https://orcid.org/0000-0002-3626-5778
Figures
Abstract
Canadian fisheries management has embraced the precautionary approach and the incorporation of ecosystem information into decision-making processes. Accurate estimation of fish stock biomass is crucial for ensuring sustainable exploitation of marine resources. Spatio-temporal models can provide improved indices of biomass as they capture spatial and temporal correlations in data and can account for environmental factors influencing biomass distributions. In this study, we developed a spatio-temporal generalized additive model (st-GAM) to investigate the relationships between bottom temperature, depth, and the biomass of three key fished species on The Grand Banks: snow crab (Chionoecetes opilio), yellowtail flounder (Limanda ferruginea), and Atlantic cod (Gadus morhua). Our findings revealed changes in the centre of gravity of Atlantic cod that could be related to a northern shift of the species within the Grand Banks or to a faster recovery of the 2J3KL stock. Atlantic cod also displayed hyperaggregation behaviour with the species showing a continuous distribution over the Grand Banks when biomass is high. These findings suggest a joint stock assessment between the 2J3KL and 3NO stocks would be advisable. However, barriers may need to be addressed to achieve collaboration between the two distinct regulatory bodies (i.e., DFO and NAFO) in charge of managing the stocks. Snow crab and yellowtail flounder centres of gravity have remained relatively constant over time. We also estimated novel indices of biomass, informed by environmental factors. Our study represents a step towards ecosystem-based fisheries management for the highly dynamic Grand Banks.
Citation: Ruiz-Diaz R, Pennino MG, Fisher JAD, Eddy TD (2024) Decadal changes in biomass and distribution of key fisheries species on Newfoundland’s Grand Banks. PLoS ONE 19(4): e0300311. https://doi.org/10.1371/journal.pone.0300311
Editor: Andrea Belgrano, Swedish University of Agricultural Sciences and Swedish Institute for the Marine Environment, University of Gothenburg, SWEDEN
Received: November 2, 2023; Accepted: February 27, 2024; Published: April 1, 2024
Copyright: © 2024 Ruiz-Diaz et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data supporting our research is not open access due to regulatory and confidentiality constraints. The dataset in question, the RV trawl sampling dataset named "7B_KG-PER-SET_DATA_FOR_GIS_MAPPING_9_Feb_2021," is owned and managed by Fisheries and Oceans Canada (DFO) Newfoundland and Labrador region. To ensure transparency and facilitate access for replication, verification, or further exploration of our findings, we have established an official channel through which researchers and interested parties can request access to the raw data. They can make a formal request to the regional headquarters via the following link: https://www.dfo-mpo.gc.ca/contact/regions/newfoundland-terreneuve-eng.html. The request process involves filling out a form specifying the data of interest, including details in the query such as: Division: 3L, 3N and 3O; Season: spring; Gear name: Campelen; Time series: 1996-2019, to ensure accurate retrieval. We understand the importance of data sharing in scientific research and are committed to promoting transparency. However, due to legal and ethical considerations, we are unable to make the data publicly available. Our research involves sensitive information, and public deposition would breach compliance with the protocol approved by our research ethics board. We believe that by directing interested parties to the official channel at DFO, we can balance the need for transparency with the necessity of adhering to legal and ethical obligations. We have ensured that the code employed for data analysis and interpretation in this study is publicly available on Figshare [DOI: https://doi.org/10.6084/m9.figshare.24916461.v1].
Funding: Fisheries & Oceans Canada (DFO) Atlantic Fisheries Fund. Natural Sciences and Engineering Research Council of Canada (NSERC). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Stock assessments aim to evaluate the status of a population by evaluating biomass and fishing mortality relative to reference points to define catch limits [1]. Time series of biomass indices are often used to calibrate stock assessment models, primarily derived from fisheries-independent data collected from scientific surveys. These surveys commonly employ a stratified-random sampling design to generate estimates of absolute biomass by using area-swept information [2, 3]. This involves dividing the study area into different strata based on specific characteristics, such as depth or habitat type. Within each stratum, random samples are collected. Random stratified sampling increases the precision of the estimates when the population is homogenously distributed among strata. However, variability in habitat preference may exist within strata, compromising the robustness of this approach [4]. This approach also requires all strata to be sampled at each sampling event, which is not always possible due to inclement weather, broken ships, among others.
Spatial correlation is a common feature of fisheries data. It occurs when observations collected at different locations are not independent of each other, as nearby things tend to be more connected than distant things [5]. If unaccounted for, spatial correlation can lead to biased estimates of biomass and abundance [6]. This also applies to the correlation of objects through time [7]. Spatio-temporal generalized mixed effect models (GLMMs) and generalized additive models (GAMs) can account for spatial heterogeneity found in fisheries survey data [8]. The latter are more flexible because they are fit using smoothing spline terms, making them especially useful for addressing non-linear relationships [9]. These models explicitly account for both spatial and temporal correlations in a dataset [10], and can incorporate information about environmental variables that may be driving species biomass and distribution [11]. Another advantage of these methods is that they employ spatial interpolation throughout the region of interest, overcoming incomplete sampling issues [9]. Spatio-temporal models can improve predictions for areas and years with little or no data and can be more precise than with design-based methods (i.e., strata-based index) and conventional GLMMs and GAMs (i.e., without spatio-temporal effects) [12, 13].
Our study focuses on the practical application of spatio-temporal Generalized Additive Models (st-GAMs) [14] to understand the biomass dynamics and distributional changes of snow crab (Chionoecetes opilio), yellowtail flounder (Limanda ferruginea), and Atlantic cod (Gadus morhua) on The Grand Banks of Newfoundland. The Grand Banks is a highly productive region where two distinct water masses, the Labrador Current and the Gulf Stream, converge. The Labrador Current brings fresh and nutrient-rich waters, while the Gulf Stream carries warmer and saltier waters. This combination of conditions makes The Grand Banks a dynamic and highly variable ecosystem [15, 16]. The Grand Banks experienced a regime shift in the early 1990s characterized by the collapse of the Atlantic cod, yellowtail flounder and other groundfish species. However, species such as snow crab and northern shrimp proliferated during that time [17]. The shift was attributed to a combination of factors, including overfishing and changes in environmental conditions [18]. In recent years, the species in this study exhibited different dynamics, with snow crab experiencing a significant decline, Atlantic cod remaining persistently low and yellowtail flounder recovered to pre-collapse levels. Currently, the yellowtail flounder fishery is operating as a Marine Stewardship Council (MSC) certified fishery [19].
The Grand Banks constitute an independent ecosystem production unit within the Newfoundland and Labrador Shelf, characterized by high ecosystem productivity and a well-defined marine community [17, 20]. We used this ecosystem production unit as our spatial scale of focus and examined how it relates to the management units of the study species. Our study contributes to the evolving landscape of ecosystem-based fisheries management through the application of advanced modeling techniques to uncover spatio-temporal patterns in snow crab, yellowtail flounder and Atlantic cod dynamics. These species were selected based on their variable responses post cod collapse, their cultural and economic fisheries importance and their distinct movement behaviors. Gadoids are streamlined swimmers, flatfish are bottom-dwellers that perform undulatory movements and snow crab have legs that allow for lateral movement. These adaptations are shaped by the specific ecological niches and lifestyles of each species. By incorporating environmental variables into our analysis (bottom temperature and depth) and considering spatial heterogeneity, we explored species-habitat associations and calculated environmental-informed biomass indices that can be used in stock assessment models. Finally, we assessed the effect of fishing on the environmental-informed biomass indices.
