Spatial Analysis of Bering Sea Groundfish Survey Data Using Generalized Additive Models

Generalized additive models (GAM) are herein applied to trawl survey data in the eastern Bering Sea with an eye to (1) detecting trends in groundfish distributions and (2) improving abundance estimates by including the trend. GAM is a statistical method, analogous to regression, but without the assumptions of normality or linearity that relate a response variable (in this case, fish abundance) to location (latitude and longitude) and associated environmental variables (e.g. depth and bottom temperature). GAM provided reasonable (i.e. high r2) fits to the spatial distribution of five flatfish species and was able to define a spatial "signature" for each species, namely their preferred depth and temperature range. GAM also gave lower average abundance and abundance variability estimates for these five flatfish species than the stratified sampling procedure previously employed.

Download Full-text

Spatio-temporal distribution of euphausiids: an important component to understanding ecosystem processes in the Gulf of Alaska and eastern Bering Sea

ICES Journal of Marine Science ◽

10.1093/icesjms/fsv272 ◽

2016 ◽

Vol 73 (8) ◽

pp. 2020-2036 ◽

Cited By ~ 7

Author(s):

Kirsten A. Simonsen ◽

Patrick H. Ressler ◽

Christopher N. Rooper ◽

Stephani G. Zador

Keyword(s):

Primary Production ◽

Bering Sea ◽

Gulf Of Alaska ◽

Additive Models ◽

Trophic Levels ◽

Production Model ◽

Trawl Survey ◽

Large Spatial Scale ◽

Eastern Bering Sea ◽

Abundance And Distribution

Abstract Euphausiids (principally Thysanoessa spp.) are found in high abundance in both the eastern Bering Sea (EBS) and the Gulf of Alaska (GOA). They are an important part of these cold-water coastal and pelagic ecosystems as a key prey item for many species, including marine mammals, seabirds, and fish, forming an ecological link between primary production and higher trophic levels. Acoustic-trawl (AT) survey methods provide a means of monitoring euphausiid abundance and distribution over a large spatial scale. Four years of AT and bottom-trawl survey data (2003, 2005, 2011, and 2013) were available from consistently sampled areas around Kodiak Island, including Shelikof Strait, Barnabas Trough, and Chiniak Trough. We identified euphausiid backscatter using relative frequency response and targeted trawling, and created an annual index of abundance for euphausiids. This index has broad application, including use in the stock assessments for GOA walleye pollock (Gadus chalcogrammus) and other species, as an ecosystem indicator, and to inform ecological research. We then used generalized additive models (GAMs) to examine the relationship between relative euphausiid abundance and potential predictors, including pollock abundance, temperature, bottom depth, and primary production. Model results were compared with an updated GAM of euphausiid abundance from the EBS to determine if the factors driving abundance and distribution were consistent between both systems. Temperature was not a strong predictor of euphausiid abundance in the GOA as in the EBS; warmer temperatures and lack of seasonal ice cover in the GOA may be a key difference between these ecosystems. Pollock abundance was significant in both the GOA and the EBS models, but was not a strongly negative predictor of euphausiid abundance in either system, a result not consistent with top-down control of euphausiid abundance.

Download Full-text

Statistical relationships between the distributions of groundfish and crabs in the eastern Bering Sea and processed returns from a single-beam echosounder

ICES Journal of Marine Science ◽

10.1093/icesjms/fsp147 ◽

2009 ◽

Vol 66 (6) ◽

pp. 1425-1432 ◽

Cited By ~ 8

Author(s):

Robert A. McConnaughey ◽

Stephen E. Syrjala

Keyword(s):

