Search

Multispecies models have existed in a fisheries context since at least the 1970s, but despite much exploration, advancement, and consideration of multispecies models, there remain limited examples of their operational use in fishery management. Given that species and fleet interactions are inherently multispecies problems and the push towards ecosystem-based fisheries management, the lack of more regular operational use is both surprising and compelling. We identify impediments hampering the regular operational use of multispecies models and provide recommendations to address those impediments. These recommendations are: (1) engage stakeholders and managers early and often; (2) improve messaging and communication about the various uses of multispecies models; (3) move forward with multispecies management under current authorities while exploring more inclusive governance structures and flexible decision-making frameworks for handling tradeoffs; (4) evaluate when a multispecies modelling approach may be more appropriate; (5) tailor the multispecies model to a clearly defined purpose; (6) develop interdisciplinary solutions to promoting multispecies model applications; (7) make guidelines available for multispecies model review and application; and (8) ensure code and models are well documented and reproducible. These recommendations draw from a global assemblage of subject matter experts who participated in a workshop entitled “Multispecies Modeling Applications in Fisheries Management”.

Environmental conditions can create spatial and temporal variability in growth and distribution processes, yet contemporary stock assessment methods often do not explicitly address the consequences of these patterns. For example, stock assessments often assume that body weight-at-age (i.e. size) is constant across the stocks’ range, and may thereby miss important spatio-temporal patterns. This is becoming increasingly relevant given climate-driven distributional shifts, because samples for estimating size-at-age can be spatially unbalanced and lead to biases when extrapolating into unsampled areas. Here, we jointly analysed data on the local abundance and size of walleye pollock (Gadus chalcogrammus) in the Bering Sea, to demonstrate a tractable first step in expanding spatially unbalanced size-at-age samples, while incorporating fine-scale spatial and temporal variation for inclusion in stock assessments. The data come from NOAA’s bottom trawl survey data and were evaluated using a multivariate spatio-temporal statistical model. We found extensive variation in size-at-age at fine spatial scales, though specific patterns differed between age classes. In addition to persistent spatial patterns, we also documented year-to-year differences in the spatial patterning of size-at-age. Intra-annual variation in the population-level size-at-age (used to generate the size-at-age matrix in the stock assessment) was largely driven by localized changes in fish size, while shifts in species distribution had a smaller effect. The spatio-temporal size-at-age matrix led to marginal improvement in the stock assessment fit to the survey biomass index. Results from our case study suggest that accounting for spatially unbalanced sampling improved stock assessment consistency. Additionally, it improved our understanding on the dynamics of how local and population-level demographic processes interact. As climate change affects fish distribution and growth, integrating spatiotemporally explicit size-at-age processes with anticipated environmental conditions may improve stock-assessment forecasts used to set annual harvest limits.

Generating accurate data for stock assessments is resource-demanding, necessitating periodic evaluation of survey sampling designs and potential impacts on stock assessments. We developed a framework for bootstrapped resampling of survey age data and calculation of input sample sizes as a function of among-bootstrap variance in age compositions. Data from this bootstrap estimator were then used to evaluate the influence of alternative sampling rates and methods on uncertainty in estimates of the overfishing limit (OFL) calculated using stock assessment models. For dusky rockfish ( Sebastes variabilis) and Pacific ocean perch ( Sebastes alutus), a 10% decrease in the number of tows sampled upon led to a predicted 5%–6% increase in the CV of OFL (log–log slope = −0.576 to −0.486), which was greater than the 0%–2% increase from a 10% decrease in otoliths-per-tow (log–log slope = −0.238 to −0.029). Application of this approach across all stocks monitored in the survey of interest is required to identify which stocks ( i) benefit the most from increased sampling of ageing structures or ( ii) cost the least in terms of OFL uncertainty owing to reduced sampling.

The spatial distributions of marine fish populations are influenced by environmental conditions, intrinsic properties of the populations, and prior distribution. The influence of these factors may not be consistent across age classes. For this study, age composition estimates for walleye pollock ( Gadus chalcogrammus) collected on bottom-trawl surveys in the Bering Sea were used to estimate range correlation indices, population centers of gravity, and effective area occupied. Age-specific density maps suggested a circular ontogenic migration during the summer feeding season, with the youngest and oldest groups most broadly distributed. Range correlation analysis among age groups and year classes provided clear evidence of a population cohort effect in the spatial distribution of the population. Variance decomposition analysis indicated that the variance in the spatial distribution of age groups during summer was influenced by the initial distribution of that cohort as recruits. Model-based analyses showed that extrinsic temperature variables affected the youngest and oldest age classes the most, but provided no indication of age-related effects for intrinsic population factors. This study showed that both cohort and age-specific factors are important drivers of spatial distribution.

