Estuarine, Coastal and Shelf Science 310 (2024) 109013

Contents lists available at ScienceDirect

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier.com/locate/ecss

Fish distribution shifts due to climate change in the Northeast Atlantic:

Using a hierarchical filtering approach on marine-estuarine
opportunist species
Anaïs Janc a,* , Chloé Dambrine a , Patrick Lambert a , Géraldine Lassalle a , Mario Lepage a ,
Jérémy Lobry a , Maud Pierre a , Trond Kristiansen b,c, Momme Butenschön d , Henrique N. Cabral a
a
INRAE, UR EABX, 33612 Cestas, France
b
Farallon Institute for Advanced Ecosystem Research, Petaluma, CA, United States
c
Norwegian Institute for Water Research, Oslo, Norway
d
Fondazione Centro Euro-Mediterraneo Sui Cambiamenti Climatici, CMCC, Bologna, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: Marine-estuarine opportunist (MEO) species are fish that occur in the continental shelf and use estuaries and/or
Species distribution models shallow coastal areas as nurseries. These commercially important resources are facing significant environmental
biomod2 modifications caused by direct and/or indirect anthropogenic climate change effects. In this study, we investi­
Flatfish species
gated the directionality and the magnitude of the distribution shifts (i.e., range size, gravity centroids, and
Meagre
Seabass
margins) in marine environment suitability for six main MEO fish species within the Northeast Atlantic expected
Temperate marine ecoregion for the end of the 21st century. In the framework of this study, we have distinguished ‘sub-boreal’ from ‘sub-
tropical’ species. The ‘hierarchical filters’ concept was adopted for modelling the potential species distributions
and combined the predictions of i) a bioclimatic model with ii) a habitat model. The bioclimatic model is based
on large-scale and time-variant variables while variables of the habitat model are fine-grained and time-
invariant. Two Intergovernmental Panel on Climate Change (IPCC) scenarios are tested: an intermediate
(SSP2-4.5) and a pessimistic one (SSP5-8.5). We applied this framework using international databases of
biodiversity occurrences, ensemble forecasting producing consensual predictions, and innovative indices of
distribution shifts. A visible north-westward shift was predicted for all six species in our study area. However, the
northward expansion was greater for ‘sub-tropical’ than for ‘sub-boreal’ species due to faster gravity centroid
displacement shifts and faster margins shifts. These range shifts may lead to major ecological impacts (e.g.,
changes in recruitment to estuarine and coastal nurseries, as well as changes in spawning grounds) that may alter
populations’ connectivity.

Abbreviations: ANN, Artificial Neural Networks; AWA, Abundance-Weighted Average; COGDs, Centres Of Gravity of the Distributions; COGEs, Centres Of Gravity
of leading edge Expansions; COGCs, Centres Of Gravity of trailing edge Contractions; CMIP6, Coupled Model Intercomparison Project Phase 6; CTA, Recursive
Partitioning; DATRAS, Database of Trawl Surveys; EMODnet, European Marine Observation and Data Network; EUNIS, European Nature Information System; FDA,
Flexible Discriminant Analysis; GAM, Generalized Additive Models; GBIF, Global Biodiversity Information Facility; GBM, Boosted Regression Trees; GLM, Gener­
alized Linear Models; ICES, International Council for the Exploitation of the Sea; IFREMER, Institut Français de Recherche pour l’Exploitation de la Mer; IPPC,
Intergovernmental Panel on Climate Change; MARS, Multivariate Adaptive Regression Splines; MaxEnt, Maximum Entropy formalism; MEO, Marine-Estuarine
Opportunist; OBIS, Ocean Biodiversity Information System; PCA, Principal Component Analysis; RF, Random Forest; SDMs, Species Distribution Models; SSP, Shared
Socioeconomic Pathways; SRC, Species Range Change; TSS, True Skill Statistic; VertNet, Vertebrate Biodiversity Data Networks.
* Corresponding author.
E-mail addresses: [email protected] (A. Janc), [email protected] (C. Dambrine), [email protected] (P. Lambert), [email protected]
(G. Lassalle), [email protected] (M. Lepage), [email protected] (J. Lobry), [email protected] (M. Pierre), [email protected]
(T. Kristiansen), [email protected] (M. Butenschön), [email protected] (H.N. Cabral).

https://doi.org/10.1016/j.ecss.2024.109013
Received 23 May 2024; Received in revised form 29 October 2024; Accepted 29 October 2024
Available online 30 October 2024
0272-7714/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

1. Introduction centroids, and margins) (Morato et al., 2020; Soultan et al., 2022). In
this context, there is a crucial need to understand the climate-induced
Marine fish species are experiencing human-induced alterations in shifts in marine environment suitability for populations of MEO fish
their ecophysiological and/or biological processes in response to climate species within the Northeast Atlantic, covering a large part of their
change (Poloczanska et al., 2013). These alterations include changes in ranges to anticipate the potential future impacts that these biological
demography and phenology (e.g., a significant increase in the propor­ redistributions may have on ecosystem functioning and human
tion of small-sized fish species – Perry et al., 2005; and 20–30% body well-being (Pecl et al., 2017).
mass reduction – Pauly and Cheung, 2018) as well as strategies of energy Species Distribution Models (SDMs) represent valuable tools to
allocation (Poloczanska et al., 2013); seasonal shifts in life cycle simulate the environmental suitability for species settlement and sur­
(Daufresne et al., 2009); and alterations in sensory system resulting in vival. SDMs have been applied exponentially, notably in the Northeast
swimming behaviour changes (Servili et al., 2022). All of these changes Atlantic over the past fifteen years (Melo-Merino et al., 2020). They are
explain why climate change currently turned 45% of marine species to mainly based on a correlative approach of the species-environment re­
be at risk of extinction and caused a decrease in the amount of ‘sus­ lationships relating species observations to environmental variables
tainable’ fish stocks by up to 4% since 1930s (Poloczanska et al., 2013; through statistically derived response curves (Guisan et al., 2017).
Penn and Deutsch, 2022). However, marine fish species also adjust their Correlative models thus allow approximating the potential niche of the
spatial distributions to follow the movements of their ecological niche species (sensu Hutchinson (1957)). They have been the most widely used
and/or prey at the same rate or not as projected climate change will models due to their simplicity of data requirements and to constant
strike (Poloczanska et al., 2013). technical progress (Peterson et al., 2015). This popularity is accompa­
Tropicalization is a prevalent process explaining climate change nied by numerous publications and the dissemination of data and tools
induced shift in marine communities (McLean et al., 2021). It has been in open access and provides relevant feedback on the most relevant
well documented in some temperate marine ecoregions as in the approaches. Indeed, it appeared that these models perform better and
Northeast Atlantic Ocean (Costa et al., 2014). Northeast Atlantic is a are more realistic when they account for some important points such as
biogeographic transition zone where fish species of different thermal the species vertical habitat dimension (Hattab et al., 2014; Ben Rais
affinity co-occur, and this area has been increasingly exposed to climate Lasram et al., 2020), and when they combine both habitat (e.g., sedi­
change particularly since the mid-20th century (Levitus et al., 2000; ments, depth) and bioclimatic (e.g., temperature, salinity) variables
Rutterford et al., 2023). Consequently, observed northward species (Sohl, 2014). In most cases, the inclusion of habitat dimension requires
range shifts (Perry et al., 2005; Pinsky et al., 2013), and/or shifts of the application of a hierarchical approach that allows respecting the
species’ distributional cores to deeper and more remote areas (Dulvy appropriate spatial extent and resolution at which coarser scale biocli­
et al., 2008), and/or species range size changes (Thorson et al., 2016) matic and finer-scale habitat variables are expected to influence the
correspond to large-scale marine habitat losses and/or gains (Morato underlying ecological processes particularly for highly mobile species
et al., 2020; Ramos Martins et al., 2021). In terms of community, these (Guisan et al., 2017; Melo-Merino et al., 2020). Moreover, hierarchical
have resulted in northern cold-water species’ declines due to their range filters avoid overfitting of a single model with too many environmental
contractions, and/or arrival of new or emerging southern warm-water predictors (Randin et al., 2006). Finally, the application of an ensemble
species due to their range expansions leading to a tropicalization of forecasting procedure for each hierarchical filter allows the variations in
fish communities (Le Marchand et al., 2020). the accuracy of predictions produced from several statistical techniques
Among the most valuable marine fish species, the marine-estuarine to be considered and provides more accurate fitting for a set of training
opportunist (MEO) fish species constitute one fifth of the species rich­ data (Hao et al., 2019).
ness of estuarine systems along the French and Iberian coasts of the The main purpose of this study was to investigate the effects of
Northeast Atlantic (Teichert et al., 2017). Indeed, MEO fish species are climate change on the environment suitability distribution for six
unique species distinguished by their complex life cycle involving bentho-demersal MEO fish species in the Northeast Atlantic Ocean. In
dependence to estuarine or coastal environments (Potter et al., 2015; the framework of this study, we have distinguished ‘sub-boreal’ (sea
Cabral et al., 2022). Specifically, juveniles of MEO fish species regularly bass, Dicentrarchus labrax; flounder, Platichthys flesus; plaice, Pleuro­
enter and use estuaries (or sometimes protected nearshore shallow nectes platessa; and common sole, Solea solea) from ‘sub-tropical’
coastal areas) as nurseries; while adults use deep coastal areas and the (meagre, Argyrosomus regius; and Senegalese sole, Solea senegalensis)
continental shelf as migratory, feeding, and/or spawning grounds giving species. These projected distributions were derived under present and
rise to seasonal migration to and from estuaries (Potter et al., 2015; late 21st century conditions based on the SSP2-4.5 (intermediate) and
Whitfield et al., 2023). MEO fish species is an ecologically (e.g., they the SSP5-8.5 (pessimistic) possible emission scenarios proposed by the
occur in high abundance in estuarine and coastal systems with a sea­ Intergovernmental Panel on Climate Change (IPCC – IPCC, 2021). These
sonal periodicity, consuming high quantities of benthic macro­ projections were made using the ‘hierarchical filters’ concept combining
invertebrates and other prey) and socio-economically (e.g., artisanal and both bioclimatic (with large-scale and time-variant variables) and
commercial fisheries, recreational fisheries, and nutrition security) habitat (with fine grained and supposedly time-invariant variables)
important group (Seitz et al., 2014). Their climate vulnerability was components of the potential species niche from consensual ensemble
classified from moderate to high depending on the species considered forecasting. Three main hypotheses were tested: (i) distribution shifts in
(Cheung and Oyinlola, 2018). For some of these species, a significant environment suitability in response to ocean warming, i.e., northward
change in their abundance in estuaries along the French Atlantic coast range shifts, deeper water shift, and/or seaward shift; (ii) a range
due to climate warming has already been documented (Hermant et al., contraction for the four ‘sub-boreal’ species and a range expansion for
2010). Most studies have focused on the impact of climate change on the two ‘sub-tropical’ species; and (iii) stronger effects under the SSP5-
larval dispersion on the continental shelf before the recruitment of 8.5 scenario than under the SSP2-4.5 scenario. Methodological choices
larvae and juveniles in estuarine nurseries (e.g., Cabral et al., 2021), the along this work used an existing framework of hierarchical modelling
suitability of estuarine habitats for juveniles (e.g., Vinagre et al., 2006) (Ben Rais Lasram et al., 2020 – https://github.com/TarekHattab/SDM)
as well as their contribution to adult populations (e.g., Le Pape et al., but incorporated improvements and/or adaptations to fit with the MEO
2003). However, the suitability of marine environment for adult pop­ species ecology and the ecology of climate change in terms of shifting
ulations on a scale covering their distribution ranges have been far less distributions.
studied. Among the studies looking at shifts in environment suitability
distribution, few have analyzed the directionality and the magnitude of
different distributional characteristics (e.g., range size, gravity