Bering Sea ◽

Environmental Control ◽

Generalized Additive Models ◽

Sampling Strategy ◽

Additive Models ◽

Surficial Sediments ◽

Single Beam ◽

Eastern Bering Sea ◽

Efficient Sampling ◽

Statistical Relationships

Abstract McConnaughey, R. A., and Syrjala, S. E. 2009. Statistical relationships between the distributions of groundfish and crabs in the eastern Bering Sea and processed returns from a single-beam echosounder. – ICES Journal of Marine Science, 66: 1425–1432. Groundfish and benthic invertebrates are not randomly distributed over the continental shelf of the eastern Bering Sea (EBS). Annual trawl surveys reveal distributional patterns that vary according to species, and substantial interannual variation in these patterns suggests some degree of environmental control. Quantitative habitat models are developed to explain the distribution and abundance of species in the EBS. Simple models based on readily available data (temperature and depth) are somewhat informative, but offer limited practical value. Earlier research in the EBS indicated that surficial sediments affect the distribution and abundance of groundfish. However, traditional sampling with grabs and cores is impractical over large areas, and an efficient sampling strategy is needed. Echosounders allow surveys of large areas, but it is unknown if they measure the relevant properties of sediments. Seabed echoes from a calibrated, single-beam echosounder were recorded over 17 000 km of trackline covering the EBS shelf. Generalized additive models were used to fit acoustic and other variables to abundance data for ten species. The final models explained 28–77% of the variability in abundances, including a marginal contribution of 2–13% by the acoustic predictors.

Download Full-text

Predicted shifts of groundfish distribution in the Eastern Bering Sea under climate change, with implications for fish populations and fisheries management

ICES Journal of Marine Science ◽

10.1093/icesjms/fsaa215 ◽

2020 ◽

Author(s):

Christopher N Rooper ◽

Ivonne Ortiz ◽

Albert J Hermann ◽

Ned Laman ◽

Wei Cheng ◽

...

Keyword(s):

Bering Sea ◽

Marine Species ◽

Species Distribution Modelling ◽

Fish Populations ◽

Trawl Survey ◽

Climate Scenarios ◽

Important Species ◽

Eastern Bering Sea ◽

Northern Latitudes ◽

Distribution Shifts

Abstract Climate-related distribution shifts for marine species are, in general, amplified in northern latitudes. The objective of this study was to predict future distributions of commercially important species in the eastern Bering Sea under six climate scenarios, by incorporating predictions of future oceanographic conditions. We used species distribution modelling to determine potential distribution changes in four time periods (2013–2017, 2030–2039, 2060–2069, and 2090-2099) relative to 1982–2012 for 16 marine fish and invertebrates. Most species were predicted to have significant shifts in the centre of gravity of the predicted abundance, the area occupied, and the proportion of the predicted abundance found in the standard bottom trawl survey area. On average the shifts were modest, averaging 35.2 km (ranging from 1 to 202 km). There were significant differences in the predicted trend for distribution metrics among climate scenarios, with the most extensive changes in distribution resulting from Representative Concentration Pathway 8.5 climate scenarios. The variability in distributional shifts among years and climate scenarios was high, although the magnitudes were low. This study provides a basis for understanding where fish populations might expand or contract in future years. This will provide managers’ information that can help guide appropriate actions under warming conditions.

Download Full-text

A Bayesian approach to choosing a design for surveying fishery resources: application to the eastern Bering Sea trawl survey

Canadian Journal of Fisheries and Aquatic Sciences ◽

10.1139/f96-286 ◽

1997 ◽

Vol 54 (2) ◽

pp. 301-311 ◽

Cited By ~ 8

Author(s):

M K McAllister ◽

E K Pikitch

Keyword(s):

Bayesian Approach ◽

Bering Sea ◽

Trawl Survey ◽

Fishery Resources ◽

Eastern Bering Sea

Download Full-text

Patterns in groundfish abundance along the Eastern Bering Sea outer continental margin

ICES Journal of Marine Science ◽

10.1093/icesjms/fst054 ◽

2013 ◽

Vol 70 (6) ◽

pp. 1181-1197 ◽

Cited By ~ 2

Author(s):

Laura A. Wigand ◽

Terrie Klinger ◽

Miles G. Logsdon

Keyword(s):