Marine population modeling, which underpins the scientific advice to support fisheries interventions, is an active research field with recent advancements to address modern challenges (e.g., climate change) and enduring issues (e.g., data limitations). Based on discussions during the 'Land of Plenty' session at the 2021 World Fisheries Congress, we synthesize current challenges, recent advances, and interdisciplinary developments in biological fisheries models (i.e., data-limited, stock assessment, spatial, ecosystem, and climate), management strategy evaluation, and the scientific advice that bridges the science-policy interface. Our review demonstrates that proliferation of interdisciplinary research teams and enhanced data collection protocols have enabled increased integration of spatiotemporal, ecosystem, and socioeconomic dimensions in many fisheries models. However, not all management systems have the resources to implement model-based advice, while protocols for sharing confidential data are lacking and impeding research advances. We recommend that management and modeling frameworks continue to adopt participatory co-management approaches that emphasize wider inclusion of local knowledge and stakeholder input to fill knowledge gaps and promote information sharing. Moreover, fisheries management, by which we mean the end-to-end process of data collection, scientific analysis, and implementation of evidence-informed management actions, must integrate improved communication, engagement, and capacity building, while incorporating feedback loops at each stage. Increasing application of management strategy evaluation is viewed as a critical unifying component, which will bridge fisheries modeling disciplines, aid management decision-making, and better incorporate the array of stakeholders, thereby leading to a more proactive, pragmatic, transparent, and inclusive management framework-ensuring better informed decisions in an uncertain world.The online version contains supplementary material available at 10.1007/s11160-022-09726-7.

Shifts in the distribution of groundfish species as oceans warm can complicate management efforts if species distributions expand beyond the extent of existing scientific surveys, changing the proportion of groundfish available to any one survey each year. We developed the first-ever model-based biomass estimates for three Bering Sea groundfishes (walleye pollock (Gadus chalcogrammus), Pacific cod (Gadus macrocephalus), and Alaska plaice (Pleuronectes quadrituberculatus)) by combining fishery-independent bottom trawl data from the U.S. and Russia in a spatiotemporal framework using Vector Autoregressive Spatio-Temporal (VAST) models. We estimated a fishing-power correction to calibrate disparate data sets and the effect of an annual oceanographic index to explain variation in groundfish spatiotemporal density. Groundfish densities shifted northward relative to historical densities, and high-density areas spanned the international border, particularly in years warmer than the long-term average. In the final year of comprehensive survey data (2017), 49%, 65%, 47% of biomass was in the western and northern Bering Sea for pollock, cod, and plaice, respectively, suggesting that availability of groundfish to the more regular eastern Bering Sea survey is declining. We conclude that international partnerships to combine past data and coordinate future data collection are necessary to track fish as they shift beyond historical survey areas.

Many marine fish species are widely distributed over large areas. Failing to acknowledge that such species may be composed of distinct populations may result in overestimation of the stock's true harvest potential. To avoid overexploitation, ways to identify population structuring are therefore needed. In this study, we developed and applied a statistical approach to identify biologically relevant population boundaries for a widely distributed marine fish species, European sprat (Sprattus sprattus). Specifically, we compiled and standardized multiple trawl-survey data sets and used a range of statistical tools to assess whether the current management boundaries adequately account for potential population structuring. Our results demonstrate regional differences in spatial abundance patterns, temporal dynamics and population demographics. These findings are in line with recent genetic studies of sprat, indicating reproductive isolation between the Baltic Sea/Kattegat and a larger cluster containing the North-, Irish-, Celtic Sea, and Bay of Biscay. Since relying on routinely collected survey data, our statistical approach can be a cost-effective complement to population genetic methods for detecting population structuring. These can be used to guide spatial management efforts and ensure sustainable exploitation, especially under climate change and the expected changes in species distributions across current management borders.