2
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

2. Material and methods downloaded from the European Marine Observation and Data Network
(EMODnet) Bathymetry (https://www.emodnet-bathymetry.eu/) at ~
2.1. Data acquisition and processing, and exploratory data analyses 1/1000th degree grid resolution, and upscaled by bilinear resampling to
1/100th degree (~750–800 m) grid resolution. The seabed type was
2.1.1. MEO fish occurrence data extracted from the EMODnet Geology (https://www.emodnet-geology.
For bioclimatic suitability distribution modelling, presence-only re­ eu/) at 1:1,000,000 scale, rasterized by means of the R terra package
cords for the six MEO fish species were downloaded from four interna­ (R Core Team, 2021), and resampled to the same spatial resolution as
tional open-access datasets of biodiversity occurrences based on the bathymetry grid. Seven EUNIS substrate classes were represented: (i)
participative science projects: (i) the Global Biodiversity Information mud, (ii) sandy mud, (iii) muddy sand, (iv) sand, (v) coarse-grained
Facility (GBIF - https://www.gbif.org/), (ii) the Ocean Biodiversity In­ sediment, (vi) mixed sediment, and (vii) rock and boulders. The two
formation System (OBIS - https://obis.org/), (iii) the Vertebrate Biodi­ habitat grids were then cropped to the desired spatial extent (n = 0.01◦
versity Data Networks (VertNet - http://www.vertnet.org/index.html), × 0.01◦ 1,647,760 cells – Fig. A.1). Due to the potential existence of
and (iv) the INaturalist (https://www.inaturalist.org/) via the rgbif, multicollinearity between the two habitat variables and convergence
robis, rvernet and rinat packages in R (R Core Team, 2021). For habitat issues of some statistical techniques from mixed variable types, a mixed
suitability distribution modelling, presence-only occurrences were Principal Component Analysis (PCA) following the ordination method of
retrieved from research samplings and/or monitoring of professional Hill and Smith (1976) was therefore performed via the R ade4 package
fisheries from two datasets: (i) the MigrenMer database (Elliott et al., (R Core Team, 2021). This method allowed defining the habitat com­
2023) regrouping the open-access fisheries-independent Database of ponents to use in the modelling approach (Ben Rais Lasram et al., 2020;
Trawl Surveys (DATRAS) maintained by the International Council for Le Marchand et al., 2020).
the Exploration of the Sea (ICES - https://www.ices.dk/data/d
ata-portals/Pages/DATRAS.aspx), the fisheries-independent French 2.2. Bioclimatic and habitat suitability distribution modelling
scientific surveys collected by the Institut Français de Recherche pour
l’Exploitation de la Mer (IFREMER) and the fisheries-dependent data 2.2.1. Filtrations and pseudo-absence data generation
belonging to the ObsMer program available upon request from the Bioclimatic and habitat filters were based on Hutchinson’s ‘niche-
French Ministry of Fisheries and Aquaculture especially for European biotope duality’ (Colwell and Rangel, 2009). This procedure reduces the
flounder; and (ii) the online DATRAS for data that were not included in under- or over-representation of some environmental conditions in
the MigrenMer database via the R icesDATRAS package (R Core Team, models by assigning equal weight to over- and under-sampled areas (Ben
2021). For both hierarchical filters, a careful data pre-processing was Rais Lasram et al., 2020). The bioclimatic component of the potential
applied for each species (see Text S1 for more details). niche was estimated by a restricted convex hull excluding the extreme
bioclimatic conditions experienced by the species from the 2.5% and
2.1.2. Environmental variables 97.5% percentiles of each bioclimatic variable (Cornwell et al., 2006;
The bioclimatic component of the potential niche was predicted from Schickele et al., 2020). The following resolutions were selected: 0.5-log
three hydrodynamic and/or biogeochemical variables: (i) the bottom mg.m− 3 for log_chl, 0.5 mL.L− 1 for o2 and 0.5 ◦ C for thetao. The spatial
seawater temperature (thetao), (ii) the bottom dissolved dioxygen con­ extent of bioclimatic models’ calibration was restricted to the coastal
centration (o2), and (iii) the log-transformed surface chlorophyll-a level waters of the European seas (i.e., Baltic Sea, North Sea, English Channel,
(log_chl) as a proxy for resource availability. Since the species studied are Celtic Sea, Irish Sea, Bay of Biscay, Cantabrian Sea, Portuguese Sea, and
all euryhaline (see for instance Able et al., 2022; Whitfield et al., 2022 Mediterranean Sea) and using the application of a precautionary
for a recent update on ecology of MEO fish species), we decided not to threshold (less than 50 km from the coast or less than 300 m deep)
consider salinity as a predictive variable. The avoidance of multi­ (Table A.2).
collinearity issues was checked (i.e., thresholds of Pearson correlation The habitat component of the potential niche was approximated in
coefficients r < 0.7 – Green (1979), and Variance Inflation Factor esti­ the same way but with a restricted surface range envelope (i.e., recti­
mates <5 – Dormann et al. (2013)). Log_chl, o2, and thetao were retrieved linear hypervolume excluding the 2.5% and 97.5% percentiles of each
from bias-adjusted, statistically downscaled, high-resolution climate PCA axis retained) from the spatialisation of scores of the seven main
ensemble projections (Kristiansen et al., 2024). The ensemble pro­ PCA axes retained (Ben Rais Lasram et al., 2020). The spatial extent of
jections were derived by averaging a set of eight individually down­ habitat models’ calibration was restricted to species-specific areas with
scaled CMIP6 models including: (i) IPSL-CM6A-LR (Institut occurrence data available across all the coastal waters of the Northeast
Pierre-Simon Laplace), (ii) MPI-ESM1-2-LR (Max Planck Institute for Atlantic (i.e., the North Sea, English Channel, Celtic Sea, Irish Sea, Bay of
Meteorology), (iii) CanESM5-CanOE (Canadian Centre for Climate Biscay, Cantabrian Sea, and Portuguese Sea) and using the same pre­
Modelling and Analysis), (iv) UKESM1-0-LL (Met Office Hadley Centre), cautionary threshold used in bioclimatic modelling.
(v) GFDL-ESM4 (NOAA/OAR Geophysical Fluid Dynamics Laboratory), As regression techniques based on presence and absence data have
(vi) CMCC-ESM2 (Fondazione Centro Euro-Mediterraneao sui Cambia­ been showed to work better than presence-only techniques, stratified
menti Climatici), (vii) CMCC-CM2-SR5 (Fondazione Centro pseudo-absence data (i.e., background data) were generated (Brotons
Euro-Mediterraneo sui Cambiamenti Climatici), and (viii) MIROC-ES2L et al., 2004). These were randomly selected within the bioclimatic and
(Research Institute for Global Change). Monthly values annually aver­ habitat space outside the restricted convex hull and restricted surface
aged were used for both historical (1993–2014) and future (2015–2099) range envelope, respectively (Barbet-Massin et al., 2012). This selection
periods under different possible emission scenarios at 1/12th degree followed the “D-designs” theory (Montgomery, 2017): as many
(~6–7 kms) grid resolution for the desired spatial extent (n = 53,431 pseudo-absence data were generated as filtered presence data for each
0.08◦ × 0.08◦ cells – Kristiansen et al., 2024) (Fig. A.4). Bioclimatic species except if too little presence data (i.e., 1000 pseudo-absences
models were monthly predicted for the present (2001–2020) and future generated if < 1000 presence data) in order to avoid biases due to un­
(2080–2099) periods. Two of the last IPCC future emission scenarios balanced prevalence and/or low sample size (Guisan et al., 2017).
(IPCC et al., 2021) were used here: the SSP2-4.5 ‘intermediate’ scenario Because the recommendation to have more than 20 data was reached, all
and the SSP5-8.5 ‘pessimistic’ scenario. six MEO fish species could be studied although there were much fewer
The habitat component of the potential niche was modelled from (i) presence data for the two ‘sub-tropical’ species (Guisan et al., 2017 –
the bathymetry, and (ii) the seabed type. These two variables were Table A.2).
considered representative of use conditions of the bottom habitat sup­
posedly time-invariant for the six MEO fish species. The bathymetry was