Spatial Pattern ◽

Continental Margin ◽

Bering Sea ◽

Local Ecological Knowledge ◽

Environmental Data ◽

Ecological Knowledge ◽

Trawl Survey ◽

Spatial Pattern Analysis ◽

Catch Per Unit Effort ◽

Eastern Bering Sea

Abstract Wigand, L. A., Klinger, T., and Logsdon, M. G. 2013. Patterns in groundfish abundance along the Eastern Bering Sea outer continental margin. – ICES Journal of Marine Science, 70: 1181–1197. Place-based management approaches require understanding the spatial arrangement and interaction of elements. To address this need, we explored the utility of spatial-pattern analysis to understand the distribution of groundfish in the Eastern Bering Sea outer continental margin. We divided this region into discrete geomorphological units to explore spatial pattern on a range of scales. We used groundfish catch per unit effort (cpue) trawl survey data collected in four years to quantify spatial autocorrelation. Global statistics indicated that groundfish cpue was dominated by clusters of low values in all years. Local statistics showed that clusters of low values in groundfish cpue were confined to the southern portion of the study area, while clusters of high values varied across the study area. Outliers were most commonly found in close proximity to the shelf–slope break. Our results reveal the existence of spatial dependency in groundfish abundance and demonstrate that spatial analysis can be used to better understand spatial arrangements of these and other living marine resources, and to quantify and validate the local ecological knowledge of resource users. Our results indicate the feasibility of using spatially explicit tools to improve integration and visualization of marine environmental data for purposes of management and conservation.

Download Full-text

Catchability of snow crab (Chionoecetes opilio) by the eastern Bering Sea bottom trawl survey estimated using a catch comparison experiment

Canadian Journal of Fisheries and Aquatic Sciences ◽

10.1139/cjfas-2013-0100 ◽

2013 ◽

Vol 70 (12) ◽

pp. 1699-1708 ◽

Cited By ~ 11

Author(s):

David A. Somerton ◽

Kenneth L. Weinberg ◽

Scott E. Goodman

Keyword(s):

Experimental Data ◽

Bering Sea ◽

Weighted Average ◽

Assessment Model ◽

Bottom Trawl ◽

Snow Crab ◽

Trawl Survey ◽

Chionoecetes Opilio ◽

Bottom Trawl Survey ◽

Eastern Bering Sea

Catchability of the eastern Bering Sea (EBS) bottom trawl survey for snow crab (Chionoecetes opilio) was estimated from experimental data to provide a constraint on the survey catchability parameters in the stock assessment model. The experiment utilized a second fishing vessel to conduct side-by-side trawling with each of two survey vessels at 92 stations using an experimental trawl assumed to capture all crabs in its path. Trawl efficiency, or the captured proportion of crabs in the trawl path, was estimated for the 83-112 Eastern otter trawl from experimental data using a nonparametric smooth function of carapace width, sediment size, and depth. Survey catchability was then estimated as the catch-weighted average of the predicted trawl efficiency at all 275 survey stations where snow crabs were captured. The fitted model indicated that trawl selectivity was greater in sand than mud and greater in shallow water than deep. At a carapace widths >95 mm, the minimum commercial size limit, the estimated survey catchability of males is considerably less than previously reported.

Download Full-text

Relating spatial distributions of acoustically determined patches of fish and plankton: data viewing, image analysis, and spatial proximity

Canadian Journal of Fisheries and Aquatic Sciences ◽

10.1139/f99-206 ◽

1999 ◽

Vol 56 (S1) ◽

pp. 188-198 ◽

Cited By ~ 13

Author(s):

Gordon Swartzman ◽

Richard Brodeur ◽

Jeffrey Napp ◽

Danny Walsh ◽

Roger Hewitt ◽

...

Keyword(s):

Bering Sea ◽

Generalized Additive Models ◽

Biological Data ◽

Additive Models ◽

Spatial Proximity ◽

Connected Component ◽

Proximity Measure ◽

Nonparametric Regression Model ◽

Acoustic Data ◽

Acoustic Images

We developed a point-and-click acoustic data viewer (FishViewer) for exploratory comparison of up to three acoustic survey transects (or three frequencies) at a time and other environmental and biological data (e.g., surface temperature and seabird abundance). FishViewer also contains image-processing tools (e.g., morphological and threshold filters) for distinguishing between fish shoals and plankton patches and for patch identification. These tools and methods are illustrated using survey data collected at three frequencies (38, 120, and 200 kHz) near the Pribilof Islands, Bering Sea, during September 1995. Data were also visualized by converting the patches identified in the acoustic images to polygons, showing the boundaries of each patch using a connected component algorithm. Proximity between these fish shoal and plankton patch polygons was examined statistically using an interval-based nonparametric regression model (generalized additive models) and a distance-based proximity measure. The methods presented for data refinement, visualization, and the establishment of fish-plankton patch proximity serve as a paradigm for scale-robust hypothesis formulation and testing of spatial patterns of fish and plankton.

