) or region-specific oceanographic features (e.g. eddies and the East Australian Current) are responsible for these effects on catch rates. Using satellite remote sensing and fishery-independent survey data, we analysed the influence of oceanographic and environmental indices on spanner crab catch rates in southern Queensland. Outputs from Generalised Additive Models (GAM) show that catch rates exhibit a large amount of variability between different regions of the fishery, with highest catch rates at fishing grounds within 40 km from the shelf break. Offshore oceanic waters, transported into various regions by different oceanographic processes, were linked to an increase in catch rates. Lower concentrations of surface chlorophyll a were also correlated with higher catch rates, but only in survey regions exposed to the effects of the Fraser Gyre and at the mouth of bays. Overall, results highlighted that the effects of environmental indices on catch rates were not homogeneous across the fishery. Rather, relationships were linked to region-specific (<100 km), highly dynamic coastal and oceanographic features that dominate different survey regions. Outcomes from this work show that the spatial variability of oceanographic features should be taken into consideration before incorporating oceanographic indices in fishery stock assessment models.

Integrating underwater video into traditional fisheries indices using a hierarchical formulation of a state-space model

Indices of abundance are commonly used in fisheries stock assessment models to represent trends in population size over time; however, an index can misrepresent such trends when catchability varies, sampling gears change or spatial sampling frames shift. Here we develop a state-space model in a Bayesian framework that combines both chevron trap catches and video counts into a single integrated index. The modeling approach accounts for variation in sampling efficiency (catchability) of both sampling gears and adjusts for aspects of changes in the spatial sampling frame (sampling intensity and spatial coverage) through time due to monitoring program development. We validate the model using a simulation study and then demonstrate its utility using data on vermilion snapper Rhomboplites aurorubens from the period 1990–2016. The index suggests high variation in the abundance of vermilion snapper, particularly for years previous to 2000 and a systematic decline in abundance between the early 1990s and 2016. This pattern culminates (2016) with vermilion snapper at about 16% of their average early 1990s abundance which is a stronger decline than is indicated by the current index used for stock assessment of the species.

From data compilation to model validation: a comprehensive analysis of a full deep-sea ecosystem model of the Chatham Rise

The Chatham Rise is a highly productive deep-sea ecosystem that supports numerous substantial commercial fisheries, and is a likely candidate for an ecosystem based approach to fisheries management in New Zealand. We present the first end-to-end ecosystem model of the Chatham Rise, which is also to the best of our knowledge, the first end-to-end ecosystem model of any deep-sea ecosystem. We describe the process of data compilation through to model validation and analyse the importance of knowledge gaps with respect to model dynamics and results. The model produces very similar results to fisheries stock assessment models for key fisheries species, and the population dynamics and system interactions are realistic. Confidence intervals based on bootstrapping oceanographic variables are produced. The model components that have knowledge gaps and are most likely to influence model results were oceanographic variables, and the aggregate species groups ‘seabird’ and ‘cetacean other’. We recommend applications of the model, such as forecasting biomasses under various fishing regimes, include alternatives that vary these components.

Enhanced global optimization methods applied to complex fisheries stock assessment models

Statistical fisheries models are frequently used by researchers and agencies to understand the behavior of marine ecosystems or to estimate the maximum acceptable catch of different species of commercial interest. The parameters of these models are usually adjusted through the use of optimization algorithms. Unfortunately, the choice of the best optimization method is far from trivial. This work proposes the use of population-based algorithms to improve the optimization process of the Globally applicable Area Disaggregated General Ecosystem Toolbox (Gadget), a flexible framework that allows the development of complex statistical marine ecosystem models. Specifically, parallel versions of the Differential Evolution (DE) and the Particle Swarm Optimization (PSO) methods are proposed. The proposals include an automatic selection of the internal parameters to reduce the complexity of their usage, and a restart mechanism to avoid local minima. The resulting optimization algorithms were called PMA (Parallel Multirestart Adaptive) DE and PMA PSO respectively. Experimental results prove that the new algorithms are faster and produce more accurate solutions than the other parallel optimization methods already included in Gadget. Although the new proposals have been evaluated on fisheries models, there is nothing specific to the tested models in them, and thus they can be also applied to other optimization problems. Moreover, the PMA scheme proposed can be seen as a template that can be easily applied to other population-based heuristics.