Material and methods
Data sources
Multispecies survey.
Fisheries and Oceans Canada (DFO) have been conducting annual stratified-random multispecies trawl surveys on The Grand Banks, located in the Northwest Atlantic since 1971 [21] (Fig 1). The survey is conducted in spring and fall and has changed sampling gear and coverage over time. Fall surveys, particularly in 2004–06, have issues with an absence of deep sets, reduced coverage and timing extensions due to vessel breakdowns and unplanned changes [22]. In contrast, the spring survey (April-June) has had more consistent coverage [mean observations of 282.5 per year], although the 2006 and 2017 surveys had lower coverage [194 and 191 observations, respectively]. To limit variability introduced by changes in the surveys, we utilized data from surveys conducted in spring (6,780 observations in total). Since 1996, spring surveys have been conducted with the CCGS Alfred Needler vessel and Campelen 1800 shrimp trawl gear, sampling up to 732 meters depth. The trawl survey collects data on the abundance, size, and biomass of numerous groundfish and shellfish species, in addition to other biological information (i.e., size, maturation status, body condition, stomach contents) [21].
Solid lines indicate the management unit boundaries of the North Atlantic Fisheries Organization (NAFO) divisions (3L, 3N and 3O).
Biomass data were used to create two response variables for each species: presence/absence and conditional-to-presence-biomass. Abiotic explanatory variables were bathymetry (here called depth) and bottom temperature, also obtained from the DFO bottom trawl survey (Table 1). These covariates were selected based on their well-known relationship with distribution and productivity of the study species [23–25]. The relationship between the study species and salinity is less understood in this region and was not considered here. Finally, we assumed that expected fish biomass is proportional to the area surveyed (referred to as swept area); thus we included swept area (log-transformed) as offset in all the models to account for the effort.
Prediction grid.
We created a 5x5 km grid for The Grand Banks with 274,461 grid points. We used gridded bathymetry data with spatial resolution of 15 arc seconds (≈0.004 º) obtained from the General Bathymetric Chart of the Oceans (GEBCO) project (https://www.gebco.net) and manipulated it with the R package marmap [26] to extract depth values at each grid point. Rasters of bottom temperature data were provided by DFO. These rasters were created using data from different sources (DFO multispecies survey, Atlantic Zone Monitoring Program (AZMP) hydrographic campaigns, International Oceanographic Campaigns (IOC), ARGO program, etc) [27]. All data were vertically averaged in 5 m bins, and a linear interpolation was applied to fill missing bins. We selected the data corresponding to the spring season (April-June) for which data were averaged on a regular 0.1° x 0.1° grid. Horizontal linear interpolation was applied to overcome missing data on grid cells (full description in [27]). We tested the correlation between these databases (i.e., GEBCO and DFO temperature interpolation) with trawl station observations to assess accuracy (Figs 1 and 2 S1 Appendix).
Fishing data.
Finally, fishing data aggregated at division level were obtained from the NAFO STLATLANT database. Strata-based indices of snow crab [28], yellowtail flounder [29] and Atlantic cod [30] were collected and compared to the new environmental-informed indices.
Spatio-temporal modelling
An exploratory analysis highlighted that species biomass data have two main features, namely strong spatial and temporal dependence and a large proportion of observed zeros (i.e., zero inflated data). To address this, we developed a delta gamma generalized additive model (GAM) using the R-package sdmTMB [31]. This separately analyzes species occurrence (biomass information is transformed to 0 and 1 according to species absence and presence, respectively) and conditional-to-presence biomass (observations with positive biomass values) and combines both predictions in a final estimate of biomass [8]. Delta gamma models are commonly found in the literature to perform analysis similar to this one [32, 33]. We considered Z(s,t) to be the spatiotemporally distributed occurrence and Y(s,t) the conditional-to-presence biomass at location s and time t. The final model formulation is described as follows (Formula 1):
(1)
where π (s, t) represents the probability of occurrence at location s and time t; and μ(s,t) and ϕ are the mean and variance of the conditional-to-presence biomass, respectively. The linear predictors, which represent the intercept of each variable associated to the parameter π (s, t) and μ (s, t), are represented by βz and βY, respectively. Survey year was added as a fixed effect in our model (Yi). f() represents any function applied to the covariates (Xi), which in the present study were smoothing terms (p-splines). Vz(s, t) and Vy(s, t) refer to the spatio-temporal structure of the occurrence and conditional-to-biomass model, respectively.
sdmTMB relies on the integrated nested Laplace approximation (INLA) to discretized the space by defining a Delaunay triangulation mesh, which in turn creates an artificial set of neighbors over the study area, and Gaussian Markov random fields to model spatial dependencies between observations [31, 34]. Our approach involved constructing a Delaunay triangulation mesh with a defined minimum distance of 20 km, resulting in a mesh comprising 467 vertices (Fig 3 in S1 Appendix). This choice was carefully considered, accounting for the spatial distribution of sampling locations and aiming to strike a balance between computational efficiency and predictive accuracy. To test which model performs better, we evaluated the spatial effect by running a model with and without the spatial component. The spatio-temporal component was included as: 1) a first-order autoregressive effect (AR1), which has a parameter (rho) that regulates the degree of autocorrelation between random field deviations from one year to the next; 2) a random walk (RW) and 3; an independent and identically distributed (iid), in which the random fields are assumed to be independent across time steps, to test which one performs better.
Model selection and validation
We calculated Pearson’s rank correlation index and the variance inflation factor (VIF) of the covariates before model runs. This helped avoid correlation and collinearity among explanatory variables [35]. We did not find any substantial correlation among the covariates (R < 0.6 and VIF < 3), allowing us to proceed with including depth and bottom temperature in the st-GAM. To model the non-linear relationship between explanatory and response factors, covariates were included to the model as random factor with a smoothing term (p-splines) [31]. Depth was log-transformed for better model convergence.
To assess the importance of bottom temperature, depth, and the spatial component on species spatial distribution, Akaike weights were used instead of stepwise variable selection, as they account for model selection uncertainty [36, 37]. By summing these weights, the relative importance of each variable can be estimated. To test model predictability, we carried out a k-fold cross-validation in which data were randomly split in k = 4 folds of equal size. In each iteration, one of the folds is held to test the data, while the other 3 are used to train the model. Then, we used the expected log pointwise predictive density (ELPD) to evaluate model predictive accuracy [31, 38].
To validate models, residuals were evaluated to ensure that spatial patterns were not detected and that residuals were normally distributed (S2 Appendix).
