https://orcid.org/0000-0003-2839-0576
Figures
Abstract
Non-targeted fish species contribute significantly to the structure and functioning of marine ecosystems, but they remain largely understudied. In this study, species distribution models (SDMs) were applied using the MaxEnt algorithm to assess current and future habitat suitability for 103 non-targeted species in the Mediterranean Sea. Based on trawl survey data from 2010 to 2022, and nine environmental variables, SDMs were calibrated and projected under three climate change scenarios (SSP1-1.9, SSP2-4.5, SSP5-8.5) for two future periods (2040–2050, 2090–2100). Among the environmental layers, depth, distance to the coast and sea bottom temperature explained most of the species’ distribution. Results indicate a general decline in suitable habitats for 97% of the species, particularly under the high emissions scenario SSP5-8.5 by 2100, independent of habitat type, with slightly greater declines for higher-trophic level species. Several species are projected to be at high risk of local or regional extinction. Moreover, future species distributions exhibited large spatial shifts of the centroids compared to the present-day, mainly explained by the contraction of suitable habitats rather than by expansion to new areas. Areas in the central and eastern basin of the Mediterranean Sea exhibited the highest levels of species turnover, indicating their potential higher vulnerability under climate change. These findings highlight the importance of integrating non-targeted species into conservation planning and fisheries management, as well as the need to prioritize areas that can sustain biodiversity in a rapidly changing Mediterranean ecosystem.
Citation: Orfanidis GA, Touloumis K, Koutrakis E, Tsikliras AC (2026) Identifying suitable habitats under climate change for non-targeted demersal fish in the Mediterranean Sea. PLOS Clim 5(4): e0000838. https://doi.org/10.1371/journal.pclm.0000838
Editor: Lalit Kumar Sharma, Zoological Survey of India, INDIA
Received: September 29, 2025; Accepted: March 17, 2026; Published: April 3, 2026
Copyright: © 2026 Orfanidis et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All data, along with R scripts supporting the findings of this study, are openly available in Figshare at https://doi.org/10.6084/m9.figshare.c.8325984.
Funding: This work was supported by the H2020 funded project “EcoScope” (Contract No. 101000302 to ACT). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: Co-author Athanassios C. Tsikliras is a PLOS ONE Editorial Board member. This does not alter the authors’ adherence to PLOS ONE Editorial policies and criteria.
Introduction
The Mediterranean Sea supports a rich diversity of marine species, including numerous demersal fish [1,2]. While commercially important species often receive significant attention in marine research and management [3], information on the distribution, status of population stocks [4] and biological parameters of non-targeted species is generally scarce. This is largely because most scientific journals prioritize research on commercially valuable species [5,6]. Non-targeted fish species may include those with low or no commercial value, which when fished are often discarded, i.e., returned back to the sea with low survival rates [7,8].
Although non-targeted species often receive little research attention compared to commercially important stocks, they contribute to ecosystem functions and services [9,10]. Despite their lack of commercial value, they contribute to biodiversity, which enhances ecosystem resilience in the face of environmental changes, such as climate shifts [11–13]. A decline in their biomass and abundance may negatively impact populations of economically valuable species [14,15], as they consist key components of the food web, interacting with commercial species and affecting the sustainability of their stocks [13,14,16].
The significant quantities of non-targeted species incidentally caught in Mediterranean multi-species fisheries, which are frequently discarded [17], emphasize the need for a deeper understanding of these species and their ecological roles. Certain fishing gears, such as bottom trawlers can generate discards that can constitute up to 55% of the total catch [18] or even 70.7% in deep sea trawling [15]. This low selectivity of many Mediterranean multi-species fisheries combined with the cascading negative effects within the marine communities [19], often results in poor condition for these non-targeted species, based on recent stock assessment studies in the northern Aegean Sea [20]. Consequently, various management measures, including, among others, improving gear selectivity [21,22] and regulations such as the landing obligation of the unwanted catch across European Union (EU Reg. No 1380/2013), have been implemented to reduce discarded catch.