3
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

2.2.2. Validation and outputs of final suitability ensemble forecasting representing displacements of southern margins of ‘sub-boreal’ species
As recommended in the literature, ensemble forecasting was used or northern margins of ‘sub-tropical’ species (Ordonez and Williams,
instead of strict selection of a single statistical technique was used for 2013). They were estimated as the relative proportion of the linear
each of the species-specific bioclimatic and habitat suitability distribu­ distance separating the range margins for the present period and that of
tion models (Marmion et al., 2009; Hao et al., 2019). Ensemble fore­ the future period (Perry et al., 2005) given the height of the suitable
casting were built with a slightly adapted code from the R package range (see Table A.8 for definition of margins, and height).
biomod2 (Thuiller et al., 2009) from nine statistical techniques It was considered that: (i) positive and negative Margins shift esti­
belonging to four statistical approaches: (i) regression-based approaches mates corresponded to a northward and southward shift, respectively;
(Generalized Linear Models – GLM; Generalized Additive Models – GAM; (ii) positively correlated SRC and Margins shift estimates for ‘sub-trop­
Multivariate Adaptive Regression Splines – MARS), (ii) ical’ species accompanied by a northward shift of northern margins
machine-learning approaches (Recursive Partitioning – CTA; Flexible would describe a gain in suitable areas at higher latitudes; and (iii)
Discriminant Analysis – FDA; Artificial Neural Networks – ANN), (iii) negatively correlated SRC and Margins shift estimates for ‘sub-boreal’
boosting and basting approaches (Random Forest – RF, Boosted species accompanied by a northward shift of southern margins would
Regression Trees – GBM), and (iv) Maximum Entropy formalism (Max­ correspond to a loss of suitable areas at lower latitudes (Soultan et al.,
Ent). Default settings were used. The predictive accuracy of models was 2022). Rates of margins displacement were calculated by dividing
tested using the commonly three-fold cross-validation. In this approach, displacement shift distances separating the range margins between
the dataset was split into three partitions (here, with a prevalence of future and present periods by the number of decades (expressed in km.
0.5). Each of the three runs was performed on two of the three partitions decade− 1).
(i.e., training data), and evaluated on the third remaining partition (i.e.,
evaluation data), each partition used only once for test, for each statis­ 2.4.2. From probabilistic environment suitability maps
tical technique (Breiman, 2017). Distribution shifts were usually referred to north-south displace­
The prevalence-independent metric, i.e., True Skill Statistic (TSS – ments but had also an east-west component (Hiddink et al., 2015). As
Allouche et al., 2006), was used as mean predictive accuracy metric for such, an alternative was to study shifts in resulting probabilistic envi­
each of the three runs of the nine statistical techniques. Only the sta­ ronment suitability predictions using three range centroids: (i) the
tistical techniques considered having a high predictive power, i.e., centres of gravity of the distributions (COGDs), (ii) the centres of leading
three-run TSS average > 0.7, were selected in ensemble forecasting. edge expansions (COGEs), and (iii) the centres of trailing edge con­
Consensual predictions were obtained by weighting the sum of proba­ tractions (COGCs) (Thorson et al., 2016; Pinsky et al., 2020; Soultan
bility values of each run of all the selected statistical techniques by the et al., 2022). COGD is the most commonly used metric to assess shifts in
predictive accuracy of its prediction (Marmion et al., 2009). The relative the spatial core of species predicted distributions (Soultan et al., 2022).
importance and response curves of these variables and estimates of the COGEs and COGCs have been much less studied than COGDs. They
uncertainty between the different cross-validation runs retained were represent part of the distribution ranges where bioclimatic conditions
provided (Guisan et al., 2017). Final projections were obtained by taking necessary to the species settlement and survival will be respectively
the arithmetic mean of the 240 (i.e., 20 years × 12 months) model gained (i.e., successful colonisations) and lost (i.e., local extirpations)
output projections. The thresholding method optimizing both sensitivity between future and present periods (Hiddink et al., 2015).
(i.e., percentage of presences correctly predicted) and specificity (i.e., Four indicators of distribution shifts were computed from the
percentage of absences correctly predicted) was chosen to transform COGDs, COGEs, and COGCs. First, the actual direction of distribution
probabilistic estimates into binary bioclimatic or habitat suitability shifts was given by the direction for the expansion (i.e., from COGDs to
prediction maps classifying areas as suitable or unsuitable (Guisan et al., COGEs) and the opposite direction for the contraction (i.e., from COGCs
2017 – Table A.6). to COGDs) (Hiddink et al., 2015) using geographical coordinates. Sec­
ond, the absolute (expressed in km) and relative (expressed in %, rela­
2.3. Environment suitability distribution combining both the bioclimatic tively to width/height of the suitable range) displacement shift distances
and habitat components of the potential species niche along the longitudinal/latitudinal axis were calculated using the COGD
for the present distribution as origin (see Table A.8 for definition of
The environment suitability distribution was obtained by combining width and height). Third, spatial variables such as bathymetry and
the bioclimatic and habitat suitability distribution models in the form of distance to the nearest coast were estimated to highlight possible
probabilistic forecasting as well as in the form of binary maps (n = 0.01◦ deepening and seawardness in distribution shifts, respectively. Finally,
× 0.01◦ 1,644,121 cells – Hattab et al., 2014; Fournier et al., 2017). To bioclimatic variables were also estimated to better understand the po­
this end, the bioclimatic suitability distribution grids were downscaled tential direct (i.e., thetao) and indirect (i.e., log_chl and o2) effects of
by bilinear resampling to the same spatial resolution and cropped to the climate change in the directionality and the magnitude of distribution
same spatial extent as habitat suitability distribution grids. Finally, shifts.
environment suitability distribution arose from the multiplication of The geographic coordinates of the COGDs were here calculated as
resampled and cropped bioclimatic and habitat suitability grids. the average values over all study area cells weighted by the values of
probabilistic environment suitability predictions (Thorson et al., 2016).
2.4. Indicators of environment suitability distribution shifts This method was shown to be more robust than the conventional
methods (e.g., the abundance-weighted average – AWA) that were
2.4.1. From binary environment suitability maps limited due to biases introduced in preferential sampling distribution or
Two indicators of distribution shifts resulting from binary environ­ its changes over time (Pinsky et al., 2013; Hiddink et al., 2015). The
ment suitability maps comparisons between the present (2001–2020) calculation of the coordinates of COGEs and COGCs, usually calculated
and future (2080–2099) periods were estimated under each of the two from binary maps as in Hiddink et al. (2015), were thus readapted in the
future emission scenarios for each species studied. First, the Species same way as COGDs but from areas where the absolute variations in
Range Change (SRC expressed in %) representing change in size of probabilistic environment suitability predictions between the future and
suitable range was calculated as the difference between the relative present periods were positive and negative, respectively. Rates of
proportion of range expansion and that of range contraction given the COGDs displacement were calculated by dividing displacement shift
size of the present suitable range using the R package biomod2 (Thuiller distances separating the COGDs between future and present periods by
et al., 2009; Soultan et al., 2022). Second, a relative latitudinal shift of the number of decades (expressed in km.decade− 1) or the temperature
range margins (Margins shift expressed in %) was computed, difference them separating (expressed in km.◦ C− 1). Unlike COGDs,

4
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

displacement shift distances for the leading (i.e., COGEs) and trailing (i. 3. Results
e., COGCs) edges were not appropriate to be interpreted as actual rates
but rather as index of the extent of distribution shifts (Hiddink et al., 3.1. Changes in bioclimatic variables under possible emission scenarios
2015).
There were significant differences in bioclimatic variables’ mean
values between the present and future periods at the scale of the coastal
waters of the Northeast Atlantic: log_chl decreased by 33% and 38%, o2

Fig. 1. Mapping of bioclimatic variables monthly values averaged (arithmetic mean of the 240 = 20 years × 12 months projections) over the whole study area: log-
transformed surface chlorophyll-a (log_chl expressed in mg.m− 3 – first row), bottom dissolved oxygen (o2 expressed in mL.L− 1 – second row), and bottom temperature
(thetao expressed in ◦ C – third row) for the present (2001–2020 – first column); and their absolute variations to the future period (2080–2099) under the SSP2-4.5
(intermediate – second column) and SSP5-8.5 (pessimistic – third column) scenarios (see Fig. A2 and Table A.1 for more details). The main areas mentioned in the
text: Bay of Biscay – BoB, Celtic Sea – CS, Irish Sea – IS, English Channel – EC, North Sea – NS, Wadden Sea – WS, Jutland – J, Skagerrak – S, Kattegat – K, and Belt Sea
– BS.

5
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

decreased by 4% and 6%, and thetao increased by 12% and 19% under 0.23 ± 0.04 and 0.63 ± 0.12 depending on variables and species.
the SSP2-4.5 and SSP5-8.5 scenarios, respectively (Fig. A.2; Table A.1). However, the variable with the greatest contribution tended to differ
Almost the whole study area and, particularly the northern Irish Sea, according to the species affinity: log_chl (except for sea bass) and o2 for
was affected by a decrease in log_chl levels, except a few well-defined the four ‘sub-boreal’ species; o2 and thetao for the two ‘sub-tropical’
areas where levels always remained constant (i.e., the Wadden Sea, species (Table A.4).
eastern British coasts, southern North Sea, coasts around the English For final habitat suitability ensemble forecasting, all the statistical
Channel and the Irish Sea, Bay of Biscay coastal waters, and Portuguese techniques except GAM and MARS for some species were retained, and
coasts especially near the Mondego catchment) or increased (i.e., the TSS averages were comprised between 0.70 and 0.94 (Table A.5). Un­
southernmost Portuguese and Spanish coasts). The whole study area was certainties between techniques were also very low with standard devi­
affected by a decrease in o2 (i.e., particularly in the northern British ation averages comprised between 0.13 ± 0.08 and 0.15 ± 0.13
coasts, Irish Sea, Celtic Sea, western English Channel, and the south­ (Fig. A.4, A.5, A.6, A.7, A.8 and A.9).
ernmost Portuguese and Spanish coasts) and an increase in thetao (i.e.,
particularly in the Belt Sea, Kattegat, Skagerrak, west coast of Jutland, 3.3. Present (2001–2020) predicted MEO fish species distributions
Wadden Sea, North Sea, English Channel, Irish Sea, and Celtic Sea)
levels (Fig. 1). Three groups of species have similar present predicted distributions
(when using probabilities >0.7): (i) sea bass and flounder (i.e., British
3.2. Validation of final bioclimatic and habitat suitability ensemble coasts especially those around the Irish Sea, west coast of Jutland,
forecasting Wadden Sea, southern North Sea, English Channel, Bay of Biscay coastal
waters, Portuguese coasts, and southern Spanish coasts, but also the
For final bioclimatic suitability ensemble forecasting, almost all the Kattegat and Belt Sea for flounder); (ii) plaice and common sole (i.e.,
statistical techniques except the GLM were retained, and the three-run almost everywhere except the south-western Celtic Sea, and the Nor­
averages of TSS varied among species between 0.71 and 0.99 wegian, Cantabrian, Galician, Portuguese and Spanish trenches); and
(Table A.3). Uncertainties between techniques (i.e., inter-technique (iii) the two ‘sub-tropical’ species (i.e., the southern Bay of Biscay
standard deviation) were generally low over the whole study area coastal waters, south-western and southern Portuguese coasts, and
since standard-deviations varied among species between 0.12 ± 0.07 southern Spanish coasts) (Figs. 2–7).
and 0.17 ± 0.06 on average (Fig. A.4, A.5, A.6, A.7, A.8 and A.9). The
contribution estimates of the three bioclimatic variables varied between

Fig. 2. Sea bass, Dicentrarchus labrax environment suitability geographic projections resulting from the combination of bioclimatic and habitat suitability predictions
weighted by a chosen predictive accuracy metric (e.g., TSS scores of statistical techniques retained in respective final ensemble forecasting): probabilistic envi­
ronment suitability predictions for the present period (2001–2020 – a); their absolute variations to the future period (2080–2099) under the SSP2-4.5 (intermediate –
b) and SSP5-8.5 (pessimistic – c) scenarios; and the resulting species range change under the SSP2-4.5 (d) and SSP5-8.5 (e) scenarios after binarisation.