Download Full-text

Evaluating total uncertainty for biomass- and abundance-at-age estimates from eastern Bering Sea walleye pollock acoustic-trawl surveys

ICES Journal of Marine Science ◽

10.1093/icesjms/fsw054 ◽

2016 ◽

Vol 73 (9) ◽

pp. 2208-2226 ◽

Cited By ~ 4

Author(s):

Mathieu Woillez ◽

Paul D. Walline ◽

James N. Ianelli ◽

Martin W. Dorn ◽

Christopher D. Wilson ◽

...

Keyword(s):

Bering Sea ◽

Walleye Pollock ◽

Target Strength ◽

Acoustic Backscatter ◽

Trawl Survey ◽

Data Types ◽

Total Uncertainty ◽

Functional Relationships ◽

Eastern Bering Sea ◽

Trawl Surveys

Abstract A comprehensive evaluation of the uncertainty of acoustic-trawl survey estimates is needed to appropriately include them in stock assessments. However, this evaluation is not straightforward because various data types (acoustic backscatter, length, weight, and age composition) are combined to produce estimates of abundance- and biomass-at-age. Uncertainties associated with each data type and those from functional relationships among variables need to be evaluated and combined. Uncertainty due to spatial sampling is evaluated using geostatistical conditional (co-) simulations. Multiple realizations of acoustic backscatter were produced using transformed Gaussian simulations with a Gibbs sampler to handle zeros. Multiple realizations of length frequency distributions were produced using transformed multivariate Gaussian co-simulations derived from quantiles of the empirical length distributions. Uncertainty due to errors in functional relationships was evaluated using bootstrap for the target strength-at-length and the weight-at-length relationships and for age–length keys. The contribution of each of these major sources of uncertainty was assessed for acoustic-trawl surveys of walleye pollock in the eastern Bering Sea in 2006–2010. This simulation framework allows a general computation for estimating abundance- and biomass-at-age variance–covariance matrices. Such estimates suggest that the covariance structure assumed in fitting stock assessment models differs substantially from what careful analysis of survey data actually indicate.

Download Full-text

Geostatistical simulations of eastern Bering Sea walleye pollock spatial distributions, to estimate sampling precision

ICES Journal of Marine Science ◽

10.1093/icesjms/fsl045 ◽

2007 ◽

Vol 64 (3) ◽

pp. 559-569 ◽

Cited By ~ 24

Author(s):

Paul D. Walline

Keyword(s):

Bering Sea ◽

Walleye Pollock ◽

Geostatistical Simulation ◽

Normal Space ◽

Total Biomass ◽

Spatial Distributions ◽

Acoustic Data ◽

Eastern Bering Sea ◽

Abundance Estimates ◽

Length Frequency Data

Abstract Walline, P. D. 2007. Geostatistical simulations of eastern Bering Sea walleye pollock spatial distributions, to estimate sampling precision. – ICES Journal of Marine Science, 64: 559–569. Sequential Gaussian and sequential indicator geostatistical simulation methods were used to estimate confidence intervals (CIs) for biomass estimates from six echo-integration trawl surveys of eastern Bering Sea walleye pollock (Theragra chalcogramma) biomass. Uncertainty in the acoustic and the length frequency data was combined in the calculation of CIs. Sampling in 2002 provided evidence for isotropy in the spatial distribution. Variogram models were characterized by long ranges (75–122 nautical miles for non-zero acoustic data, for example) compared with the smallest dimension of the survey area (∼100 nautical miles) and small nugget effects (∼20% of the semi-variance in transformed normal space for acoustic data). The 95% CIs obtained for the abundance estimates did not vary greatly between years and were similar to those from a one-dimensional transitive geostatistical analysis, i.e. ± 5–9% of estimated total biomass.

Download Full-text