Understanding the distribution patterns of non-targeted demersal fish species is important for effective conservation strategies and ecosystem management, especially in the face of increasing anthropogenic pressures, such as overfishing, habitat degradation, and climate change [23–25]. Recent advances in species distribution modeling (SDM) have provided valuable tools for assessing and predicting species distributions based on environmental variables [26,27]. Several methods are commonly applied for SDM and their selection depends on the available data (e.g., presence only or presence-absence), the complexity of the relationships being modeled and the objectives of the study. Some of the most widely applied methods include Generalized Additive Models (GAM) [28,29], Artificial Neural Networks [30], Boosted Regression Trees [31], Random Forest [32], Maximum Entropy (MaxEnt) [33], or a compilation of them, called ensemble models [34–37].
Among these modeling approaches, MaxEnt has gained significant attention due to its ability to produce robust predictions with limited occurrence data [33]. MaxEnt correlates environmental conditions with species occurrences and identifies areas with similar environmental profiles where the species is likely to occur [33]. Several studies have demonstrated that MaxEnt often outperforms traditional methods, such as Generalized Linear Models (GLM) and GAM [38,39]. Its reliance on presence-only data and environmental variables makes it especially suitable for studying non-targeted species, where data may be sparse [40–42]. MaxEnt has been successfully applied to the distribution of fish species in both marine [43–45] and freshwater ecosystems [46,47].
In addition to modeling current species distributions, the potential impacts of climate change on species distribution modelling have gained increasing attention [45,48]. Climate change may impact marine ecosystems in multiple ways, including shifting species distribution [49,50] and changing biodiversity [51,52]. It may also reshape community structure [53], modify food webs [54,55], and influence the productivity of marine organisms and ecosystems [56,57]. Despite the growing use of SDMs for marine fishes, they are rarely applied in marine conservation and planning efforts [44,58]. Moreover, most research mainly focuses on current and future distributions of invasive species, such as the devil firefish, Pterois miles [59,60], or commercially important fish, such as anglerfishes Lophius spp., surmullet Mullus surmuletus, red mullet Mullus barbatus, European hake Merluccius merluccius, common sole Solea solea, common pandora Pagellus erythrinus, European seabass Dicentrarchus labrax and gilthead seabream Sparus aurata [34], while studies on non-targeted fish species in the Mediterranean Sea remain limited [61].
This study addresses these gaps by developing SDMs for 103 key non-targeted fish species primarily associated with demersal habitats across the entire Mediterranean Sea using fisheries independent data and the MaxEnt algorithm. The main objectives of this study were to: (1) develop and evaluate SDMs for 103 selected non-targeted species associated with demersal trawl surveys; (2) identify the environmental variables that influence their distribution; (3) estimate short and long term potential future distributions for each species under three climate change scenarios; (4) identify areas where could potentially retain most of the species under climate change; and (5) provide insights that can inform conservation strategies and fisheries management in the Mediterranean Sea. By integrating ecological modeling with survey data, this research contributes to a better understanding of the distribution dynamics of non-targeted demersal species in the Mediterranean Sea and can be used to improve sustainable fisheries management.
Materials and methods
Species occurrences
Species presence data was collected during the International Bottom Trawl Survey in the Mediterranean (MEDITS; [62]) from 2010 to 2022 (Fig 1). This survey is being conducted since 1994, by eight EU member states (Croatia, Cyprus, France, Greece, Italy, Malta, Slovenia and Spain), across the northern Mediterranean Sea coastline. In the past, 28 different fishing vessels have been used, a fact that may have introduced varying levels of sampling bias. However, all the parties follow a standardized protocol [63], on which the sampling stations have been selected to cover most of the distribution areas of exploited and potentially exploitable demersal species, with hauls towed at depths ranging from 10 to 800 meters. On average, approximately 1166 sampling stations were surveyed annually over the 12-year period (2010–2022), providing the sufficient spatial and temporal coverage of non-targeted demersal fish occurrences across the survey area.
Red symbols indicate the positions of hauls where samples were collected. Basemap derived from Natural Earth (https://www.naturalearthdata.com).