6
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Fig. 3. Flounder, Platichthys flesus environment suitability geographic projections resulting from the combination of bioclimatic and habitat suitability predictions
weighted by a chosen predictive accuracy metric (e.g., TSS scores of statistical techniques retained in respective final ensemble forecasting): probabilistic envi­
ronment suitability predictions for the present period (2001–2020 – a); their absolute variations to the future period (2080–2099) under the SSP2-4.5 (intermediate –
b) and SSP5-8.5 (pessimistic – c) scenarios; and the resulting species range change under the SSP2-4.5 (d) and SSP5-8.5 (e) scenarios after binarisation.

3.4. Future (2080–2099) predicted MEO fish species distributions under north-eastern Irish Sea, but also the south-western Irish coasts and
the two emission scenarios southern English coasts for sea bass. But they decreased in the northern
Irish Sea, Western English Channel, Bay of Biscay continental shelf, and
SRC and northward Margins shift were positively correlated for the Portuguese and Spanish coasts, but also in the Celtic Sea and central
two ‘sub-tropical’ species under both possible emission scenarios North Sea for common sole. Second, flounder and plaice exhibited a
reflecting an increase in the number of suitable areas at northern global decrease in all areas especially in the northern Irish Sea, Celtic
boundaries (Figs. 6 and 7 and A.10). Conversely, SRC and northward Sea, Western English Channel, Bay of Biscay continental shelf, and the
Margins shift were negatively correlated for two ‘sub-boreal’ species (i.e., southernmost Portuguese and Spanish coasts, but also the southern
flounder and plaice) highlighting a decrease in the number of suitable Wadden Sea for flounder. And finally, the two ‘sub-tropical’ species
areas at southern boundaries (Figs. 3 and 4, and A.10). However, ‘sub- increased almost everywhere especially in the Kattegat, west coast of
tropical’ species, and more specifically meagre, showed significant SRC Jutland, Wadden Sea, southern North Sea, eastern English Channel, the
expansion and northward Margins shift under the SSP2-4.5 scenario and coasts around the English Channel, British coasts especially those
even more pronounced under the SSP5-8.5 scenario (Fig. A.10; around the Irish Sea, and Bay of Biscay coastal waters. However, they
Tables A.7, A.8 and A.9). Moreover, margins displacement were faster decreased highly localized in the ‘Basque’ Bay of Biscay coastal waters,
for ‘sub-tropical’ species (~79–97 km decade− 1) than for ‘sub-boreal’ but also south-western Portuguese coasts for meagre or the southern­
species (~0–3 km.decade− 1) (Table A.9). most Portuguese and Spanish coasts for Senegalese sole (Figs. 2–7).
There were significant increases in probabilistic environment suit­ A global and visible north-westward shift of increases in probabilistic
ability predictions’ mean values from the present to future periods and environment suitability predictions was highlighted under the SSP2-4.5
even more pronounced under the SSP5-8.5 scenario for ‘sub-tropical’ scenario and even more pronounced under the SSP5-8.5 scenario for the
species. Conversely, there were significant decreases in probabilistic six MEO fish species (Fig. 8). This shift, mostly northward (69–75%)
environment suitability predictions’ mean values for ‘sub-boreal’ spe­ rather than westward (25–28%), was towards shallower areas closer to
cies (except for sea bass) (Fig. A.11; Table A.10). We identified three the coasts for ‘sub-boreal’ species (except for sea bass) and towards areas
groups of species with similar distribution of absolute variations of their further from the coasts for ‘sub-tropical’ species (Fig. A.13; Table 1).
future environment suitability probabilities under both possible emis­ COGDs tended to shift towards areas with lower log_chl and o2 levels and
sion scenarios. First, sea bass and common sole exhibited similar trends. higher thetao level under the SSP2-4.5 scenario and even more pro­
They especially increased on the eastern British coasts, Kattegat, west nounced the SSP5-8.5 scenario for the six MEO fish species (Fig. A.13).
coast of Jutland, Wadden Sea, southern North Sea, and coasts around the However, displacement shifts of COGDs were greater and even more

7
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Fig. 4. Plaice, Pleuronectes platessa environment suitability geographic projections resulting from the combination of bioclimatic and habitat suitability predictions
weighted by a chosen predictive accuracy metric (e.g., TSS scores of statistical techniques retained in respective final ensemble forecasting): probabilistic envi­
ronment suitability predictions for the present period (2001–2020 – a); their absolute variations to the future period (2080–2099) under the SSP2-4.5 (intermediate –
b) and SSP5-8.5 (pessimistic – c) scenarios; and the resulting species range change under the SSP2-4.5 (d) and SSP5-8.5 (e) scenarios after binarisation.

pronounced under the SSP5-8.5 scenario for the two ‘sub-tropical’ greater northward range expansion was also observed in our results
(73–81 km/~10 km.decade− 1/~43 km◦ C− 1 to north or 3–4% relative especially for ‘sub-tropical’ species contrary to ‘sub-boreal’ species.
northward shift; and 45–50 km/~6 km.decade− 1/~27 km◦ C− 1 to west Decreases in chlorophyll-a and dissolved dioxygen levels seemed to be
or 4–6% relative westward shift) than for the four ‘sub-boreal’ (8–24 the main drivers in the loss of environment suitability especially for
km/~1–3 km.decade− 1/~4–12 km◦ C− 1 to north or 0–1% relative ‘sub-boreal’ species while increase in temperature led the gain of envi­
northward shift; and 5–14 km/~1–2 km.decade− 1/~2–7 km◦ C− 1 to ronment suitability for ‘sub-tropical’ species. Finally, all of these
west or 0–1% relative westward shift) species (Fig. A.12; Table 1). modelling conclusions were globally and unsurprisingly even greater
Displacement shifts were nevertheless much greater for COGCs under the SSP5-8.5 scenario than the SSP2-4.5 scenario at the horizon of
(particularly much more importantly for the two ‘sub-tropical’ species) the 21st century end.
and COGEs than for COGDs under the SSP2-4.5 scenario and even more
pronounced under the SSP5-8.5 scenario for the six MEO fish species
4.1. Relevance and accuracy of the data, method and models
(Fig. A.12; Table 1). ‘Sub-boreal’ species tented to contract from areas
with lower log_chl and o2 levels and ‘sub-tropical’ species from areas
Application of ensemble forecasting to build consensual environ­
with lower o2 level and higher thetao level under the SSP2-4.5 scenario
ment suitability predictions is now recognized to provide the most
and even more pronounced under the SSP5-8.5 scenario (Fig. A.13).
reliable and robust correlative models’ outputs (Hao et al., 2019) while
the ‘hierarchical filters’ concept aimed to both improve models’ pre­
4. Discussion
dictive accuracy and limit sampling bias (Hattab et al., 2014; Fournier
et al., 2017; Ben Rais Lasram et al., 2020; Schickele et al., 2020). In
This study re-adapted the ‘hierarchical filters’ method employed in
addition, two main aspects of the original Ben Rais Lasram et al. (2020)
Hattab et al. (2014), Fournier et al. (2017), Ben Rais Lasram et al.
framework were changed. First, finer temporal resolution (i.e., a
(2020), and Schickele et al. (2020) in focusing on a relevant and
monthly scale instead of a decadal-scale) and larger temporal extent (i.
consistent match between the particular case of MEO species ecology
e., over twenty years instead of one decade) were chosen for the present
and distribution and the time-space frame of climate. Anthropogenic
and future projections of species distribution. These different choices
climate change was predicted to induce a global and visible
allowed to capture a more robust estimation of the trend in the impacts
north-westward, more to the north than to the west, distribution shift for
of climate change on the spatial distribution and obtain more accurate
six of the main bentho-demersal MEO fish species contributing to coastal
final projections (Castillo-Escrivà et al., 2020). Second, the spatial extent
fisheries and seafood production within the Northeast Atlantic. A
of bioclimatic models’ calibration was not on a global scale but limited

8
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Fig. 5. Common sole, Solea solea environment suitability geographic projections resulting from the combination of bioclimatic and habitat suitability predictions
weighted by a chosen predictive accuracy metric (e.g., TSS scores of statistical techniques retained in respective final ensemble forecasting): probabilistic envi­
ronment suitability predictions for the present period (2001–2020 – a); their absolute variations to the future period (2080–2099) under the SSP2-4.5 (intermediate –
b) and SSP5-8.5 (pessimistic – c) scenarios; and the resulting species range change under the SSP2-4.5 (d) and SSP5-8.5 (e) scenarios after binarisation.

to the coastal waters of the European seas and using the application of a ecology. On the one hand, these predictions were obtained from the
precautionary threshold. The spatial extent of habitat models’ calibra­ combination of the bioclimatic and habitat components of the potential
tion was species-specific: it was delineated to areas with occurrence data species niche that were judged to be accurate during models’ evaluation
available across all the coastal waters of the Northeast Atlantic and using process, and finally resulted from consensual predictions by at least
the same precautionary threshold used in bioclimatic modelling. These seven statistical techniques over the nine proposed by Biomod2
choices allowed to take into account biogeographic history and/or (Tables A.3 and A.5). On the other hand, the resulting response curves of
dispersal ability of the six MEO fish species or avoid over-predictions of each bioclimatic variable were carefully controlled and considered
the species range due to presence and pseudo-absence data conflicts plausible since smoothed unimodal physiological responses reflecting
(Hanberry et al., 2012). Nevertheless, some limitations regarding the more realistic ellipse-like niche visualisations were the most frequently
quantity and quality of occurrence data should be pointed out. The data encountered in natural systems (Fig. A.3 – Guisan et al., 2017). More­
used in our analysis are coming from international voluntary databases, over, the optimal temperature ranges obtained from these response
although these data are largely derived from scientific monitoring. curves were in the similar range to those found in the literature (e.g.,
These data are limited for the two ‘sub-tropical’ species (e.g., in the Baensch and Riehl, 1997; Quéro and Vayne, 1997). Finally, present
database that was used in our analyses only 113 and 63 observations distributions of environmentally highly suitable areas were consistent
were available for meagre and for the Senegalese sole, respectively – with knowledge on the species essential habitats in the present situation.
Table A.2), particularly in the south of the study area for some ‘sub-­ Indeed, for instance, the main identified spawning and/or feeding
boreal’ species. In addition to the limited data availability, there may be grounds in the Northeast Atlantic coastal zone part are included in the
taxonomic confusion between some species (e.g., sea bass and spotted present predicted distribution of highly suitable areas (Muus et al.,
bass, or common sole and Senegalese sole – Muus et al., 1988; Quéro and 1988; Quéro and Vayne, 1997). All these suggest that we can be
Vayne, 1997). These potential problems highlight the need to improve reasonably confident in the models’ transferability and thus predictions
species identification skills and increase participatory reporting in all for the end of the 21st century (Randin et al., 2006).
sampled and/or exploited areas, which is one of the main challenges for
sustainable biomonitoring scientists in the 21st century (Ferreira et al.,
4.2. Distribution shift directionality and magnitude
2021).
At the same time, this study provided interpretable present distri­
One of the main trends highlighted by our results was the north-
butions of environment suitability for species settlement and survival as
westward shift in distribution for the six MEO fish species studied in
many methodological choices were made in agreement with the species
the Northeast Atlantic. Though not focusing on the same species,