Biomass index and centre of gravity calculation
Once models were fit, we predicted species biomass over the study area using a 5x5 km grid. We summed up the densities calculated in the predictions and multiplied them by cell area to calculate the biomass index. In the case of snow crab, species catchability in the bottom trawl survey was found to be lower than 1 [39]. To account for this issue, a conversion factor was calculated using a Delury depletion regression analysis on fishery catch rate data from logbooks from 2000 to 2016 [40] and biomass estimates were adjusted by a factor of 0.126. Agreements between the new biomass indices (spatially aggregated biomass data) and strata-based indices (strata aggregated biomass data) reported in the stock assessment of snow crab [28], yellowtail flounder [29] and Atlantic cod [30] were assessed by using the coefficient of determination, R2. For snow crab, the strata-based indices were estimated using fall survey data, whereas our environmental-informed indices were estimated using spring data. The models used in the assessment of all stocks except 3NO Atlantic cod require biomass indices as input (snow crab assessment uses biomass trends, yellowtail flounder a Schaefer surplus production model and 2J3KL Atlantic cod a state-space model). Therefore, the new biomass indices developed in the present study are comparable and could be used in the assessment of the species.
Finally, we estimated the centre of gravity of the populations using the following (Formula 2):
(2)
where xi is the location (x or y-coordinates) of the grid cell, wi is the species biomass estimated at each grid cell i, and n is the total number of grid points in the study area (n = 274,461) [41]. We also calculated the species centre of gravity directly from the data using mean weight.
Fishing impact
We were unable to account for the impact of fishing due to a lack of spatially resolved fishing effort data that could be integrated into the model. Despite this limitation, we endeavored to explore the fishing effect on species biomass using a linear regression model with species biomass at the divisional level (3L, 3N, and 3O) as the response variable and fishing catch data aggregated at the same level (NAFO STLATLANT database) as the explanatory variable. We only investigated yellowtail flounder and snow crab because the commercial Atlantic cod fishery on The Grand Banks has remained closed since 1992 due to the slow recovery after its collapse, with species being captured as bycatch and in recreational (known as the food fishery) and small vessel, inshore commercial fisheries [42].
Results
For all species, models that included spatial effects produced better fits. Similarly, models that included bottom temperature and depth effects as covariates performed better (Table 2). Our results showed that the spatial effect explained most of the variability in the biomass data, followed by depth and temperature, respectively (Table 2-ΔAIC values). We also tested different spatio-temporal configurations (AR1, RW and iid) and found that the autoregressive spatio-temporal structure (AR1) had higher predictive accuracy for snow crab and yellowtail flounder, while independent and identically distribution (iid) performed better for Atlantic cod (Table 2-ΔELPD values). AR1 has a rho parameter that indicates the degree of correlation from one year to the next. We found that yellowtail flounder biomass had a rho = 0.73 and snow crab biomass had a rho = 0.71.
Model structure indicates the different configurations tested in the model. Note that bottom temperature (temp) and depth were added as smoothing terms (p-spline). df indicates the degree of freedom, AIC is the Akaike information criteria, and ΔAIC indicates differences among AICs. ELPD is the expected log pointwise predictive density, and differences among ELPDs are indicated by ΔELPD. The best model configuration is in bold.
Spatial and covariate effects
The spatial random field represents consistent deviations in space through time that are not accounted for by depth and bottom temperature covariates. Higher spatial deviations were found in the north of The Grand Banks for snow crab occurrence, while they were higher near the nose and tail of the Banks for snow crab biomass (Fig 2a and 2b). Higher spatial deviations of yellowtail flounder probability of occurrence and biomass were both found in the southern part of The Grand Banks (Fig 2h and 2i). Higher spatial deviations of Atlantic cod occurrence were found in the western part of the Banks and in the 3O division, while spatial deviations were higher in the south (3NO division) and around The Grand Banks’ periphery for Atlantic cod biomass (Fig 2o and 2p).
Spatial random field deviations of snow crab probability of occurrence (a) and biomass (b), yellowtail flounder probability of occurrence (h) and biomass (i), and Atlantic cod probability of occurrence (o) and biomass (p). Temperature smoothed effects on snow crab probability of occurrence (c) and biomass (d), yellowtail flounder probability of occurrence (j) and biomass (k), and Atlantic cod probability of occurrence (q) and biomass (r). Depth smoothed effect on snow crab probability of occurrence (e) and biomass (f), yellowtail flounder probability of occurrence (l) and biomass (m), and Atlantic cod probability of occurrence (s) and biomass (t). Note that depth has been log transformed. Biomass density units are kg/25 km2. Delta gamma combined biomass prediction of snow crab (g), yellowtail flounder (n) and Atlantic cod (u) over the Grand Banks. Predictions made on a 5x5 km grid.
Higher probability of snow crab occurrence and biomass was associated with colder temperatures, below 0 ºC (Fig 2c and 2d), and at depths of about 100 m for occurrence and 450 m for biomass (Fig 2e and 2f). For yellowtail flounder, both higher probability of occurrence and higher biomass were predicted for temperatures close to 3 ºC and shallower depths around 80 m (Fig 2j–2m). Atlantic cod probability of occurrence was predicted to be highest at temperatures close to 3 ºC and depths of 300 m (Fig 2q and 2r). Highest biomass of Atlantic cod was predicted to occur at 5 ºC and at depths between 200 and 400 m (Fig 2s and 2t).
Combined predictions of the delta gamma models indicate that snow crab biomass is higher in the north of the Banks, with a hotspot located in the northeast, and in the southeastern edge of the Banks (Fig 2g). The yellowtail flounder biomass hotspot is found south of the Banks (Fig 2n). Finally, Atlantic cod biomass is higher in the north and periphery of the Banks, as well as in the southeast (Fig 2u).
Biomass indices
A decline in the snow crab relative biomass index was observed over time, reaching a minimum of 33.72 t in 2016 (Fig 3a). Although there has been a small recovery since then, the current biomass index of snow crab was 85% lower in 2019 than at the beginning of the time series in 1996. Yellowtail flounder relative biomass has fluctuated over time, peaking in 2006 (597.88 t), 2012 (538.31 t) and 2008 (424.83 t) (Fig 3b). However, the biomass plummeted to its lowest value in 2016 (119.96 t) and has remained relatively low since then, with a biomass of 171.12 t in 2019. Similarly, Atlantic cod biomass has fluctuated over time, reaching its highest value in 2013 (150.07 t), with smaller peaks in 1999 (100.6 t) and 2006 (100.04 t) (Fig 3c). Since 2014, the relative biomass of Atlantic cod has been on the decline, hitting a record low of 19.30 t in 2017, a similar biomass level when the stock collapsed in the mid 1990s (31.12 and 20.39 t in 1996 and 1997, respectively). As of 2019, the biomass of Atlantic cod on The Grand Banks was rather low compared to the historical series (43.17 t).
Shaded areas indicate the 95% confidence interval. Black dashed line indicates the strata-based index of the species (units in tonnes x1000). R2 is the coefficient of determination.