Aim There has been a wide interest in the effect of biotic interactions on species' occurrences and abundances at large spatial scales, coupled with a vast development of the statistical methods to study them. Still, evidence for whether the effects of within-trophic-level biotic interactions (e.g. competition and heterospecific attraction) are discernible beyond local scales remains inconsistent. Here, we present a novel hypothesis-testing framework based on joint dynamic species distribution models and functional trait similarity to dissect between environmental filtering and biotic interactions. Location France and Finland. Taxon Birds. Methods We estimated species-to-species associations within a trophic level, independent of the main environmental variables (mean temperature and total precipitation) for common species at large spatial scale with joint dynamic species distribution (a multivariate spatiotemporal delta model) models. We created hypotheses based on species' functionality (morphological and/or diet dissimilarity) and habitat preferences about the sign and strength of the pairwise spatiotemporal associations to estimate the extent to which they result from biotic interactions (competition, heterospecific attraction) and/or environmental filtering. Results Spatiotemporal associations were mostly positive (80%), followed by random (15%), and only 5% were negative. Where detected, negative spatiotemporal associations in different communities were due to a few species. The relationship between spatiotemporal association and functional dissimilarity among species was negative, which fulfils the predictions of both environmental filtering and heterospecific attraction. Main conclusions We showed that processes leading to species aggregation (mixture between environmental filtering and heterospecific attraction) seem to dominate assembly rules, and we did not find evidence for competition. Altogether, our hypothesis-testing framework based on joint dynamic species distribution models and functional trait similarity is beneficial in ecological interpretation of species-to-species associations from data covering several decades and biogeographical regions. peerReviewed

Weighting data appropriately in stock assessment models is necessary to diagnose model mis-specification, estimate uncertainty, and when combining data sets. Age- and length-composition data are often fitted using a multinomial distribution and then reweighted iteratively, and the Dirichlet-multinomial (“DM”) likelihood provides a model-based alternative that estimates an additional parameter and thereby “self-weights” data. However, the DM likelihood requires specifying an input sample size (ninput), which is often unavailable and results are sensitive to ninput. We therefore introduce the multivariate-Tweedie (MVTW) as alternative with three benefits: (1) it can identify both overdispersion (downweighting) or underdispersion (upweighting) relative to the ninput; (2) proportional changes in ninput are exactly offset by parameters; and (3) it arises naturally when expanding data arising from a hierarchical sampling design. We use an age-structured simulation to show that the MVTW (1) can be more precise than the DM in estimating data weights, and (2) can appropriately upweight data when needed. We then use a real-world state-space assessment to show that the MVTW can easily be adapted to other software. We recommend that stock assessments explore the sensitivity to specifying DM, MVTW, and logistic-normal likelihoods, particularly when the DM estimates an effective sample size approaching ninput.

Spatial management has been adopted worldwide to mitigate habitat impacts while achieving fisheries management objectives. However, there is little theory or practice for predicting the impact of spatial regulations on future fishery production; this would provide scientific basis for greater flexibility in fisheries management when balancing fishery and conservation goals. We propose that predicting changes in fishery production resulting from human activities within specific habitats is a “Grand Challenge” for habitat science in the coming decade(s). We then outline three difficulties in resolving this Grand Habitat Challenge, including: (i) stage-structured responses to habitat impacts, (ii) nonlocal responses, and (iii) mechanistic associations among habitat variables. We next discuss analytical approaches to address each difficulty, respectively: (i) ongoing developments for spatial demographic models; (ii) individual movement models and rank-reduction approaches to identify regional variability; (iii) causal analysis involving structural equation models. We demonstrate nonlocal effects in detail using a diffusion-taxis movement model applied to sablefish (Anoplopoma fimbria) in the Gulf of Alaska and discuss all three approaches for deep-sea corals. Despite isolated progress to resolve individual difficulties, we argue that resolving this Grand Habitat Challenge will require a coordinated commitment from science agencies worldwide.

Abundance indices from scientific surveys are key stock assessment inputs, but when the availability of fish varies in space and time, the estimated indices and associated uncertainties do not accurately reflect changes in population abundance. For example, indices for many semi-pelagic species rely on acoustic and bottom trawl gear that differ in water column coverage, and so spatiotemporal trends in fish vertical distribution affect the availability of fish to each gear type. The gears together cover the whole water column, and so in principle allow estimation of more accurate, combined indices of the whole population. Here, we extend previous methods and develop a vertically integrated index, which accounts for spatiotemporal correlation and works with data unbalanced spatially or unpaired from distinct surveys. Using eastern Bering Sea walleye pollock (Gadus chalcogrammus) as an example, we identified clear spatial and temporal patterns in vertical distribution and gear availability from 2007 to 2018. Estimated acoustic annual vertical availability ranged from 0.339 to 0.888 among years, and from 0.588 to 0.911 for the bottom trawl survey. Our results highlight the importance of accounting for the spatiotemporal and vertical distribution of semi-pelagic fish to estimate more accurate indices, and provide important context for gear availability.

