Modeling nearshore fish habitats using Alaska as a regional case study
Introduction
The nearshore marine environment, including coastal and inshore areas of the continental shelf, provides important habitat for many marine species during at least part of their life history. Therefore, understanding the spatial extent and ecological importance of nearshore habitats for marine species allows natural resource managers to protect and restore habitats to ensure the sustainable use of natural resources. In the United States (U.S.), the Magnuson-Stevens Fishery Conservation and Management Act (MSA) requires that the National Marine Fisheries Service (NMFS) and regional fishery management councils describe and map essential fish habitat (EFH) – habitats that are necessary to fish and shellfish species throughout their life history – and recommend actions to conserve these areas from adverse human impacts1 . NMFS may provide conservation recommendations such as gear modifications or restrictions and time and area closures in the case of fishing activities and alternative site selection and timing of work in the case of non-fishing activities (Limpinsel et al., 2017). EFH regulations provide an approach to organize the information necessary to describe and identify EFH, where designations rely at a minimum on distribution data (i.e., EFH Level 1 information). Whenever possible, designations are based on more detailed population-level information, including habitat-related densities or abundance (Level 2), survival, growth and reproduction within habitats (Level 3), and production rates by habitat (Level 4). EFH designations are periodically reviewed and updated by the regional fishery management councils to ensure that the best available scientific information is used to describe and identify EFH (NMFS National Standard 2 Scientific Information2).
Alaska is the largest U.S. state, with an intricate coastline composed of many bays, fjords and islands, extending a distance greater than the coastlines of all the other U.S. states combined (Shalowitz, 1964). Alaska spans five large marine ecosystems, including the Gulf of Alaska (GOA), Aleutian Islands, Bering Sea, and Chukchi and Beaufort seas of the U.S. Arctic region (Fig. 1a). Coastal areas of Alaska host a diverse array of shallow, nearshore habitats including eelgrass (Zostera marina) and kelp beds, sand beaches, and exposed or sheltered rocky shores (Dean et al., 2000; Johnson et al., 2012; Pirtle et al., 2012), which are affected by human activities that take place nearshore or in upland terrestrial locations. Human activities that affect Alaska nearshore habitats include, among others, urban development, oil and gas exploration and extraction, mining, timber harvest, municipal and industrial waste, and vessel traffic from a variety of industries (Harris et al., 2008; Johnson et al., 2012; Limpinsel et al., 2017).
Alaska nearshore areas provide habitat for numerous fish species. In Alaska, like in other marine regions, many fish and shellfish species undertake ontogenetic habitat shifts, whereby individuals migrate offshore into deeper waters as they grow, to meet the ecological demands of survival, growth, or reproduction. Thus, Alaska nearshore habitats serve as the nursery areas (i.e., the distribution areas of the early juvenile life stage) for many ecologically and economically important demersal fish and shellfish species with offshore life stages that are targeted by fisheries. These include commercially important gadoids (e.g., Pacific cod Gadus macrocephalus; walleye pollock Gadus chalcogrammus; Abookire et al., 2001; Laurel et al., 2009), flatfishes (Norcross et al., 1999; Hurst, 2016), sablefish (Anoplopoma fimbria; Courtney and Rutecki, 2011), and red king crab (Paralithodes camtschaticus; Loher and Armstrong, 2000).
The early juvenile stages of demersal fish populations have long been identified as being vulnerable to human impacts in nearshore areas (Beck et al., 2003; Lellis-Dibble et al., 2008; Johnson et al., 2012). However, species of Pacific salmon, and forage fishes such as Pacific herring (Clupea pallasii), Pacific sand lance (Ammodytes personatus), capelin (Mallotus villosus), and eulachon (Thaleichthys pacificus) are at least as vulnerable as the early juvenile stages of demersal fishes to disturbances in nearshore areas, as they use nearshore areas for feeding and shelter as juveniles as well as spawning as adults (Pahlke, 1985; Robards et al., 1999; Cooney, 2007; Harris et al., 2008; Johnson et al., 2008; Miller et al., 2016). The consequences of human impacts on nearshore areas for forage fish are concerning not only for forage fish populations, but also for the many fish, seabird, and marine mammal predator populations that prey upon them (Springer and Speckman, 1997; Mundy and Hollowed, 2005). Thus, there is a critical need to model and map Alaska nearshore areas because of their importance as habitat for numerous marine species of economic and ecological importance.