Our model effectively generated an environmental-informed biomass index that showed trends consistent with strata-based indices reported in the stock assessments of the studied species. For yellowtail flounder, our biomass index had a high correlation (R2 = 0.92), with the strata-based index falling within the confidence interval. For snow crab, the correlation with the strata-index was also high (R2 = 0.84). However, some disparities were evident between the indices, as our analysis indicates slightly higher biomass estimates, particularly at the beginning of the time series. Note that we are comparing our biomass index created using spring data with the strata-based biomass index created using fall data. Unfortunately, no established index exists for Atlantic cod in the 3LNO division since this species is considered two separate stocks (i.e., 2J3KL and 3NO). As a result, we compared our biomass index to the index used to assess the 3NO stock, as it more accurately represents cod biomass within The Grand Banks, resulting in a correlation of R2 = 0.63.
Centre of gravity
Centre of gravity indicates the central point of a population distribution. In the case of snow crab, the centre of gravity shifted slightly toward the northwest of The Grand Banks over time (Fig 4a). The centre of gravity of yellowtail flounder has remained relatively stable (Fig 4b). Atlantic cod had the greatest changes in centre of gravity, first moving eastward and then northward at the end of the time series (Fig 4c). We also calculated the species centre of gravity directly from the data using mean weight, which displayed similar trends but are slightly more spread out (Fig 4 in S1 Appendix). This difference can be attributed to the spatio-temporal model predicting biomass in years and locations with sampling gaps.
Points indicate mean values and bars indicate variance. Colors represent years (from 1996 to 2019), with more recent years in yellow.
Fishing effect
Our findings suggest a negative relationship between snow crab biomass and fishing (measured as catch) in divisions 3L and 3N, although these relationships were not statistically significant (p = 0.1 and 0.9, respectively). Similarly, we observed a negative relationship between yellowtail flounder biomass and fishing in divisions 3N and 3O. These relationships were also not statistically significant, although the fishing effect was more pronounced in division 3N (p = 0.08 and 0.48, respectively) (Fig 5 in S1 Appendix).
Discussion
Our findings highlight the importance of considering spatial heterogeneity in fisheries survey data, as the spatial component accounted for the majority of observed variance for all three species (see Table 2). The spatial component explains variance that is not accounted for by the covariates depth and temperature. We observed that spatial effects differed between the occurrence and biomass processes for snow crab and Atlantic cod, but not for yellowtail flounder. These differences have been found in other studies and may be indicating spatial differences between species life stages, which are ignored in the occurrence analysis but get weighted in the biomass analysis [43]. Likewise, we noted a disparity in the relationship between biomass and depth for snow crab in both the occurrence and biomass processes, which is likely linked to the preference of adult snow crab for deeper waters [44]. Biomass hotspots are more restrictive than occurrence hotspots since biomass is higher only in areas with suitable conditions (e.g., environmental, reduce competition, prey availability), whereas individuals have a wide spatial range where they can be found [43].
The new biomass indices presented in this study overcome issues related with gaps in sampling by interpolating among missing data points. Missing data were important in the years 2015 and 2017 due to incomplete sampling during those periods. [21, 22]. They also account for species habitat-presence (i.e., depth and temperature) when predicting in unsampled locations and address spatio-temporal correlation. Predicted model biomass estimates align closely with those obtained using strata-based methods, with some disagreements for snow crab and Atlantic cod. In the case of snow crab, the assessment of the stock is done using data from the fall trawl survey. This is because spring data are considered to be less reliable because certain population components may be sampled relatively poorly during this time, coinciding with the mating and moulting periods [45]. In our analysis, we applied the conversion factor used to account for low catchability of snow crab in the fall bottom trawl survey [39, 40]. The development of this conversion factor was created using data from 2000 to 2016. Thus, the disparities observed in the biomass indices, particularly during the early stages of the temporal series, could potentially be explained by the exclusion of certain years and differences between fall and spring data. Spring data correspond to the pre-fishery season; the higher biomass observed in the new biomass index compared to the fall index could also be attributed to this. For Atlantic cod, differences are mostly due to the comparison of 3NO to 3LNO Atlantic cod.
The models used in the assessment of all stocks included in our analysis, except 3NO Atlantic cod, require biomass indices as input (snow crab assessment uses biomass trends, yellowtail flounder a Schaefer surplus production model and 2J3KL Atlantic cod a state-space model). Therefore, the new biomass indices developed in the present study could be used in the assessment of the species. Use of spatio-temporal indices to fit stock assessments has been shown to improve estimate precision compared to design-based indices [12, 46]. Consequently, this approach has been adopted by governmental bodies such as DFO and the United States National Oceanic and Atmospheric Administration (NOAA) to conduct the assessment of species such as northern shrimp (Pandalus borealis) [47] and yelloweye rockfish (Sebastes ruberrimus) [48]. However, it is important to acknowledge that calculating these indices is computationally intensive and presents implementation challenges. For example, spatial confounding (i.e., unaccounted spatial effects influencing the relationships between predictors and response variables) may exist, leading to bias in predictions [49]. Thus, accurately defining the spatial component is crucial in spatio-temporal models [50].
Centre of gravity has been used to evaluate the impacts of climate, fishing pressure and other anthropogenic factors on the average location of marine populations [51, 52]. A shift in species centre of gravity may create challenges and risks for managing resources when species move outside of historical fishing areas or management boundaries [53, 54]. The temperature distribution over The Grand Banks is not uniform, with the south and the north warmer than the centre. Additionally, this area undergoes natural cyclical periods of cold and warmth [27]. Changes in species distribution (here reflected as changes in the centre of gravity) can be an early sign of water warming on The Grand Banks due to natural variability and/or to climate change [55, 56]. However, shifts in species distribution may also be influenced by other factors such as competition, prey availability, habitat degradation and fishing [51, 57]. Our results indicate that snow crab is the most sensitive species to warming as its biomass declined as temperature increases, in agreement with other observations in Newfoundland and Labrador, where cold events have been associated with higher recruitment of snow crab [58, 59]. Yellowtail flounder tolerates a wide range of temperatures, with a preference for 3 ºC. This agrees with the literature, stating that yellowtail flounder can survive wide fluctuations in temperature [60, 61]. The persistence in the yellowtail centre of gravity on The Grand Banks is likely related to a weak current regime allowing for the retention of eggs and larvae in the southern part of The Grand Banks [62]—hypothesized as a nursery ground [63]. Atlantic cod has preference for warmer waters compared to yellowtail flounder and snow crab, favoring temperatures around 5 ºC. The northward shift in the center of gravity of Atlantic cod could be explained by a faster recovery of the 3L component of the 2J3KL stock, whereas the 3NO component may been experiencing a slower recovery. However, it may also be related to a northern shift of the species as a response to warming in the region due to a northern shift of the Gulf Stream [55].