9
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Fig. 6. Meagre, Argyrosomus regius environment suitability geographic projections resulting from the combination of bioclimatic and habitat suitability predictions
weighted by a chosen predictive accuracy metric (e.g., TSS scores of statistical techniques retained in respective final ensemble forecasting): probabilistic envi­
ronment suitability predictions for the present period (2001–2020 – a); their absolute variations to the future period (2080–2099) under the SSP2-4.5 (intermediate –
b) and SSP5-8.5 (pessimistic – c) scenarios; and the resulting species range change under the SSP2-4.5 (d) and SSP5-8.5 (e) scenarios after binarisation.

ensemble climate projections, current and future periods (observed or distribution of other Northeast Atlantic ‘sub-boreal’ (e.g., mackerel,
predicted), modelling methodologies and/or spatial extent used in this COGD shift of 16 km.decade− 1 or 25 km◦ C− 1 – Bruge et al., 2016), and
study, several previous studies also highlighted a similar pattern for ‘sub-tropical’ (e.g., albacore Thunnus alalunga, northward margins shifts
other fish species in the Northeast Atlantic. For cod Gadus morhua, the of 74 km.decade− 1 – Chust et al., 2019) fish species. More fluctuating
marine environment suitability would decrease in the southern area of dynamics at the range margins may explain the much faster shift rates
the North Sea but increased in the remaining North Sea (Núñez-Riboni than at the range core, with intermittent suitable conditions fostering
et al., 2019). For mackerel Scomber scombrus, the centre of gravity in the rapid colonisation events (Hastings et al., 2018). Moreover, the fact that
summer feeding area would have shifted from the Norwegian Sea core to ‘sub-tropical’ species are probably still not in equilibrium with their
Svalbard in the north and west to Greenland during the last decade environment and are still adjusting with the environmental conditions in
(Boyd et al., 2020). This northward shift pattern was also observed in present days may explain the observed patterns (Guisan et al., 2017). An
other taxa of the Northeast Atlantic from zooplankton (e.g., copepods observation that has already highlighted in the Bay of Biscay where the
species – Chust et al., 2014b) to top predators (e.g., Balearic shearwater abundances of cold-water flatfish species decreased, and that of
Puffinus mauretanicus – Wynn et al., 2007), including also marine plants warm-water ones increased (Hermant et al., 2010). Thus, globally,
(e.g., temperate seaweed species – Neiva et al., 2016). This pattern has transient dynamics are characterised by rapid colonisation of
been shown also in other northern temperate regions (e.g., marine warm-water species and slower extirpation of cold-water species in
species on the U.S. Northeast continental shelf – Fredston et al., 2021). response to climate change (Pinsky et al., 2020). This can be related with
This study put forward a greater northward expansion for ‘sub- the potential difference in the influence of climate change components.
tropical’ species than for ‘sub-boreal’ species in relation with a faster Indeed, indirect effects (i.e., decreases in chlorophyll-a and dissolved
northward shift of COGDs (i.e., ~10 km.decade− 1 or 43 km◦ C− 1 versus dioxygen levels) seemed to influence the loss of environment suitability
~ 1–3 km.decade− 1 or 4–12 km◦ C− 1) and faster northward margins especially for ‘sub-boreal’ species, while direct effects (i.e., increases in
shifts (i.e., ~79–97 km.decade− 1 versus 0–3 km.decade− 1). However, temperature) were more related to gains of environment suitability in
interpretations concerning the differences in northward expansion rates particular for ‘sub-tropical’ species.
between ‘sub-boreal’ and ‘sub-tropical’ species should be taken with a
grain of salt since the entire distribution area of species and in particular 4.3. Comparison of climate change impact between regions or across the
that of the two ‘sub-tropical’ species were not fully captured in the studied area
models’ calibration. Nevertheless, these differences were also observed
in other cases, and for instance regarding the predicted future spawning The main areas where the environment suitability decreased were

10
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Fig. 7. Senegalese sole, Solea senegalensis environment suitability geographic projections resulting from the combination of bioclimatic and habitat suitability
predictions weighted by a chosen predictive accuracy metric (e.g., TSS scores of statistical techniques retained in respective final ensemble forecasting): probabilistic
environment suitability predictions for the present period (2001–2020 – a); their absolute variations to the future period (2080–2099) under the SSP2-4.5 (inter­
mediate – b) and SSP5-8.5 (pessimistic – c) scenarios; and the resulting species range change under the SSP2-4.5 (d) and SSP5-8.5 (e) scenarios after binarisation.

the central and northern Irish Sea, Celtic Sea, Western English Channel, towards shallower areas closer to the coast for flounder, plaice and
and Bay of Biscay continental shelf for ‘sub-boreal’ species, but also the common sole or areas more far from the coast for seabass and ‘sub-­
southern Wadden Sea for flounder or along the southernmost Portu­ tropical’ species. However, this difference in future use of the seabed
guese and Spanish coasts for the six MEO fish species. This should be put between ‘sub-boreal’ and ‘sub-tropical’ species may probably be biased
in relation with low present levels of chlorophyll-a and dissolved oxygen and due to the topology of the continental shelf in the Bay of Biscay.
levels becoming limiting in the future (Fig. 1 and A.3). This interpre­ Indeed, as sub-tropical species move northward, the continental shelf
tation is consistent with the negative trophic amplification, i.e., the becomes larger and their distribution can expanse further from the
decrease in primary production and zooplankton biomass in response to coast.
increase in ocean stratification that has been demonstrated in these Finally, the existence of three groups of species in our results with
areas (Chust et al., 2014a). However, this negative trophic amplification similar future distribution of suitable thermic habitats may be explained
could likely be worse than expected if the Atlantic Meridian Overturning by finer thermal affinity differences. Flounder and plaice are the most
Circulation (AMOC) weakens over the 21st century (Collins et al., 2019). cold-water ‘sub-boreal’ species and will not gain suitable areas due to
The contraction effect of chlorophyll-a and dissolved oxygen in these their lower tolerance to warmer temperatures (>20 ◦ C – Fig. A.3). Sea
areas but also within the central North Sea would lead to an increase in bass and common sole would gain suitable habitats in some places but to
potential species occupancy in areas where chlorophyll-a and dissolved a lesser extent than the two warm-water species (i.e., meagre and Sen­
oxygen levels would remain unchanged over time and still be sufficient egalese sole). This pattern of lower gains in suitable habitats is also
for species settlement and survival (Fig. A.3). These areas would include highlighted for other highly cold-water (e.g., Acadian redfish Sebastes
the eastern British coasts, Kattegat, west of Jutland, Wadden Sea, fasciatus, American plaice Hippoglossoides platessoides, Atlantic cod
southern North Sea, and coasts around the north-eastern Irish Sea for sea Gadus morhua, haddock Melanogrammus aeglefinus, and thorny skate
bass, common sole and the two ‘sub-tropical’ species; but also the Amblyraja radiata), cool-water (e.g., spiny dogfish Squalus acanthias, and
remaining North Sea for seabass or the remaining coasts around the Irish American lobster Homarus americanus), and warm-water (e.g., striped
Sea and English Channel, and the Bay of Biscay coastal waters for bass Morone saxatilis, summer flounder Paralichthys dentatus, and
‘sub-tropical’ species. In addition, the northward range expansion being Atlantic croaker Micropogonias undulatus) species on the U.S. Northeast
greater for ‘sub-tropical’ than for ‘sub-boreal’ species was mainly continental shelf (Kleisner et al., 2017). A relevant addition to this
explained by the fact that areas were currently calculated as unsuitable present work would be to expand the spatial extent of projected future
but would become suitable largely due to increasing temperatures distributions beyond the North Sea (e.g., the Norwegian Sea, Baltic Sea,
(Fig. A.3). All these elements combined could explain distribution shifts Barents Sea, White Sea, or even the Irminger Sea, Denmark Strait, and

11
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Fig. 8. Indicators of distribution shifts based on probabilistic environment suitability predictions – the centres of gravity of the distributions (COGDs – filled squares),
the centres of leading edge expansions (COGEs – squares with a triangle inside), and the centres of trailing edge contractions (COGCs – circles with a cross inside) of
geographic coordinates for the present (2001–2020 – blue) and future (2080–2099) periods under the SSP2-4.5 (intermediate – orange) and SSP5-8.5 (pessimistic –
dark) scenarios. The specific case of Platichthys flesus COGE under the SSP5-8.5 scenario was due to a too low number of cells very localised within the Cantabrian
trench where the absolute variations in probabilistic environment suitability predictions between the future and present periods were positive (see Fig. 3c). (For
interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Greenland Sea) where a positive trophic amplification would be reached relevant to use joint species distribution models such as the Hierarchical
in these areas in response to ocean stratification (Chust et al., 2014a). Modelling of Species Communities, in order to include the influence of
biotic and random processes, species-specific life-history traits and
inter-species phylogenetic relationships in addition to the environ­
4.4. Future research mental influence (Tikhonov et al., 2017).
Regarding the specific case of MEO, next step would be to consider
To go further and notably to better understand transient dynamics connectivity between the marine habitats of adults and the estuarine
and fine-scale organisms-environment relationships, future research habitat of juveniles (Gillanders et al., 2003; Whitfield et al., 2023).
should take into account a number of key-processes that are not included Hence, a relevant perspective could be to consider several functioning
in this study such as population growth and dispersal dynamics, as well habitat types through ontogenetic life-history stages of MEO fish species
as stochasticity in demographic processes (Singer et al., 2018). In (i.e., growth, feeding, spawning grounds, and migration corridors –
addition, to better discriminate potential and realized niches, it could be

12
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

Table 1
Indicators of distribution shifts based on probabilistic environment suitability predictions – the centres of gravity of the distributions (COGDs), the centres of leading
edge expansions (COGEs), and the centres of trailing edge contractions (COGCs) of the actual displacement shift distance (north-westward displacement shift expressed
in km) and components of this distance along the longitudinal (westward component displacement shift expressed in km) or the latitudinal (northward component
displacement shift expressed in km) axis for the present (2001–2020) and future (2080–2099) periods under the SSP2-4.5 (intermediate) and SSP5-8.5 (pessimistic)
scenarios. The percentages in brackets represented the relative contribution of the longitudinal or latitudinal displacement shift to the actual displacement shift.
Species Thermal affinity North-westward displacement shift Westward component displacement shift Northward component displacement shift
(km) (km) (km)