In the present study, we develop and demonstrate a practical approach to model and map Alaska nearshore fish habitats that rely on binomial and delta-Gamma generalized additive models (GAMs), fish survey data collected by multiple gear types, and very fine-scale habitat information. Our modeling approach leverages the information provided by two large databases for Alaska: a large fish survey database called the “NMFS Nearshore Fish Atlas of Alaska database” (hereafter referred to as the “Nearshore Fish Atlas” or the “NFA”; National Marine Fisheries Service (NMFS, 2020a); and a large habitat database called ShoreZone (Cook et al., 2017; National Marine Fisheries Service (NMFS, 2020b). Although our approach was designed for Alaska nearshore areas, it could easily be applied to other marine regions where survey and habitat databases are available. Our modeling approach was ddeveloped specifically to be simple enough to be employed and adapted by fisheries scientists and resource managers for a variety of purposes. There are two key steps to our approach: (1) constructing a strip of coastline consisting of ∼10 m coastline segments for the study area to be able to generate predictions from fitted GAMs, which is referred to as a “predictive coastline”; and (2) using the fitted and validated GAMs and the predictive coastline to predict spatial patterns of probability of encounter and density at the very fine spatial scales at which localized ecological processes operate for life stages of demersal fishes in the nearshore areas. In the following, we first provide brief overviews of the study areas, the NFA and ShoreZone. We then detail our modeling approach, before demonstrating it for Pacific cod early juveniles of the northern southeastern Alaska (NSEA) area and walleye pollock early juveniles of Prince William Sound (PWS) (Figs. 1b-c). Next, we discuss how the information generated by our approach will support natural resource management in Alaska, and we highlight some avenues for future ecological and modeling research.
Section snippets
Study areas
In the present study, we focus on two nearshore areas of the GOA: the northern southeastern Alaska (NSEA) area in the case of Pacific cod (Fig. 1b) and Prince William Sound (PWS) in the case of walleye pollock (Fig. 1c). The NSEA area covers southeastern Alaska from about 56–59 degrees latitude, with the exclusion of inside waters around Juneau and outside waters near Yakutat. We constructed predictive coastlines (strips of coastline consisting of ∼10 m coastline segments that are needed to
Application to Pacific cod early juveniles of the NSEA area
The final binomial GAM of Pacific cod early juveniles of the NSEA area included the effect of year, the tensor product smooth between eastings and northings, and the eelgrass factor. Coastal type, wave exposure, rockweed, and soft brown kelps were all found to have a non-significant effect on the probability of encounter of Pacific cod early juveniles. This model explained 39.1% of the deviance in the encounter/non-encounter data. The median AUC of the final binomial GAM equaled 0.90 (CI:
Discussion
In the present study, we demonstrated the utility of compiling large fish survey and habitat databases for nearshore ecosystems by using this information in a practical approach that employs species distribution models (SDMs) to generate very fine-scale nearshore EFH information. Specifically, we designed a GAM approach that used the NFA and ShoreZone databases to produce very fine-scale maps of probability of encounter (EFH level 1 information) and density (EFH level 2 information) for
Funding
This work was funded by the NOAA, National Marine Fisheries Service (NMFS), Office of Habitat Conservation.
CRediT authorship contribution statement
Arnaud Grüss: Conceptualization, Methodology, Software, Validation, Formal analysis, Data curation, Writing - original draft, Visualization. Jodi L. Pirtle: Conceptualization, Writing - review & editing, Funding acquisition. James T. Thorson: Conceptualization, Methodology, Software, Writing - review & editing, Funding acquisition. Mandy R. Lindeberg: Conceptualization, Writing - review & editing. A. Darcie Neff: Data curation, Writing - review & editing. Steve G. Lewis: Software, Data
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The scientific results and conclusions, as well as any views or opinions expressed herein, are those of the author(s) and do not necessarily reflect those of NOAA or the Department of Commerce. This work was funded by the NOAA, National Marine Fisheries Service (NMFS), Office of Habitat Conservation. The update to the Nearshore Fish Atlas of Alaska was funded by the Alaska Region and Alaska Fisheries Science Center (AFSC)’s Essential Fish Habitat Research Plan. We are very grateful to the
References (109)
- et al.