Management of natural resources is a complex task that should consider ecological processes and how they relate to administrative boundaries. When management units are solely defined based on these administrative boundaries, decisions can have unintended consequences for the ecosystem [64]. Our analysis of Atlantic cod showed a continuous distribution over The Grand Banks during years of higher biomass (Fig 6 in S1 Appendix), indicating hyperaggregation behavior (i.e., aggregation of fish in a location during a period of low abundance [65]). While this behavior can decrease individual competition and maximize fitness, it can also increase the vulnerability of the species to fishing because of range contraction [66]. Previous studies have documented the mixing of Atlantic cod stocks [67, 68], suggesting that a joint assessment of the population and management decisions at the 2J3KL and 3NO management units, similar to the approach taken with snow crab, would be prudent. However, the pursuit of such collaborative efforts may face institutional barriers, given that these two Atlantic cod stocks fall under the purview of distinct regulatory bodies, namely DFO and NAFO. NAFO is a regional fisheries management organization responsible of the management of high seas fishery resources while DFO manages resources within the Canadian economic exclusive zone. Both entities apply the precautionary approach and reference points to manage the resources [69, 70], however, differences in management strategies, regulatory frameworks, and governance structures across these governing entities may exist.
Ocean warming (natural or driven by climate change) has been identified as the primary driver of snow crab decline in Newfoundland, while fishing and competition may have had localized impacts [71]. Overfishing is generally blamed for the decline of yellowtail flounder stocks in the early 1990s, but the productivity of the species was also strongly influenced by climatic conditions during the collapse and subsequent recovery [63]. In the present analysis, even though we could not directly account for the fishing effect in our models, we assessed the effect of fishing on species biomass at a coarser spatial resolution—the NAFO division level. We found that, even though the negative effect of fishing (i.e., landed catches) on species biomass was important in certain divisions (such as 3L and 3N for snow crab, and 3N for yellowtail flounder), these relationships were not statistically significant (refer to Fig 5 in S1 Appendix). The divisions with strongest fishing effect overlaped with those in which species had a biomass hotspot. This is not surprising, since fishers harvest on aggregations of individuals and not homogeneously across the entire area. However, the non-significance (p > 0.05) of the fishing effect on the new biomass indices suggests that additional factors are likely contributing to the observed decline in species biomass.
Our analysis is based on spring survey data, and therefore, the distribution patterns we observed may differ during other seasons. Species distribution is highly influenced by seasonal cycles, particularly in temperate areas, due to variations in environmental factors, light availability, and nutrient supply [72, 73]. Atlantic cod of the 2J3KL stock seek refuge near the continental shelf edge in winter and move to shallow coastal waters and onto The Grand Banks plateau during spring and summer [74]. Similarly, snow crab undergo seasonal migrations related to moulting and mating [44]. It is likely that we are missing information on species affinity for habitats that rely on seasonal variations. In addition, we used aggregated size and sex information of species to predict biomass. Distribution differences among species life stages may exist, including potential shifts in habitat preferences between juvenile and adult stages [75].
As species ranges shift in different directions and rates, it is likely that predator-prey interactions will also change [76]. The availability and abundance of prey directly impacts the population dynamics and distribution of predators [77, 78]. When prey species become scarce, predators may experience declines in body condition and overall population size [79]. The diminishing abundance of capelin, the primary prey of Atlantic cod, has been identified as a contributing factor to the decline in Atlantic cod growth potential [79] and body condition [80] of the 2J3KL stock. Fisheries harvest of other prey items of the Atlantic cod (i.e. snow crab and northern shrimp) could further exacerbate the issue of food limitation, hampering stock recovery [79, 80]. On the other hand, predators like Atlantic cod can play a regulatory role in shaping prey populations. In the Barents Sea, Atlantic cod has been identified as a regulator of the snow crab population, impacting both its distribution and productivity [81, 82]. A comparable phenomenon could potentially exist between these species on The Grand Banks. The examination of the distribution maps (Fig 2g and 2u) reveals limited overlap between Atlantic cod and snow crab in The Grand Banks. The limited overlap could be primarily attributed to the preference of snow crab for lower temperatures (below 0 ºC), however snow crab distribution may also be regulated by Atlantic cod presence. Future studies could explore predator-prey dynamics and species competition using alternative methodologies such as joint species distribution models [83, 84]. Additionally, incorporating substrate type in future research could provide valuable insights into the habitat preferences of demersal and benthic species of The Grand Banks. Future research could enhance ecosystem understanding by incorporating predation-prey interactions [85, 86], integrate multiple sources of data [13] and forecast climate change impacts on The Grand Banks under different emissions scenarios [87]. Our work could serve as a foundation for the development of spatial management strategies, including the establishment of conservation areas and spatial closures [88, 89].
Conclusions
The introduction of novel indices offers a practical avenue for informed decision-making and underscores the importance of comprehensive approaches to fisheries management.
Our study revealed a continuous distribution of Atlantic cod across The Grand Banks and a possible northern shift of the species, emphasizing the need for joint management of the 2J3KL and 3NO stocks. The research presented here holds promise for enhancing the sustainability of Canadian fisheries by improving our understanding of the interactions between environmental variables and species distributions.
Acknowledgments
We would like to thank Sean Anderson and Eric Ward for their prompt response in the sdmTMB discussion forum, helping to improve the present analysis. We would also like to thank Frederic Cyr for providing bottom temperature data and those who assist with the dataset: Andrew Cuff, Andrew Roberts, Mariano Koen-Alonso and Darrel Mullowney. We would also thank the DFO technicians and scientists involved in the research surveys. We extend our sincere gratitude to the reviewers for their valuable insights and constructive feedback, which significantly contributed to enhancing the quality and clarity of this manuscript. Advanced computing resources are provided by the Digital Research Alliance of Canada, the organization responsible for digital research infrastructure in Canada, and ACENET, the regional partner in Atlantic Canada. ACENET is funded by Industry Science & Economic Development, the provinces of New Brunswick, Newfoundland & Labrador, Nova Scotia and Prince Edward Island, as well as the Atlantic Canada Opportunities Agency. The present study has been proofread using ChatGPT [version GPT-3.5] to enhanced clarity and readability for effective communication. After using this tool, R.R.D. reviewed and edited the content as needed and take full responsibility for the content of the publication.
References
- 1. Punt A. Those who fail to learn from history are condemned to repeat it: A perspective on current stock assessment good practices and the consequences of not following them. Fisheries Research 2023. 261.
- 2.
Martell S.J.D. Fisheries Management. In Encyclopedia of Ecology, edited by Sven Erik Jørgensen and Brian D. Fath, 1572–82. Oxford: Academic Press 2008. p. 1572–82. https://doi.org/10.1016/B978-008045405-4.00206-8
- 3. Smith SJ. Use of Statistical Models for the Estimation of Abundance from Groundfish Trawl Survey Data. Can J Fish Aquat Sci. 1990 May;47(5):894–903.
- 4.
Morin PJ. Community ecology. 2nd ed. Chichester, West Sussex; Hoboken, NJ: Wiley; 2011. 407 p.
- 5. Tobler WR. A Computer Movie Simulating Urban Growth in the Detroit Region. Economic Geography. 1970 Jun 1;46:234–40.