SSP2-4.5 SSP5-8.5 SSP2-4.5 SSP5-8.5 SSP2-4.5 SSP5-8.5

COGDs

Dicentrarchus labrax Sub-boreal 20.97 24.67 10.72 (26%) 12.58 (26%) 18.03 (74%) 21.23 (74%)
Platichthys flesus Sub-boreal 10.48 14.91 5.34 (26%) 7.57 (26%) 9.02 (74%) 12.84 (74%)
Pleuronectes platessa Sub-boreal 14.46 27.31 7.27 (25%) 13.70 (25%) 12.50 (75%) 23.64 (75%)
Solea solea Sub-boreal 6.45 9.76 3.27 (26%) 4.92 (25%) 5.56 (74%) 8.43 (75%)
Argyrosomus regius Sub-tropical 43.14 94.74 22.61 (27%) 49.80 (27%) 36.79 (73%) 80.84 (73%)
Solea senegalensis Sub-tropical 34.04 85.38 17.93 (28%) 44.99 (28%) 28.96 (72%) 72.77 (72%)
​ ​ COGEs
Dicentrarchus labrax Sub-boreal 262.40 251.87 135.04 (26%) 129.58 (26%) 226.94 (74%) 217.78 (74%)
Platichthys flesus Sub-boreal 175.06 183.06 90.61 (26%) 92.94 (26%) 150.65 (74%) 156.79 (73%)
Pleuronectes platessa Sub-boreal 98.55 163.33 49.71 (25%) 82.60 (25%) 85.36 (75%) 141.65 (75%)
Solea solea Sub-boreal 212.17 221.05 108.11 (25%) 112.66 (25%) 183.84 (75%) 191.57 (75%)
Argyrosomus regius Sub-tropical 202.27 241.96 106.76 (27%) 127.89 (27%) 172.95 (73%) 207.05 (73%)
Solea senegalensis Sub-tropical 174.03 230.55 92.10 (28%) 122.09 (28%) 148.51 (72%) 197.06 (73%)
​ ​ COGCs
Dicentrarchus labrax Sub-boreal 510.40 554.54 257.28 (25%) 276.98 (25%) 433.98 (72%) 472.68 (73%)
Platichthys flesus Sub-boreal 154.07 171.41 78.16 (26%) 86.70 (26%) 132.12 (74%) 147.08 (74%)
Pleuronectes platessa Sub-boreal 321.03 424.17 160.22 (25%) 210.63 (25%) 275.43 (74%) 363.44 (73%)
Solea solea Sub-boreal 219.78 293.66 110.66 (25%) 146.67 (25%) 188.58 (74%) 252.12 (74%)
Argyrosomus regius Sub-tropical 924.34 1270.65 467.79 (26%) 636.03 (25%) 776.41 (71%) 1062.15 (70%)
Solea senegalensis Sub-tropical 546.43 1686.88 280.18 (26%) 835.98 (25%) 461.59 (71%) 1401.54 (69%)

Levin and Stunz, 2005) as proposed in multi-state distribution models the work reported in this paper.
(Frans et al., 2018). The next step would be to collect a biological
database with estuarine occurrences but also, more importantly, to Acknowledgments
obtain environmental variables and their projections in estuaries. This
would be a great challenge for the European Union, which is investing in Special thanks to A. Charbonnel (INRAE) and A. Schickele (Observ­
this subject of reducing observation gaps in the land-sea continuum atoire Océanologique de Villefranche-sur-Mer) for providing relevant
through satellite imagery. Nonetheless, more process-driven modelling modelling advice; S. Elliott from the common unit Management of
approaches require more data and are time-consuming. Therefore, the Diadromous Fish in their Environment OFB-INRAE-Institut Agro-UPPA
degree of biological realism chosen in modelling approaches should be for providing the MigrenMer dataset; and E. Quinton (INRAE) for
analyzed in the light of the work objectives. facilitating access to computing servers. Thanks to Actea (https://www.
actea.earth) for producing the downscaled climate projections relied on
Funding for this analysis.

This work was supported by the FutureMARES (Climate Change and Appendix A. Supplementary data
Future Marine Ecosystem Services and Biodiversity) project funded by
the European Union’s Horizon 2020 research and innovation pro­ Supplementary data to this article can be found online at https://doi.
gramme under grant agreement No 869300. org/10.1016/j.ecss.2024.109013.

CRediT authorship contribution statement Data availability

Anaïs Janc: Writing – original draft, Visualization, Methodology, Data will be made available on request.
Formal analysis, Conceptualization. Chloé Dambrine: Writing – review
& editing, Validation. Patrick Lambert: Writing – review & editing, References
Supervision, Methodology, Conceptualization. Géraldine Lassalle:
Writing – review & editing, Supervision, Project administration, Meth­ Able, K.W., Simenstad, C.A., Strydom, N.A., Bradley, M., Sheaves, M., 2022. Habitat use
odology, Funding acquisition, Conceptualization. Mario Lepage: and connectivity. In: Fish and Fisheries in Estuaries. John Wiley & Sons, Ltd,
pp. 188–254. https://doi.org/10.1002/9781119705345.ch4.
Writing – review & editing, Validation. Jérémy Lobry: Writing – review
Allouche, O., Tsoar, A., Kadmon, R., 2006. Assessing the accuracy of species distribution
& editing, Validation. Maud Pierre: Writing – review & editing, Vali­ models: prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. 43,
dation. Trond Kristiansen: Writing – review & editing, Validation, 1223–1232. https://doi.org/10.1111/j.1365-2664.2006.01214.x.
Resources. Momme Butenschön: Writing – review & editing, Valida­ Baensch, H.A., Riehl, R., 1997. Baensch Aquaria Atlas.
Barbet-Massin, M., Jiguet, F., Albert, C.H., Thuiller, W., 2012. Selecting pseudo-absences
tion, Resources. Henrique N. Cabral: Writing – review & editing, for species distribution models: how, where and how many?: How to use pseudo-
Validation, Project administration, Funding acquisition. absences in niche modelling? Methods Ecol. Evol. 3, 327–338. https://doi.org/
10.1111/j.2041-210X.2011.00172.x.
Ben Rais Lasram, F., Hattab, T., Nogues, Q., Beaugrand, G., Dauvin, J., Halouani, G., Le
Declaration of competing interest Loc’h, F., Niquil, N., Leroy, B., 2020. An open-source framework to model present
and future marine species distributions at local scale. Ecol. Inform. 59, 101130.
https://doi.org/10.1016/j.ecoinf.2020.101130.
The authors declare that they have no known competing financial Boyd, R.J., Sibly, R., Hyder, K., Walker, N., Thorpe, R., Roy, S., 2020. Simulating the
interests or personal relationships that could have appeared to influence summer feeding distribution of Northeast Atlantic mackerel with a mechanistic