Non-stationary Gaussian models with physical barriers
Spat. Stat.
(2019) - et al.
Generalized additive modelling and zero inflated count data
Ecol. Modell.
(2002) - et al.
Harbour porpoise distribution can vary at small spatiotemporal scales in energetic habitats
Deep. Sea Res. Part II
(2017) - et al.
Understanding patterns of fish backscatter, size and density around petroleum platforms of the US Gulf of Mexico using hydroacoustic data
Fish. Res.
(2021) - et al.
Eelgrass patch size and proximity to the patch edge affect predation risk of recently settled age 0 cod (Gadus)
J. Exp. Mar. Biol. Ecol.
(2009) - et al.
Using delta generalized additive models to produce distribution maps for spatially explicit ecosystem models
Fish. Res.
(2014) - et al.
Representing species distributions in spatially-explicit ecosystem models from presence-only data
Fish. Res.
(2019) - et al.
Evaluation of the impacts of different treatments of spatio-temporal variation in catch-per-unit-effort standardization models
Fish. Res.
(2019) - et al.
Making the most of available monitoring data: A grid-summarization method to allow for the combined use of monitoring data collected at random and fixed sampling stations
Fish. Res.
(2020) Shallow-water habitat use by Bering Sea flatfishes along the central Alaska Peninsula
J. Sea Res.
(2016)
Stock, recruitment and moderating processes in flatfish
J. Sea Res.
Comparative habitat associations in juvenile Pacific cod and other gadids using seines, baited cameras and laboratory techniques
J. Exp. Mar. Biol. Ecol.
Temporal and ontogenetic shifts in habitat use of juvenile Pacific cod (Gadus macrocephalus)
J. Exp. Mar. Biol. Ecol.
Effects of habitat complexity and relative larval supply on the establishment of early benthic phase red king crab (Paralithodes camtschaticus Tilesius, 1815) populations in Auke Bay, Alaska
J. Exp. Mar. Biol. Ecol.
Think outside the grids: an objective approach to define spatial strata for catch and effort analysis
Fish. Res.
Evaluating the predictive performance of habitat models developed using logistic regression
Ecol. Modell.
Habitat suitability models for groundfish in the Gulf of Alaska
Deep. Sea Res. Part II
Standardization of catch and effort data in a spatially-structured shark fishery
Fish. Res.
A Beginner’s Guide to Generalised Additive Mixed Models With R. Highland Statistics, Newburgh, UK
Juvenile groundfish habitat in Kachemak Bay, Alaska, during late summer
AlasSka Fishery Res. Bull.
Habitat associations and diet of young-of-the-year Pacific cod (Gadus macrocephalus) near Kodiak
Alaska. Marine Biology
The role of nearshore ecosystems as fish and shellfish nurseries
Issues Ecol.
Seasonal composition and abundance of juvenile and adult marine finfish and crab species in the nearshore zone of Kodiak Island’s eastside during April 1978 through March 1979, in: outer Continental Shelf Environmental Assessment Program
Final Reports of Principal Investigators
Environmental and structural drivers of fish distributions among petroleum platforms across the US Gulf of Mexico
Mar. Coast. Fish. Dyn. Manag. Ecosyst. Sci.
Trade-offs in covariate selection for species distribution models: a methodological comparison
Ecography
ShoreZone Coastal Imaging and Habitat Mapping Protocol. Report Prepared by Coastal and Ocean Resources, Victoria, bc, Canada
Pacific herring
Applying Bayesian spatiotemporal models to fisheries bycatch in the Canadian Arctic
Can. J. Fish. Aquat. Sci.