- 6. McGilliard CR, Punt AE, Methot RD, Hilborn R. Accounting for marine reserves using spatial stock assessments. Can J Fish Aquat Sci. 2015 Feb;72(2):262–80.
- 7.
Cressie N, Wikle CK. Statistics for Spatio-Temporal Data. John Wiley & Sons; 2015. 612 p.
- 8. Thorson JT, Shelton AO, Ward EJ, Skaug HJ. Geostatistical delta-generalized linear mixed models improve precision for estimated abundance indices for West Coast groundfishes. ICES Journal of Marine Science. 2015 Jun 1;72(5):1297–310.
- 9. Martínez-Minaya J, Cameletti M, Conesa D, Pennino MG. Species distribution modeling: a statistical review with focus in spatio-temporal issues. Stoch Environ Res Risk Assess. 2018 Nov;32(11):3227–44.
- 10. Redding DW, Lucas TCD, Blackburn TM, Jones KE. Evaluating Bayesian spatial methods for modelling species distributions with clumped and restricted occurrence data. Bosso L, editor. PLoS ONE. 2017 Nov 30;12(11):e0187602. pmid:29190296
- 11. Lloret-Lloret E, Pennino MG, Vilas D, Bellido JM, Navarro J, Coll M. Main drivers of spatial change in the biomass of commercial species between summer and winter in the NW Mediterranean Sea. Marine Environmental Research. 2021 Feb;164:105227. pmid:33478732
- 12. Shelton AO, Thorson JT, Ward EJ, Feist BE. Spatial semiparametric models improve estimates of species abundance and distribution. Can J Fish Aquat Sci. 2014 Nov;71(11):1655–66.
- 13. Grüss A, Thorson JT. Developing spatio-temporal models using multiple data types for evaluating population trends and habitat usage. ICES Journal of Marine Science. 2019 Dec 1;76(6):1748–61.
- 14. Thorson JT, Pinsky ML, Ward EJ. Model-based inference for estimating shifts in species distribution, area occupied and centre of gravity. Methods in Ecology and Evolution. 2016;7(8):990–1002.
- 15. Han G, Lu Z, Wang Z, Helbig J, Chen N, de Young B. Seasonal variability of the Labrador Current and shelf circulation off Newfoundland. Journal of Geophysical Research: Oceans [Internet]. 2008 [cited 2022 Nov 23];113(C10).
- 16. Han G, Ma Z, deYoung B, Foreman M, Chen N. Simulation of three-dimensional circulation and hydrography over the Grand Banks of Newfoundland. Ocean Modelling. 2011 Jan 1;40(3):199–210.
- 17.
Koen-Alonso, M., and Cuff, A. Status and trends of the fish community in the Newfoundland Shelf (NAFO Div. 2J3K), Grand Bank (NAFO Div. 3LNO) and Southern Newfoundland Shelf (NAFO Div. 3Ps) Ecosystem Production Units. NAFO SCR 2018 Doc. 18/070.
- 18. Dempsey DP, Koen-Alonso M, Gentleman WC, Pepin P. Compilation and discussion of driver, pressure, and state indicators for the Grand Bank ecosystem, Northwest Atlantic. Ecological Indicators. 2017 Apr 1;75:331–9.
- 19.
Knapman, P., Cook, R., and Blyth-Skyrme, R. OCI Grand Bank Yellowtail Flounder Trawl: Public Comment Draft Report June 2020.
- 20. Pepin P., Higdon J., Koen-Alonso M., Fogarty M. and Ollerhead N. Application of ecoregion analysis to the identification of Ecosystem Production Units (EPUs) in the NAFO Convention Area. NAFO SCR 2014 Doc. 14/069.
- 21.
Rideout, R M, and Ings, D.W. Temporal And Spatial Coverage Of Canadian (Newfoundland And Labrador Region) Spring And Autumn Multi-Species RV Bottom Trawl Surveys, With An Emphasis On Surveys Conducted In 2018. NAFO SCR 2019 Doc. 19/15
- 22.
Brodie, W. and Stansbury, D. A brief description of Canadian Multispecies Survey in SA2+ Divisions 3KLMNO from 1995–2006. NAFO SCR 2007 Doc. 07/18
- 23. Windle MJ, Rose GA, Devillers R, Fortin MJ. Spatio-temporal variations in invertebrate− cod− environment relationships on the Newfoundland–Labrador Shelf, 1995− 2009. Marine Ecology Progress Series. 2012 Nov 26;469:263–78.
- 24.
Colbourne, E.B. and Walsh, S.J. The Distribution and Abundance of Yellowtail Flounder (Limanda ferruginea) in Relation to Bottom Temperatures in NAFO Divisions 3LNO Based on Multi-Species Surveys from 1990–2005. NAFO SCR 2006 Doc. 06/23
- 25. Drinkwater KF. The response of Atlantic cod (Gadus morhua) to future climate change. Ices journal of marine science. 2005 Jan 1;62(7):1327–37.
- 26. Pante E. and Simon-Bouhet B. Marmap: A Package for Importing, Plotting and Analyzing Bathymetric and Topographic Data in R. PLoS One 2013 8 (9): e73051. pmid:24019892
- 27.
Cyr, F., Snook, S., Bishop, C., Galbraith, P.S., Chen, N., and Han, G. Physical Oceanographic Conditions on the Newfoundland and Labrador Shelf during 2021. DFO Can. Sci. Advis. Sec. Res. 2022 Doc. 2022/040. iv + 48 p.
- 28.
Government of Canada Fisheries and Oceans a. "Assessment of Newfoundland and Labrador (Divisions 2HJ3KLNOP4R) Snow Crab". DFO 2022 Can. Sci. Advis. Sec. Sci. Advis. Rep. 2022/012.
- 29. Parsons D.M., Rideout R. and Rogers R. 2021 Assessment of Yellowtail Flounder in NAFO Divisions 3LNO using a Stock Production Model in a Bayesian framework. NAFO SCR 2021 Doc. 21/018
- 30. Rideout R M, Rogers R. and Ings D.W. “An Updated Assessment of the Cod Stock in NAFO Divisions 3NO”. NAFO SCR 2021 Doc. 21/031
- 31. Anderson SC, Ward EJ, English PA, Barnett LAK. sdmTMB: an R package for fast, flexible, and user-friendly generalized linear mixed effects models with spatial and spatiotemporal random fields. bioRxiv; 2022. p. 2022.03.24.485545.
- 32. Paradinas I, Conesa D, López-Quílez A, Esteban A, Martín López L, Bellido J, et al. Assessing the spatiotemporal persistence of fish distributions: a case study on two red mullet species (Mullus surmuletus and M. barbatus) in the western Mediterranean. Mar Ecol Prog Ser. 2020 Jun 25;644:173–85.
- 33. Pennino MG, Izquierdo F, Paradinas I, Cousido M, Velasco F, Cerviño S. Identifying persistent biomass areas: The case study of the common sole in the northern Iberian waters. Fisheries Research. 2022 Apr;248:106196.