Diet-based annual biomass indices can potentially use predator stomach contents to provide information about prey biomass and may be particularly useful for species that are otherwise poorly sampled, including ecologically important forage fishes. However, diet-based biomass indices may be sensitive to underlying ecological dynamics between predators and prey, such as predator functional responses and changes in overlap in space and time. To evaluate these factors, we fit spatio-temporal models to stomach contents of five Atlantic herring (Clupea harengus) predators and survey catch data for predators and Atlantic herring. We identified drivers of variation in stomach contents, evaluated spatial patterns in stomach content data, and produced predator-specific indices of seasonal Atlantic herring biomass. After controlling for spatio-temporal processes and predator length, diet-based indices of biomass shared similar decadal trends but varied substantially between predators and seasons on shorter time scales. Diet-based indices reflected prey biomass more than prey availability, but weak correlations indicated that not all biological processes were controlled for. Results provide potential guidance for developing diet-based biomass indices and contribute to a body of evidence demonstrating the utility of predator diet data to provide information about relative prey biomass.

Developing Species Distribution Models (SDM) for marine exploited species is a major challenge in fisheries ecology. Classical modelling approaches typically rely on fish research survey data. They benefit from a standardized sampling design and a controlled catchability, but they usually occur once or twice a year and they may sample a relatively small number of spatial locations. Spatial monitoring of commercial data (based on logbooks crossed with Vessel Monitoring Systems) can provide an additional extensive data source to inform fish spatial distribution. We propose a spatial hierarchical framework integrating both data sources while accounting for preferential sampling (PS) of commercial data. From simulations, we demonstrate that PS should be accounted for in estimation when PS is actually strong. When commercial data far exceed scientific data, the later bring little information to spatial predictions in the areas sampled by commercial data, but bring information in areas with low fishing intensity and provide a validation dataset to assess the integrated model consistency. We applied the framework to three demersal species (hake, sole, and squids) in the Bay of Biscay that emphasize contrasted PS intensity and we demonstrate that the framework can account for several fleets with varying catchabilities and PS behaviours.

Climate change is rapidly affecting the seasonal timing of spatial demographic processes. Consequently, resource managers require information from models that simultaneously measure seasonal, interannual, and spatial variation. We present a spatio-temporal model that includes annual, seasonal, and spatial variation in density and then highlight two important uses: (i) standardizing data that are spatially unbalanced within multiple seasons and (ii) identifying interannual changes in seasonal timing (phenology) of population processes. We demonstrate these uses with two contrasting case studies: three bottom trawl surveys for yellowtail flounder (Limanda ferruginea) in the Northwest Atlantic Ocean from 1985 to 2017 and pelagic tows for copepodite stage 3+ copepod (Calanus glacialis/marshallae) densities in the eastern Bering Sea from 1993 to 2016. The yellowtail analysis illustrates how data from multiple surveys can be used to infer density hot spots in an area that is not sampled one or more surveys. The copepod analysis assimilates seasonally unbalanced samples to estimate an annual index of the seasonal timing of copepod abundance and identifies a positive correlation between this index and cold-pool extent. We conclude by discussing additional potential uses of seasonal spatio-temporal models and emphasize their ability to identify climate-driven shifts in the seasonal timing of fish movement and ecosystem productivity.

Ocean management involves monitoring data that are used in biological models, where estimates inform policy choices. However, few science organizations publish results from a recurring, quantitative process to optimize effort spent measuring fish age. We propose that science organizations could predict the likely consequences of changing age-reading effort using four independent and species-specific analyses. Specifically we predict the impact of changing age collections on the variance of expanded age-composition data (“input sample size”, Analysis 1), likely changes in the variance of residuals relative to stock-assessment age-composition estimates (“effective sample size”, Analysis 2), subsequent changes in the variance of stock status estimates (Analysis 3), and likely impacts on management performance (Analysis 4). We propose a bootstrap estimator to conduct Analysis 1 and derive a novel analytic estimator for Analysis 2 when age-composition data are weighted using a Dirichlet-multinomial likelihood. We then provide two simulation studies to evaluate these proposed estimators and show that the bootstrap estimator for Analysis 1 underestimates the likely benefit of increased age reads while the analytic estimator for Analysis 2 is unbiased given a plausible mechanism for model misspecification. We conclude by proposing a formal process to evaluate changes in survey efforts for stock assessment.