Inshore Movement and Habitat Use by Juvenile Sablefish Anoplopoma fimbria, Implanted With Acoustic Tags in Southeast Alaska. AFSC Processed Report 2011-01
The distribution of nearshore fishes in kelp and eelgrass communities in Prince William Sound, Alaska: associations with vegetation and physical habitat characteristics
Environ. Biol. Fishes
Spatio-temporal analysis of commercial trawler data using General Additive models: patterns of Loliginid squid abundance in the north-east Atlantic
Ices J. Mar. Sci.
Methods to account for spatial autocorrelation in the analysis of species distributional data: a review
Ecography
Substrate mapping to inform ecosystem science and marine spatial planning around the Main hawaiian Islands
A Refined Description of Essential Fish Habitat for Pacific Salmon Within the U.S. Exclusive Economic Zone in Alaska
Species distribution models: ecological explanation and prediction across space and time
Annu. Rev. Ecol. Evol. Syst.
Novel methods improve prediction of species’ distributions from occurrence data
Ecography
Spatial distribution and conservation of speckled hind and warsaw grouper in the Atlantic Ocean off the southeastern US
PLoS One
Pacific sand lance, Ammodytes hexapterus, with notes on related Ammodytes species
Employing predictive spatial models to inform conservation planning for seabirds in the Labrador Sea
Front. Mar. Sci.
Exploring a new class of non-stationary spatial Gaussian random fields with varying local anisotropy
Stat. Sin.
Developing spatio-temporal models using multiple data types for evaluating population trends and habitat usage
Ices J. Mar. Sci.
Exploring the spatial distribution patterns of South African Cape hakes using generalised additive models
Afr. J. Mar. Sci.
Assisting Ecosystem-Based Fisheries Management Efforts Using a Comprehensive Survey Database, a Large Environmental Database, and Generalized Additive Models
Mar. Coast. Fish. Dyn. Manag. Ecosyst. Sci.
Producing distribution maps for a spatially-explicit ecosystem model using large monitoring and environmental databases and a combination of interpolation and extrapolation
Front. Mar. Sci.
Monitoring programs of the US Gulf of Mexico: inventory, development and use of a large monitoring database to map fish and invertebrate spatial distributions
Rev. Fish Biol. Fish.
The meaning and use of the area under a receiver operating characteristic (ROC) curve
Radiology
Cross-shelf habitat occupancy probabilities for juvenile groupers in the Florida keys coral reef ecosystem
Mar. Coast. Fish. Dyn. Manag. Ecosyst. Sci.
Eelgrass Habitat and Faunal Assemblages in the City and Borough of Juneau, Alaska. U.S. Dep. Commer
Generalized Additive Models
Transport, distribution, and abundance of larval and juvenile walleye pollock (Theragra chalcogramma) in the western Gulf of Alaska
Can. J. Fish. Aquat. Sci.
Cited by (14)
Research progress and framework on the simulation and assessment of fish habitat degradation in lakes
2024, Ecological IndicatorsCatch per unit effort modelling for stock assessment: A summary of good practices
2024, Fisheries ResearchCatchment-scale stream network spatio-temporal models, applied to the freshwater stages of a diadromous fish species, longfin eel (Anguilla dieffenbachii)
2023, Fisheries ResearchCitation Excerpt :Then, spatial and spatio-temporal variation and, therefore, probability of encounter were predicted over the whole study area using spatial interpolation (Grüss et al., 2020a). There was then a “transferability” issue where predictions could not be made at a very fine spatial resolution, as the estimation of spatial and spatio-temporal autocorrelation was based on data points that were mostly very far from one another (Grüss et al., 2021a). Thus, Charsley et al. (in press) was unable to predict probabilities of encounter in the Waitaki catchment, but still managed to estimate low (0–0.4) probabilities of encounter for locations close to the Waitaki catchment for the period 1974–2014.
Long-term monitoring and integrated research – Understanding ecosystem processes in the Gulf of Alaska
2022, Deep-Sea Research Part II: Topical Studies in OceanographyEvaluating the impacts of environmental and fishery v ar iability on the distribution of big e y e tuna in the Pacific Ocean
2023, ICES Journal of Marine Science