- 34. Lindgren F. and Rue H. Bayesian spatial modelling with R-INLA. Journal of statistical software, 2015 63(19).
- 35. Zuur AF, Ieno EN, Elphick CS. A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution. 2010 Mar 1;1(1):3–14.
- 36. Akaike H. Likelihood of a model and information criteria. Journal of Econometrics. 1981 May 1;16(1):3–14.
- 37. Burnham KP, Anderson DR. Multimodel Inference: Understanding AIC and BIC in Model Selection. Sociological Methods & Research. 2004 Nov 1;33(2):261–304.
- 38. Vehtari A, Gelman A, Gabry J. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Stat Comput. 2017 Sep;27(5):1413–32.
- 39. Dawe E.G., Walsh S.J. and Hynick E.M. Capture efficiency of a multi-species survey trawl for Snow Crab (Chionoecetes opilio) in the Newfoundland region. Fisheries Research 2010 101, no. 1–2: 70–79.
- 40.
Mullowney, D., Coffey, W., Baker, K., Evans, G., Fiander, D., Colbourne, E., et al. An Assessment of Newfoundland and Labrador Snow Crab (Chionoecetes opilio) in 2016. DFO 2017 Can. Sci. Advis. Sec. Res. Doc. 2017/081.
- 41. Hollowed A.B. "Spatial and temporal distributions of Pacific hake, Merluccius productus, larvae and estimates of survival during early life stages." Calif. Coop. Oceanic Fish. Invest. Rep 33 1992: 100–123.
- 42.
Government of Canada Fisheries and Oceans. “Groundfish Newfoundland and Labrador Region NAFO Subarea 2 + Divisions 3KLMNO.” June 23, 2020. https://www.dfo-mpo.gc.ca/fisheries-peches/ifmp-gmp/groundfish-poisson-fond/2020/groundfish-poisson-fond-2_3klmno-eng.htm.
- 43. Izquierdo F, Paradinas I, Cerviño S, Conesa D, Alonso-Fernández A, Velasco F, et al. Spatio-Temporal Assessment of the European Hake (Merluccius merluccius) Recruits in the Northern Iberian Peninsula. Front Mar Sci. 2021 Jan 28;8:614675.
- 44. Mullowney D, Morris C, Dawe E, Zagorsky I, Goryanina S. Dynamics of snow crab (Chionoecetes opilio) movement and migration along the Newfoundland and Labrador and Eastern Barents Sea continental shelves. Rev Fish Biol Fisheries. 2018 Jun;28(2):435–59.
- 45.
Mullowney, D., Dawe, E., Coffey, W., Skanes, K., Hynick, E., Fiander, D., et al. An Assessment of Newfoundland and Labrador Snow Crab (Chionoecetes opilio) in 2013. DFO 2014 Can. Sci. Advis. Sec. Res. Doc. 2014/114. vi + 206 p
- 46. Cao J., Thorson J.T., Richards R.A. and Chen Y. Spatiotemporal index standardization improves the stock assessment of northern shrimp in the Gulf of Maine. Canadian Journal of Fisheries and Aquatic Sciences, 2017 74(11), pp.1781–1793.
- 47.
Government of Canada Fisheries and Oceans. Assessment of Northern Shrimp (Pandalus borealis) and Striped Shrimp (Pandalus montagui) in the Eastern and Western Assessment Zones, February 2023. DFO 2023 Can. Sci. Advis. Sec. Sci. Advis. Rep. 2023/013.
- 48.
Haggarty, D.R., Huynh, Q.C., Forrest, R.E., Anderson, S.C., Bresch, M.J. and Keppel, E.A. Evaluation of potential rebuilding strategies for Inside Yelloweye Rockfish (Sebastes ruberrimus) in British Columbia. DFO 2021 Can. Sci. Advis. Sec. Doc. 2021/008
- 49. Clayton D.G., Bernardinelli L. and Montomoli C. Spatial Correlation in Ecological Analysis. International Journal of Epidemiology. 1993 Dec 1;22(6):1193–202. pmid:8144305
- 50. Commander CJC, Barnett LAK, Ward EJ, Anderson SC, Essington TE. The shadow model: how and why small choices in spatially explicit species distribution models affect predictions. PeerJ. 2022 Feb 14;10:e12783. pmid:35186453
- 51. Adams CF, Alade LA, Legault CM, O’Brien L, Palmer MC, Sosebee KA, et al. Relative importance of population size, fishing pressure and temperature on the spatial distribution of nine Northwest Atlantic groundfish stocks. PLOS ONE. 2018 Apr 26;13(4):e0196583. pmid:29698454
- 52. Friedland KD, Smoliński S, Tanaka KR. Contrasting patterns in the occurrence and biomass centers of gravity among fish and macroinvertebrates in a continental shelf ecosystem. Ecology and Evolution. 2021;11(5):2050–63. pmid:33717441
- 53. Ojea E, Lester SE, Salgueiro-Otero D. Adaptation of Fishing Communities to Climate-Driven Shifts in Target Species. One Earth. 2020 Jun 19;2(6):544–56.
- 54. Sumaila UR, Palacios-Abrantes J, Cheung W. Climate change, shifting threat points, and the management of transboundary fish stocks. Ecology and Society. 2020 Dec 22; 25(4).
- 55. Gonçalves Neto A, Langan JA, Palter JB. Changes in the Gulf Stream preceded rapid warming of the Northwest Atlantic Shelf. Commun Earth Environ. 2021 Apr 20;2(1):1–10.
- 56. Saba VS, Griffies SM, Anderson WG, Winton M, Alexander MA, Delworth TL, et al. Enhanced warming of the Northwest Atlantic Ocean under climate change. Journal of Geophysical Research: Oceans. 2016;121(1):118–32.
- 57. Robinson NM, Nelson WA, Costello MJ, Sutherland JE, Lundquist CJ. A Systematic Review of Marine-Based Species Distribution Models (SDMs) with Recommendations for Best Practice. Front Mar Sci. 2017 Dec 18;4:421.
- 58. Dawe EG, Parsons DG, Colbourne EB, John’s S. Relationships of sea ice extent and bottom water temperature with abundance of snow crab (Chionoecetes opilio) on the Newfoundland—Labrador Shelf. ICES CM 2008 (2008): B02
- 59. Marcello LA, Mueter FJ, Dawe EG, Moriyasu M. Effects of temperature and gadid predation on snow crab recruitment: comparisons between the Bering Sea and Atlantic Canada. Marine Ecology Progress Series. 2012 Nov 26;469:249–61.
- 60. Perry RI, Smith SJ. Identifying Habitat Associations of Marine Fishes Using Survey Data: An Application to the Northwest Atlantic. Can J Fish Aquat Sci. 1994 Mar;51(3):589–602.
- 61. Walsh SJ. Factors influencing distribution of juvenile yellowtail flounder (Limanda ferruginea) on the grand bank of Newfoundland. Netherlands Journal of Sea Research. 1992 Jun 1;29(1):193–203.