13
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

individual-based model. Prog. Oceanogr. 183, 102299. https://doi.org/10.1016/j. Fredston, A., Pinsky, M., Selden, R.L., Szuwalski, C., Thorson, J.T., Gaines, S.D.,
pocean.2020.102299. Halpern, B.S., 2021. Range edges of North American marine species are tracking
Breiman, L., 2017. Classification and Regression Trees. Routledge, New York. https:// temperature over decades. Glob. Change Biol. 27, 3145–3156.
doi.org/10.1201/9781315139470. Gillanders, B., Able, K., Brown, J., Eggleston, D., Sheridan, P., 2003. Evidence of
Brotons, L., Thuiller, W., Araújo, M.B., Hirzel, A.H., 2004. Presence-absence versus connectivity between juvenile and adult habitats for mobile marine fauna: an
presence-only modelling methods for predicting bird habitat suitability. Ecography important component of nurseries. Mar. Ecol. Prog. Ser. 247, 281–295. https://doi.
27, 437–448. https://doi.org/10.1111/j.0906-7590.2004.03764.x. org/10.3354/meps247281.
Bruge, A., Alvarez, P., Fontán, A., Cotano, U., Chust, G., 2016. Thermal niche tracking Green, R.H., 1979. Sampling Design and Statistical Methods for Environmental
and future distribution of atlantic mackerel spawning in response to ocean warming. Biologists. John Wiley & Sons.
Front. Mar. Sci. 3. Guisan, A., Thuiller, W., Zimmermann, N.E., 2017. Habitat Suitability and Distribution
Cabral, H., Drouineau, H., Teles-Machado, A., Pierre, M., Lepage, M., Lobry, J., Reis- Models: with Applications in R. Cambridge University Press.
Santos, P., Tanner, S.E., 2021. Contrasting impacts of climate change on connectivity Hanberry, B.B., He, H.S., Palik, B.J., 2012. Pseudoabsence generation strategies for
and larval recruitment to estuarine nursery areas. Prog. Oceanogr. 196, 102608. species distribution models. PLoS One 7, e44486. https://doi.org/10.1371/journal.
https://doi.org/10.1016/j.pocean.2021.102608. pone.0044486.
Cabral, H.N., Borja, A., Fonseca, V.F., Harrison, T.D., Teichert, N., Lepage, M., Leal, M.C., Hao, T., Elith, J., Guillera-Arroita, G., Lahoz-Monfort, J.J., 2019. A review of evidence
2022. Fishes and estuarine environmental health. In: Fish and Fisheries in Estuaries. about use and performance of species distribution modelling ensembles like
John Wiley & Sons, Ltd, pp. 332–379. https://doi.org/10.1002/9781119705345. BIOMOD. Divers. Distrib. 25, 839–852. https://doi.org/10.1111/ddi.12892.
ch6. Hastings, A., Abbott, K.C., Cuddington, K., Francis, T., Gellner, G., Lai, Y.-C.,
Castillo-Escrivà, A., Mesquita-Joanes, F., Rueda, J., 2020. Effects of the temporal scale of Morozov, A., Petrovskii, S., Scranton, K., Zeeman, M.L., 2018. Transient phenomena
observation on the analysis of aquatic invertebrate metacommunities. Front. Ecol. in ecology. Science 361, eaat6412.
Evol. 8. Hattab, T., Albouy, C., Ben Rais Lasram, F., Somot, S., Le Loc’h, F., Leprieur, F., 2014.
Cheung, W.W.L., Oyinlola, M.A., 2018. Vulnerability of flatfish and their fisheries to Towards a better understanding of potential impacts of climate change on marine
climate change. J. Sea Res. 140, 1–10. https://doi.org/10.1016/j. species distribution: a multiscale modelling approach. Glob. Ecol. Biogeogr. 23,
seares.2018.06.006. 1417–1429. https://doi.org/10.1111/geb.12217.
Chust, G., Allen, J.I., Bopp, L., Schrum, C., Holt, J., Tsiaras, K., Zavatarelli, M., Hermant, M., Lobry, J., Bonhommeau, S., Poulard, J.-C., Le Pape, O., 2010. Impact of
Chifflet, M., Cannaby, H., Dadou, I., Daewel, U., Wakelin, S.L., Machu, E., warming on abundance and occurrence of flatfish populations in the Bay of Biscay
Pushpadas, D., Butenschon, M., Artioli, Y., Petihakis, G., Smith, C., Garçon, V., (France). J. Sea Res. 64, 45–53. https://doi.org/10.1016/j.seares.2009.07.001.
Goubanova, K., Le Vu, B., Fach, B.A., Salihoglu, B., Clementi, E., Irigoien, X., 2014a. Hiddink, J.G., Burrows, M.T., García Molinos, J., 2015. Temperature tracking by North
Biomass changes and trophic amplification of plankton in a warmer ocean. Glob. Sea benthic invertebrates in response to climate change. Glob. Change Biol. 21,
Change Biol. 20, 2124–2139. https://doi.org/10.1111/gcb.12562. 117–129. https://doi.org/10.1111/gcb.12726.
Chust, G., Castellani, C., Licandro, P., Ibaibarriaga, L., Sagarminaga, Y., Irigoien, X., Hill, M.O., Smith, A.J.E., 1976. Principal component analysis of taxonomic data with
2014b. Are Calanus spp. shifting poleward in the North Atlantic? A habitat multi-state discrete characters. Taxon 25, 249–255. https://doi.org/10.2307/
modelling approach. ICES J. Mar. Sci. 71, 241–253. https://doi.org/10.1093/ 1219449.
icesjms/fst147. Hutchinson, G.E., 1957. Concluding remarks. Population studies : animal ecology and
Chust, G., Goikoetxea, N., Ibaibarriaga, L., Sagarminaga, Y., Arregui, I., Fontán, A., demography. Cold Spring Harb. Symp. Quantative Biol. 22, 415–427.
Irigoien, X., Arrizabalaga, H., 2019. Earlier migration and distribution changes of IPCC, 2021. In: Masson-Delmotte, V., Zhai, P., Pirani, A., Connors, S.L., Péan, C.,
albacore in the Northeast Atlantic. Fish. Oceanogr. 28, 505–516. https://doi.org/ Berger, S., Caud, N., Chen, Y., Goldfarb, L., Gomis, M.I., Huang, M., Leitzell, K.,
10.1111/fog.12427. Lonnoy, E., Matthews, J.B.R., Maycock, T.K., Waterfield, T., Yelekçi, O., Yu, R.,
Colwell, R.K., Rangel, T.F., 2009. Hutchinson’s duality: the once and future niche. Proc. Zhou, B. (Eds.), Climate Change 2021: the Physical Science Basis. Contribution of
Natl. Acad. Sci. 106, 19651–19658. https://doi.org/10.1073/pnas.0901650106. Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on
Collins, M., Sutherland, M., Bouwer, L., Cheong, S.-M., Frölicher, T., Jacot Des Climate Change. Cambridge University Press, Cambridge, United Kingdom and New
Combes, H., Koll Roxy, M., Losada, I., McInnes, K., Ratter, B., Rivera-Arriaga, E., York, NY, USA. https://doi.org/10.1017/9781009157896 (in press).
Susanto, R.D., Swingedouw, D., Tibig, L., 2019. Extremes, Abrupt Changes and Kleisner, K.M., Fogarty, M.J., McGee, S., Hare, J.A., Moret, S., Perretti, C.T., Saba, V.S.,
Managing Risk. In: Pörtner, H.-O., Roberts, D.C., Masson-Delmotte, V., Zhai, P., 2017. Marine species distribution shifts on the U.S. Northeast Continental Shelf
Tignor, M., Poloczanska, E., Mintenbeck, K., Alegría, A., Nicolai, M., Okem, A., under continued ocean warming. Prog. Oceanogr. 153, 24–36. https://doi.org/
Petzold, J., Rama, B., Weyer, N.M. (Eds.), IPCC Special Report on the Ocean and 10.1016/j.pocean.2017.04.001.
Cryosphere in a Changing Climate. Cambridge University Press, Cambridge, UK and Kristiansen, T., Butenschön, M., Peck, M.A., 2024. Statistically downscaled CMIP6 ocean
New York, NY, USA, pp. 589–655. https://doi.org/10.1017/9781009157964.008. variables for European waters. Sci. Rep. 14, 1209. https://doi.org/10.1038/s41598
Cornwell, W.K., Schwilk, D.W., Ackerly, D.D., 2006. A trait-based test for habitat -024-51160-1.
filtering: convex hull volume. Ecology 87, 1465–1471. https://doi.org/10.1890/ Le Marchand, M., Hattab, T., Niquil, N., Albouy, C., Le Loc’h, F., Ben Rais Lasram, F.,
0012-9658(2006)87[1465:ATTFHF]2.0.CO;2. 2020. Climate change in the Bay of Biscay: changes in spatial biodiversity patterns
Costa, B.H. e, Assis, J., Franco, G., Erzini, K., Henriques, M., Gonçalves, E.J., Caselle, J.E., could be driven by the arrivals of southern species. Mar. Ecol. Prog. Ser. 647, 17–31.
2014. Tropicalization of fish assemblages in temperate biogeographic transition https://doi.org/10.3354/meps13401.
zones. Mar. Ecol. Prog. Ser. 504, 241–252. https://doi.org/10.3354/meps10749. Le Pape, Olivier, Chauvet, F., Mahévas, S., Lazure, P., Guérault, D., Désaunay, Y., 2003.
Daufresne, M., Lengfellner, K., Sommer, U., 2009. Global warming benefits the small in Quantitative description of habitat suitability for the juvenile common sole (Solea
aquatic ecosystems. Proc. Natl. Acad. Sci. 106, 12788–12793. https://doi.org/ solea, L.) in the Bay of Biscay (France) and the contribution of different habitats to
10.1073/pnas.0902080106. the adult population. J. Sea Res. 50, 139–149. https://doi.org/10.1016/S1385-1101
Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G., (03)00059-5.
Gruber, B., Lafourcade, B., Leitão, P.J., Münkemüller, T., McClean, C., Osborne, P.E., Levin, P.S., Stunz, G.W., 2005. Habitat triage for exploited fishes: can we identify
Reineking, B., Schröder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2013. essential “Essential Fish Habitat?”. Estuar. Coast Shelf Sci. 64, 70–78. https://doi.
Collinearity: a review of methods to deal with it and a simulation study evaluating org/10.1016/j.ecss.2005.02.007.
their performance. Ecography 36, 27–46. https://doi.org/10.1111/j.1600- Levitus, S., Antonov, J.I., Boyer, T.P., Stephens, C., 2000. Warming of the world ocean.
0587.2012.07348.x. Science 287, 2225–2229. https://doi.org/10.1126/science.287.5461.2225.
Dulvy, N.K., Rogers, S.I., Jennings, S., Stelzenmüller, V., Dye, S.R., Skjoldal, H.R., 2008. Marmion, M., Parviainen, M., Luoto, M., Heikkinen, R.K., Thuiller, W., 2009. Evaluation
Climate change and deepening of the North Sea fish assemblage: a biotic indicator of of consensus methods in predictive species distribution modelling. Divers. Distrib.
warming seas. J. Appl. Ecol. 45, 1029–1039. https://doi.org/10.1111/j.1365- 15, 59–69. https://doi.org/10.1111/j.1472-4642.2008.00491.x.
2664.2008.01488.x. McLean, M., Mouillot, D., Maureaud, A.A., Hattab, T., MacNeil, M.A., Goberville, E.,
Elliott, S.A.M., Deleys, N., Beaulaton, L., Rivot, E., Réveillac, E., Acou, A., 2023. Lindegren, M., Engelhard, G., Pinsky, M., Auber, A., 2021. Disentangling
Fisheries-dependent and -Independent data used to model the distribution of tropicalization and deborealization in marine ecosystems under climate change.
diadromous fish at-sea. Data Brief 48, 109107. https://doi.org/10.1016/j.dib.20 Curr. Biol. 31, 4817–4823.e5. https://doi.org/10.1016/j.cub.2021.08.034.
23.109107. Melo-Merino, S.M., Reyes-Bonilla, H., Lira-Noriega, A., 2020. Ecological niche models
Ferreira, C.C., Stephenson, P.J., Gill, M., Regan, E.C., 2021. Biodiversity monitoring and and species distribution models in marine environments: a literature review and
the role of scientists in the twenty-first century. In: Ferreira, C.C., Klütsch, C.F.C. spatial analysis of evidence. Ecol. Model. 415, 108837. https://doi.org/10.1016/j.
(Eds.), Closing the Knowledge-Implementation Gap in Conservation Science: ecolmodel.2019.108837.
Interdisciplinary Evidence Transfer across Sectors and Spatiotemporal Scales, Montgomery, D.C., 2017. Design and Analysis of Experiments. John Wiley & Sons.
Wildlife Research Monographs. Springer International Publishing, Cham, pp. 25–50. Morato, T., González-Irusta, J.-M., Dominguez-Carrió, C., Wei, C.-L., Davies, A.,
https://doi.org/10.1007/978-3-030-81085-6_2. Sweetman, A.K., Taranto, G.H., Beazley, L., García-Alegre, A., Grehan, A.,
Fournier, A., Barbet-Massin, M., Rome, Q., Courchamp, F., 2017. Predicting species Laffargue, P., Murillo, F.J., Sacau, M., Vaz, S., Kenchington, E., Arnaud-Haond, S.,
distribution combining multi-scale drivers. Glob. Ecol. Conserv. 12, 215–226. Callery, O., Chimienti, G., Cordes, E., Egilsdottir, H., Freiwald, A., Gasbarro, R.,
https://doi.org/10.1016/j.gecco.2017.11.002. Gutiérrez-Zárate, C., Gianni, M., Gilkinson, K., Wareham Hayes, V.E., Hebbeln, D.,
Frans, V.F., Augé, A.A., Edelhoff, H., Erasmi, S., Balkenhol, N., Engler, J.O., 2018. Hedges, K., Henry, L.-A., Johnson, D., Koen-Alonso, M., Lirette, C., Mastrototaro, F.,
Quantifying apart what belongs together: a multi-state species distribution Menot, L., Molodtsova, T., Durán Muñoz, P., Orejas, C., Pennino, M.G., Puerta, P.,
modelling framework for species using distinct habitats. Methods Ecol. Evol. 9, Ragnarsson, S.Á., Ramiro-Sánchez, B., Rice, J., Rivera, J., Roberts, J.M., Ross, S.W.,
98–108. https://doi.org/10.1111/2041-210X.12847. Rueda, J.L., Sampaio, Í., Snelgrove, P., Stirling, D., Treble, M.A., Urra, J., Vad, J., van
Oevelen, D., Watling, L., Walkusz, W., Wienberg, C., Woillez, M., Levin, L.A.,
Carreiro-Silva, M., 2020. Climate-induced changes in the suitable habitat of cold-