Estimating fish condition, the relative weight of an individual fish given its body length, is a convenient way to relate the physiological health and energetic status of fishes to their productivity. Despite evidence of density-dependence effects on condition in some species, previous research has not jointly estimated synchronous changes in condition and density operating at fine spatial scales (a few km). Therefore, we developed a spatio-temporal modeling approach that simultaneously estimates correlated variation in density (measured as numbers per area) and condition. We applied our approach to 6 eastern Bering Sea (EBS) groundfish species (4 flatfishes and 2 gadoids) for the period 1992-2016, and estimated correlations in spatial variation (unmeasured variation that is stable over time) and spatio-temporal variation (unmeasured variation that changes between years). Spatial variation in density had a strong significant negative association with spatial variation in condition for 3 flatfishes and a positive association for one gadoid. Spatio-temporal variation in density had a significant association with spatio-temporal variation in condition for one flatfish (negative) and one gadoid (positive). Moreover, for the 6 study species, bottom temperature was identified as an important predictor of both density and condition. The increasing trend in bottom temperatures between 1992 and 2016 was accompanied by an overall increase in the abundance-weighted condition of 5 species. We conclude that forecasts of changes in weight-at-age within some EBS groundfish assessments will require an understanding of both density-dependence and bottom temperature effects on fish condition to better prepare for future climate and exploitation changes.

How to properly weight composition data is an important ongoing research topic for fisheries stock assessments, and multiple methods for weighting composition data have been developed. Although several studies indicated that properly accounting for time-varying selectivity can reduce estimation biases in population biomass and management-related quantities, no study to date has compared the performance of widely used data-weighting methods when allowing for time-varying selectivity. Here, we conducted four simulation experiments on this topic, aiming to provide guidance on weighting age-composition data given time-varying selectivity. The first simulation experiment showed that over-weighting should be avoided in general and even when estimating time-varying selectivity. The second simulation experiment compared three data-weighting methods (McAllister–Ianelli, Francis, and Dirichlet-multinomial), within which the Dirichlet-multinomial method outperformed the other two methods when selectivity is time-varying. The third and fourth simulation experiments further showed that given time-varying selectivity, the Dirichlet-multinomial method still performed well when age-composition data were over-dispersed and when the level of selectivity variation needed to be simultaneously estimated. Our simulation results support using the Dirichlet-multinomial method when estimating time-varying fishery selectivity. Also, the simulations suggest that improving stock assessments by accounting for time-varying selectivity requires simultaneously addressing data weighting and time-varying selectivity.

Spatiotemporal predictions of bycatch (i.e., catch of nontargeted species) have shown promise as dynamic ocean management tools for reducing bycatch. However, which spatiotemporal model framework to use for generating these predictions is unclear. We evaluated a relatively new method, Gaussian Markov random fields (GMRFs), with two other frameworks, generalized additive models (GAMs) and random forests. We fit geostatistical delta-models to fisheries observer bycatch data for six species with a broad range of movement patterns (e.g., highly migratory sea turtles versus sedentary rockfish) and bycatch rates (percentage of observations with nonzero catch, 0.3%–96.2%). Random forests had better interpolation performance than the GMRF and GAM models for all six species, but random forests performance was more sensitive when predicting data at the edge of the fishery (i.e., spatial extrapolation). Using random forests to identify and remove the 5% highest bycatch risk fishing events reduced the bycatch-to-target species catch ratio by 34% on average. All models considerably reduced the bycatch-to-target ratio, demonstrating the clear potential of species distribution models to support spatial fishery management.

Fisheries managers use biological reference points (BRPs) as targets or limits on fishing and biomass to maintain productive levels of fish stock biomass. There are multiple ways to calculate BRPs when biological parameters are time varying. Using summer flounder (Paralichthys dentatus) as a case study, we investigated time-varying approaches in concert with climate-linked population models to understand the impact of environmentally driven variability in natural mortality, recruitment, and size-at-age on two commonly used BRPs [B0(t) and F35%(t)]. We used the following two approaches to calculate time-varying BRPs: dynamic-BRP and moving-average-BRP. We quantified the variability and uncertainty of different climate dependencies and estimation approaches, attributed BRP variation to variation in life-history processes, and evaluated how using different approaches impacts estimates of stock status. Results indicate that the dynamic-BRP approach using the climate-linked natural mortality model produced the least variable reference points compared to others calculated. Summer flounder stock status depended on the estimation approach and climate model used. These results emphasize that understanding climate dependencies is important for summer flounder reference points and perhaps other species, and careful consideration is warranted when considering what time-varying approach to use, ideally based upon simulation studies within a proposed set of management procedures.