- 62. Simpson MR, Walsh SJ. Changes in the spatial structure of Grand Bank yellowtail flounder: testing MacCall’s basin hypothesis. Journal of Sea Research. 2004 May 1;51(3):199–210.
- 63. Brodie WB, Walsh SJ, Maddock Parsons D. An evaluation of the collapse and recovery of the yellowtail flounder (Limanda ferruginea) stock on the Grand Bank. ICES Journal of Marine Science. 2010 Dec 1;67(9):1887–95.
- 64. Kerr LA, Cadrin SX, Kovach AI. Consequences of a mismatch between biological and management units on our perception of Atlantic cod off New England. ICES Journal of Marine Science. 2014 Sep 1;71(6):1366–81.
- 65. Rose GA, Kulka DW. Hyperaggregation of fish and fisheries: how catch-per-unit-effort increased as the northern cod (Gadus morhua) declined. Can J Fish Aquat Sci. 1999 Nov 30;56(S1):118–27.
- 66. Burgess MG, Costello C, Fredston-Hermann A, Pinsky ML, Gaines SD, Tilman D, et al. Range contraction enables harvesting to extinction. Proceedings of the National Academy of Sciences. 2017 Apr 11;114(15):3945–50. pmid:28351981
- 67. Smedbol RK, Shelton PA, Swain DP, Fréchet A, Chouinard GA. Review of population structure, distribution and abundance of cod (Gadus morhua) in Atlantic Canada in a species-at-risk context. DFO 2002 Can. Sci. Adv. Sec. Doc. 2002/082.
- 68. Tulk FJ, Cadigan NG, Brattey J, Robert D. Spatial synchronicity in recruitment of Atlantic cod (Gadus morhua) stocks off Newfoundland and Labrador and the Flemish Cap. Fisheries Research. 2017 Jul 1;191:49–59.
- 69.
Government of Canada Fisheries and Oceans. A Harvest Strategy Compliant with the Precautionary Approach. DFO 2006 Can. Sci. Advis. Sec. Sci. Advis. Rep. 2006/023.
- 70.
Northwest Atlantic Fisheries Organization. NAFO Precautionary approach framework. Serial No. N5062. 2004 NAFO/FC Doc. 04/18.
- 71. Mullowney D., Dawe E.G., Colbourne E.B. and Rose G.A. A review of factors contributing to the decline of Newfoundland and Labrador snow crab (Chionoecetes opilio). Reviews in Fish Biology and Fisheries 2014 24: 639–657.
- 72.
Kaiser MJ, Williams PJ leB, Jennings S, Thomas DN, Barnes DKA. Marine Ecology: Processes, Systems, and Impacts. OUP Oxford; 2011. 520 p.
- 73. Lloret-Lloret E, Lloret-Lloret Elena, Navarro J, Giménez J, López N, Albo-Puigserver M, et al. The Seasonal Distribution of a Highly Commercial Fish Is Related to Ontogenetic Changes in Its Feeding Strategy. Frontiers in Marine Science. 2020;7.
- 74.
Government of Canada Fisheries and Oceans b. Stock Assessment of Northern Cod (NAFO Divisions 2J3KL) in 2021. DFO 2022 Can. Sci. Advis. Sec. Sci. Advis. Rep. 2022/041.
- 75. Tanaka KR, Chang JH, Xue Y, Li Z, Jacobson L, Chen Y. Mesoscale climatic impacts on the distribution of Homarus americanus in the US inshore Gulf of Maine. Can J Fish Aquat Sci. 2019 Apr;76(4):608–25.
- 76. Selden RL, Batt RD, Saba VS, Pinsky ML. Diversity in thermal affinity among key piscivores buffers impacts of ocean warming on predator–prey interactions. Global Change Biology. 2018;24(1):117–31. pmid:28731569
- 77. Sydeman WJ, Poloczanska E, Reed TE, Thompson SA. Climate change and marine vertebrates. Science. 2015 Nov 13;350(6262):772–7. pmid:26564847
- 78. Richardson DE, Palmer MC, Smith BE. The influence of forage fish abundance on the aggregation of Gulf of Maine Atlantic cod (Gadus morhua) and their catchability in the fishery. Can J Fish Aquat Sci. 2014 Sep;71(9):1349–62.
- 79. Petrie B, Frank KT, Leggett WC. Temporal evolution of critical traits and their relationship to cod stock collapse and recovery. Canadian Journal of Fisheries and Aquatic Sciences. 2022 Jun 8;79(11):1765–79.
- 80. Regular PM, Buren AD, Dwyer KS, Cadigan NG, Gregory RS, Koen-Alonso M, et al. Indexing starvation mortality to assess its role in the population regulation of Northern cod. Fisheries Research. 2022 Mar 1;247:106180.
- 81. Durant JM, Holt RE, Ono K, Langangen Ø. Predatory walls may impair climate warming associated population expansion. Ecology 2023. pmid:37342068
- 82. Holt RE, Hvingel C, Agnalt AL, Dolgov AV, Hjelset AM, Bogstad B. Snow crab (Chionoecetes opilio), a new food item for North-east Arctic cod (Gadus morhua) in the Barents Sea. ICES Journal of Marine Science. 2021 Mar;78(2):491–501.
- 83. Clark JS, Gelfand AE, Woodall CW, Zhu K. More than the sum of the parts: forest climate response from joint species distribution models. Ecological Applications. 2014;24(5):990–9. pmid:25154092
- 84. Pollock LJ, Tingley R, Morris WK, Golding N, O’Hara RB, Parris KM, et al. Understanding co-occurrence by modelling species simultaneously with a Joint Species Distribution Model (JSDM). Methods in Ecology and Evolution. 2014;5(5):397–406.
- 85. Rutterford LA, Simpson SD, Bogstad B, Devine JA, Genner MJ. Sea temperature is the primary driver of recent and predicted fish community structure across Northeast Atlantic shelf seas. Global Change Biology 2023. pmid:36896634
- 86. Laman EA, Rooper CN, Turner K, Rooney S, Cooper DW, Zimmermann M. Using species distribution models to describe essential fish habitat in Alaska. Can J Fish Aquat Sci. 2018 Aug;75(8):1230–55.
- 87. Thompson PL, Rooper CN, Nephin J, Park AE, Christian JR, Davies SC, et al. Response of Pacific halibut (Hippoglossus stenolepis) to future climate scenarios in the Northeast Pacific Ocean. Fisheries Research. 2023 Feb 1;258:106540.
- 88.
Lawler JJ, Wiersma YF, Huettmann F. Using Species Distribution Models for Conservation Planning and Ecological Forecasting. In: Drew CA, Wiersma YF, Huettmann F, editors. Predictive Species and Habitat Modeling in Landscape Ecology: Concepts and Applications. New York, NY: Springer; 2011. p. 271–90. https://doi.org/10.1007/978-1-4419-7390-0_14
- 89. Abad E, Abad E, Abad E, Pennino MG, Valeiras J, Vilela R, et al. Integrating spatial management measures into fisheries: The Lepidorhombus spp. case study. Marine Policy. 2020 Jun 1;116:103739.