14
A. Janc et al. Estuarine, Coastal and Shelf Science 310 (2024) 109013

water corals and commercially important deep-sea fishes in the North Atlantic. Glob. Rutterford, L.A., Simpson, S.D., Bogstad, B., Devine, J.A., Genner, M.J., 2023. Sea
Change Biol. 26, 2181–2202. https://doi.org/10.1111/gcb.14996. temperature is the primary driver of recent and predicted fish community structure
Muus, B.J., Dahlström, P., Bovet, P., 1988. Guide des poissons de mer et de pêche across Northeast Atlantic shelf seas. Glob. Change Biol. 29, 2510–2521. https://doi.
(biologie- pêche- importance économique). Guid. Nat. org/10.1111/gcb.16633.
Neiva, J., Serrão, E.A., Assis, J., Pearson, G.A., Coyer, J.A., Olsen, J.L., Hoarau, G., Schickele, A., Leroy, B., Beaugrand, G., Goberville, E., Hattab, T., Francour, P.,
Valero, M., 2016. Climate oscillations, range shifts and phylogeographic patterns of Raybaud, V., 2020. Modelling European small pelagic fish distribution:
north atlantic fucaceae. In: Hu, Z.-M., Fraser, C. (Eds.), Seaweed Phylogeography. methodological insights. Ecol. Model. 416, 108902. https://doi.org/10.1016/j.
Springer, Netherlands, Dordrecht, pp. 279–308. https://doi.org/10.1007/978-94- ecolmodel.2019.108902.
017-7534-2_11. Seitz, R.D., Wennhage, H., Bergström, U., Lipcius, R.N., Ysebaert, T., 2014. Ecological
Núñez-Riboni, I., Taylor, M.H., Kempf, A., Püts, M., Mathis, M., 2019. Spatially resolved value of coastal habitats for commercially and ecologically important species. ICES
past and projected changes of the suitable thermal habitat of North Sea cod (Gadus J. Mar. Sci. 71, 648–665. https://doi.org/10.1093/icesjms/fst152.
morhua) under climate change. ICES J. Mar. Sci. 76, 2389–2403. https://doi.org/ Servili, A., Lévêque, E., Mouchel, O., Devergne, J., Lebigre, C., Roussel, S., Mazurais, D.,
10.1093/icesjms/fsz132. Zambonino-Infante, J.-L., 2022. Ocean acidification alters the acute stress response
Ordonez, A., Williams, J.W., 2013. Climatic and biotic velocities for woody taxa of a marine fish. https://doi.org/10.2139/ssrn.4116088.
distributions over the last 16 000 years in eastern North America. Ecol. Lett. 16, Singer, A., Schweiger, O., Kühn, I., Johst, K., 2018. Constructing a hybrid species
773–781. https://doi.org/10.1111/ele.12110. distribution model from standard large-scale distribution data. Ecol. Model. 373,
Pauly, D., Cheung, W.W.L., 2018. Sound physiological knowledge and principles in 39–52. https://doi.org/10.1016/j.ecolmodel.2018.02.002.
modeling shrinking of fishes under climate change. Glob. Change Biol. 24, e15–e26. Sohl, T.L., 2014. The relative impacts of climate and land-use change on conterminous
https://doi.org/10.1111/gcb.13831. United States bird species from 2001 to 2075. PLoS One 9, e112251. https://doi.org/
Pecl, G.T., Araújo, M.B., Bell, J.D., Blanchard, J., Bonebrake, T.C., Chen, I.-C., Clark, T. 10.1371/journal.pone.0112251.
D., Colwell, R.K., Danielsen, F., Evengård, B., Falconi, L., Ferrier, S., Frusher, S., Soultan, A., Pavón-Jordán, D., Bradter, U., Sandercock, B.K., Hochachka, W.M.,
Garcia, R.A., Griffis, R.B., Hobday, A.J., Janion-Scheepers, C., Jarzyna, M.A., Johnston, A., Brommer, J., Gaget, E., Keller, V., Knaus, P., Aghababyan, K.,
Jennings, S., Lenoir, J., Linnetved, H.I., Martin, V.Y., McCormack, P.C., Maxhuni, Q., Vintchevski, A., Nagy, K., Raudonikis, L., Balmer, D., Noble, D.,
McDonald, J., Mitchell, N.J., Mustonen, T., Pandolfi, J.M., Pettorelli, N., Popova, E., Leitão, D., Øien, I.J., Shimmings, P., Sultanov, E., Caffrey, B., Boyla, K., Radišić, D.,
Robinson, S.A., Scheffers, B.R., Shaw, J.D., Sorte, C.J.B., Strugnell, J.M., Sunday, J. Lindström, Å., Velevski, M., Pladevall, C., Brotons, L., Karel, Š., Rajković, D.Z.,
M., Tuanmu, M.-N., Vergés, A., Villanueva, C., Wernberg, T., Wapstra, E., Chodkiewicz, T., Wilk, T., Szép, T., Turnhout, C. van, Foppen, R., Burfield, I.,
Williams, S.E., 2017. Biodiversity redistribution under climate change: impacts on Vikstrøm, T., Mazal, V.D., Eaton, M., Vorisek, P., Lehikoinen, A., Herrando, S.,
ecosystems and human well-being. Science 355, eaai9214. https://doi.org/10.1126/ Kuzmenko, T., Bauer, H.-G., Kalyakin, M.V., Voltzit, O.V., Sjeničić, J., Pärt, T., 2022.
science.aai9214. The future distribution of wetland birds breeding in Europe validated against
Penn, J.L., Deutsch, C., 2022. Avoiding ocean mass extinction from climate warming. observed changes in distribution. Environ. Res. Lett. 17, 024025. https://doi.org/
Science 376, 524–526. https://doi.org/10.1126/science.abe9039. 10.1088/1748-9326/ac4ebe.
Perry, A.L., Low, P.J., Ellis, J.R., Reynolds, J.D., 2005. Climate change and distribution Teichert, N., Pasquaud, S., Borja, A., Chust, G., Uriarte, A., Lepage, M., 2017. Living
shifts in marine fishes. Science 308, 1912–1915. https://doi.org/10.1126/ under stressful conditions: fish life history strategies across environmental gradients
science.1111322. in estuaries. Estuar. Coast Shelf Sci. 188, 18–26. https://doi.org/10.1016/j.
Peterson, A.T., Papeş, M., Soberón, J., 2015. Mechanistic and correlative models of ecss.2017.02.006.
ecological niches. https://doi.org/10.1515/eje-2015-0014. Thorson, J.T., Pinsky, M.L., Ward, E.J., 2016. Model-based inference for estimating shifts
Pinsky, M.L., Selden, R.L., Kitchel, Z.J., 2020. Climate-driven shifts in marine species in species distribution, area occupied and centre of gravity. Methods Ecol. Evol. 7,
ranges: scaling from organisms to communities. Annu. Rev. Mar. Sci. 12, 153–179. 990–1002. https://doi.org/10.1111/2041-210X.12567.
https://doi.org/10.1146/annurev-marine-010419-010916. Thuiller, W., Lafourcade, B., Engler, R., Araújo, M.B., 2009. Biomod – a platform for
Pinsky, M.L., Worm, B., Fogarty, M.J., Sarmiento, J.L., Levin, S.A., 2013. Marine taxa ensemble forecasting of species distributions. Ecography 32, 369–373. https://doi.
track local climate velocities. Science 341, 1239–1242. https://doi.org/10.1126/ org/10.1111/j.1600-0587.2008.05742.x.
science.1239352. Tikhonov, G., Abrego, N., Dunson, D., Ovaskainen, O., 2017. Using joint species
Poloczanska, E.S., Brown, C.J., Sydeman, W.J., Kiessling, W., Schoeman, D.S., Moore, P. distribution models for evaluating how species-to-species associations depend on the
J., Brander, K., Bruno, J.F., Buckley, L.B., Burrows, M.T., Duarte, C.M., Halpern, B.S., environmental context. Methods Ecol. Evol. 8, 443–452. https://doi.org/10.1111/
Holding, J., Kappel, C.V., O’Connor, M.I., Pandolfi, J.M., Parmesan, C., Schwing, F., 2041-210X.12723.
Thompson, S.A., Richardson, A.J., 2013. Global imprint of climate change on marine Vinagre, C., Fonseca, V., Cabral, H., Costa, M.J., 2006. Habitat suitability index models
life. Nat. Clim. Change 3, 919–925. https://doi.org/10.1038/nclimate1958. for the juvenile soles, Solea solea and Solea senegalensis, in the Tagus estuary:
Potter, I.C., Tweedley, J.R., Elliott, M., Whitfield, A.K., 2015. The ways in which fish use defining variables for species management. Fish. Res. 82, 140–149. https://doi.org/
estuaries: a refinement and expansion of the guild approach. Fish Fish. 16, 230–239. 10.1016/j.fishres.2006.07.011.
https://doi.org/10.1111/faf.12050. Whitfield, A.K., Able, K.W., Blaber, S.J.M., Elliott, M., Franco, A., Harrison, T.D.,
Quéro, J.C., Vayne, 1997. Les poissons de mer des peches françaises. Delachaux et Potter, I.C., Tweedley, J.R., 2022. Fish assemblages and functional groups. In: Fish
niestlé, pp. 245–246. Paris Fr. and Fisheries in Estuaries. John Wiley & Sons, Ltd, pp. 16–59. https://doi.org/
R Core Team, 2021. R: a Language and Environment for Statistical Computing. R 10.1002/9781119705345.ch2.
Foundation for Statistical Computing, Vienna, Austria. Whitfield, A.K., Houde, E.D., Neira, F.J., Potter, I.C., 2023. Importance of marine-
Ramos Martins, M., Assis, J., Abecasis, D., 2021. Biologically meaningful distribution estuarine-riverine connectivity to larvae and early juveniles of estuary-associated
models highlight the benefits of the Paris Agreement for demersal fishing targets in fish taxa. Environ. Biol. Fishes 106, 1983–2009. https://doi.org/10.1007/s10641-
the North Atlantic Ocean. Glob. Ecol. Biogeogr. 30, 1643–1656. https://doi.org/ 023-01474-2.
10.1111/geb.13327. Wynn, R.B., Josey, S.A., Martin, A.P., Johns, D.G., Yésou, P., 2007. Climate-driven range
Randin, C.F., Dirnböck, T., Dullinger, S., Zimmermann, N.E., Zappa, M., Guisan, A., expansion of a critically endangered top predator in northeast Atlantic waters. Biol.
2006. Are niche-based species distribution models transferable in space? J. Biogeogr. Lett. 3, 529–532. https://doi.org/10.1098/rsbl.2007.0162.
33, 1689–1703. https://doi.org/10.1111/j.1365-2699.2006.01466.x.

15