Fishing communities are increasingly required to adapt to environmentally driven changes in the availability of fish stocks. Here, we examined trends in the distribution and biomass of five commercial target species (dover sole, thornyheads, sablefish, lingcod, and petrale sole) on the US west coast to determine how their availability to fishing ports changed over 40 years. We show that the timing and magnitude of stock declines and recoveries are not experienced uniformly along the coast when they coincide with shifts in species distributions. For example, overall stock availability of sablefish was more stable in southern latitudes where a 40% regional decline in biomass was counterbalanced by a southward shift in distribution of >200 km since 2003. Greater vessel mobility and larger areal extent of fish habitat along the continental shelf buffered northerly ports from latitudinal changes in stock availability. Landings were not consistently related to stock availability, suggesting that social, economic, and regulatory factors likely constrain or facilitate the capacity for fishers to adapt to changes in fish availability. Coupled social–ecological analyses such as the one presented here are important for defining community vulnerability to current and future changes in the availability of important marine species.

The birth and survival of juveniles (termed recruitment) is highly variable in marine populations, and this variation is not precisely predicted by spawning output or environmental effects in isolation. Variation in recruitment after accounting for differences in spawning output (termed “recruitment deviations”) is commonly estimated in age-structured populations models that are fitted to data including samples of length or age-composition (“age-structured integrated models”, ASIM). However, recruitment deviations often are not estimated when fitting only to catch and abundance-index data (“age-structured production models”, ASPM), or catch data given an assumed value for final stock status (“age-structured stock-reduction analysis”, ASSRA). We use a simulation and real-world example to illustrate the potential benefits of estimating recruitment deviations in data-rich (ASIM), data-moderate (ASPM), and data-poor (ASSRA) models, while using either marginal likelihood or tuning algorithms with penalized likelihood estimation. The simulation experiment shows that marginal and penalized likelihood methods are both unbiased and precise for ASIM when estimating recruitment deviations, but biased and imprecise when not. The ASPM is somewhat biased both with or without estimating recruitment deviations, but is more precise when estimating recruitment deviations than not. The ASPM with recruitment deviations estimated using marginal likelihood has somewhat better performance than using penalized likelihood, although maximum likelihood also sometimes does not converge. The ASSRA has similar precision with or without estimating recruitment deviations, but estimating recruitment deviations is important to have reasonable estimates of confidence intervals. Both marginal and penalized likelihood can accurately estimate the variance of recruitment deviations for ASIM and ASPM. Estimating recruitment deviations is necessary to accurately estimate uncertainty in spawning output, and results are robust to mis-specifying the recruitment model (e.g., when simulating data with regime-shift dynamics). A sensitivity analysis shows mis-specifying maturity-at-age causes bias in all models, but bias is somewhat mitigated when estimating recruitment deviations. Finally, a case-study involving Pacific hake off the US West Coast emphasizes that it is necessary to estimate recruitment deviations to closely match results from the original assessment.

Stock assessments are important to sustainable ocean management, but developing assessments remains time-consuming despite increased computation power and access to shared software. Improved efficiency in developing stock assessments could allow an increased rate of new assessments, increased attention to biological mechanisms in existing assessments, or accelerated testing of existing methods. I argue that the efficiency of the stock-assessment enterprise is hindered by a reliance upon ad hoc “tuning algorithms” that are conducted independently of standard parameter estimation. I present three examples where tuning algorithms are widely used: (1) determining the variance of recruitment, (2) bias-correcting recruitment deviations, and (3) determining the effective sample size for compositional data, and summarize why each tuning algorithm was originally developed. I then review recent research showing that each task can be replaced with parameter estimation involving random effects. Finally, I explain how model development, peer-review, and model testing would each be improved if tuning algorithms were replaced by parameter estimation, and outline the steps required to transition existing stock assessments to modern parameter estimation involving mixed effects.

Recruitment is one of the most variable biological processes driving fisheries population dynamics and the characteristics and drivers of the variation differ among stocks. Understanding, describing (modeling), estimating, and predicting (forecasting) this variation is essential to appropriate stock assessment and fisheries management. Most modern stock assessments estimate annual variation in recruitment, particularly those that include age or size composition data. Changes over time are often partitioned into those related to the spawning biomass due to density dependence (the stock-recruitment relationship) and those related to other factors such as environmental conditions. Several methods have been used to model recruitment inside stock assessment models, and they differ by how recruitment is represented and how statistical inference is conducted. Virtual population analyses make few statistical assumptions about recruitment and instead calculate recruitment as the sum of the mortality-adjusted catches of a cohort while surplus production models generally imply a stock-recruitment relationship in their production function. Integrated statistical stock assessment models have modelled recruitment using a stock-recruitment relationship, autocorrelation, a function of covariates, regime shifts, or a combination of these. Statistical inference has typically been conducted using state-space (hierarchical) models, whether using Bayesian or maximum likelihood methods, or inferences are approximated using penalized likelihood approaches, which may not be statistically reliable. All these methods have specific issues that need to be addressed and there are tradeoffs in their use. Future temporal variation also needs to be considered when providing management advice. We review the theory and practice of modelling temporal variation in recruitment in fisheries stock assessment, provide advice on good practices, and recommend important research.

Multiple processes act at different stages and different intensities within the timeline between spawning and the age designated as “recruitment”. However, common practice is to model only a single step between spawning stock and recruits. Reasons for this practice include lack of data on the intermediate stages, lack of understanding of the mechanisms and functional form governing intermediate stages, and lack of computational resources to model a multistage process in the appropriate statistical framework. We develop a state-space framework and, using a simulation study, we explore the estimation of multistage stock-recruit functions. We evaluated four different functions (Ricker, Beverton-Holt, Shepherd, and Generalized), and examined the effects on estimation of several factors, including the form of density dependence, the magnitude of measurement error associated with each stage, type of prior on measurement error, and the magnitude of process error between stages. Three-parameter stock-recruit functions (Shepherd, Generalized) correctly identified the form of density dependence in each stage, although the Shepherd model exhibited problems with convergence. Model misspecification resulted in bias, especially in parameters specifying measurement and process error; an informative prior on measurement error improved precision and bias. The Deviance Information Criterion selected the Ricker model too often, even when the true model was Beverton-Holt. Sequential density-dependent stages, even multiple overcompensatory stages, lead to an overall function that appears fairly flat, suggesting that a function capable of producing asymptotic dynamics is a practical default. The common practice of bypassing stages between the first (spawning stock) and the last (recruits) worked reasonably well, except when fitting a Ricker model, most likely because the true function was nearly flat over most of the range of the first stage. An application to data on North Sea herring illustrates that a multi-stage stock-recruit model can generate a stock-recruit function that is intermediate between Ricker and Beverton-Holt models, and that does not match existing three-parameter forms.

Theoretical and applied research suggests that survival rates during early life stages will increase when spawning biomass is reduced in marine fishes (termed “recruitment compensation”). However, the magnitude of recruitment compensation is generally difficult to estimate for individual fish stocks, and its average value for marine fishes remains highly contested. Scientists and managers for Pacific rockfishes (Sebastes spp.) on the US West Coast have used a regional meta-analysis to estimate the likely distribution of the steepness parameter of the Beverton-Holt stock-recruit relationship using stock assessment models since 2007, and the method has been updated every two years as new assessments are conducted (i.e., five biennial updates). Here, we provide a short history of this approach, its methodological assumptions, changes in results over time, and ongoing efforts to validate its assumptions. While the regional meta-analysis has been successful in ensuring a consistent approach to treatment of steepness across assessments, the estimates of mean steepness have been unexpectedly variable as the meta-analysis has been updated. Specifically, we show that the estimated average value of steepness for West Coast rockfish increased markedly from 2007 (average: 0.75), before decreasing somewhat again in the 2017 update. We also show that this value has a strong impact on rockfish rebuilding plans, and showcase the example of canary and widow rockfishes, where the estimated rates of rebuilding are strongly influenced by the assumed value of steepness. We conclude by discussing the bias-variance tradeoff between using global and regional meta-analysis, as well as the likely implications of difficult-to-validate assumptions including: (1) no recruitment autocorrelation within each stock; (2) no correlations among stocks; and (3) no bias from individual stocks resulting from mis-specification of the stock assessment models used in the meta-analysis.