The distinction between targeted and non-targeted fish stocks for analysis was based on previous literature [20,64]. Species that are primary targets of fisheries were excluded, while non-targeted species (S1 Table) were included for further analysis. Occurrence tables for each non-targeted species were then created using the collected survey data. To address spatial autocorrelation and sampling bias, a spatial thinning procedure was applied using the “thin” function from the “spThin” R package [65]. This procedure removed occurrences of the same species within a 10 km radius. Species with fewer than 50 presence records after thinning or that appeared for less than 5 years in the dataset were excluded from further analysis to ensure robust model calibration and evaluation. Previous studies have shown that SDMs calibrated with very small sample sizes or temporally few occurrences may produce unstable or unreliable predictions, particularly under future climate projections [40,41].
This process resulted in 103 Actinopterygian species, primarily associated with demersal habitats and sampled through bottom trawl surveys, being retained for modeling, with the number of records per species ranged from 50 (for greater pipefish, Syngnathus acus) to 858 (for brown comber, Serranus hepatus). A complete list of the species included in the analysis is provided in S1 Table.
Environmental layers
The selection of environmental ocean layers for the analysis was based on the biology and ecology of the selected species. Sea surface and bottom temperature, as well as depth are among the primary factors determining the spatial and vertical distribution of demersal species [66], as they directly affect metabolic rates and physiological tolerances [67]. Dissolved oxygen concentration at the seabed is another important variable, especially in regions prone to hypoxia, which can severely constrain habitat suitability for sensitive species [68]. Salinity and bottom water currents further influence species assemblages by affecting reproduction, behavior [69], as well as larval survival and dispersal [70]. In addition, surface chlorophyll-a concentration serves as a proxy for primary productivity, indirectly affecting benthic food availability and trophic interactions in demersal communities [71]. Finally, topographic features such as slope and proximity to the coastline influence substrate type and habitat complexity, which are key determinants of demersal fish community structure [72] and can be further modified by fishing activities, thereby influencing species distribution [73].
Therefore, the environmental variables were selected to ensure that the environmental niche models captured the key ecological gradients relevant to the distribution of non-targeted demersal fish species in the Mediterranean Sea. Environmental variables included mean and range of sea surface (SST) and sea bottom (SBT) temperature (°C), mean sea bottom salinity (SBS; PSU), mean sea bottom dissolved oxygen concentration (DO; mmol/m3), mean surface chlorophyll a concentration (CHL; mg/m3) and mean bottom sea water speed (SWS; m/s). Furthermore, topographic data layers including the mean depth (meters), the slope of the seabed (degrees) and the distance from the nearest shore, estimated based on a raster layer of coastline boundaries, were integrated into the models. All layers had a spatial resolution of 3 arc minutes and were downloaded from the Bio-ORACLE database v.3 [74,75].
Multicollinearity among the selected variables was tested using Pearson correlation coefficients (r), as well as the variance inflation factor (VIF). In the case of collinear variables (i.e., r > 0.7 or VIF > 10), only a single variable was retained, to avoid redundancy. Thus, the final selection of environmental layers included the SBT mean, SBT range, SBS, DO, CHL, SWS, depth, slope and distance from shore.
Distribution modelling
Habitat suitability for each species was predicted with the MaxEnt algorithm, a machine learning method that models species distributions by estimating the probability distribution of maximum entropy, constrained by environmental variables and presence-only data [33], implemented within the R environment [76]. Specifically, the “dismo” package [77] was used to run the distribution models for each species separately. Furthermore, the “ENMevaluate” function from the “ENMeval” package [78] was used for model optimization. This process involved testing several feature classes (linear, quadratic, linear-quadratic, linear-quadratic-hinge and linear-quadratic-hinge-product), as well as a range of regularization multipliers (from 0.5 to 3 with a step of 0.5). A total of 36 unique combinations of feature classes and regularization multipliers were tested per species to determine the best-performing model based on the Akaike Information Criterion (AIC) criterion.
All models were validated with k-fold cross validation, with the number of folds (k) depending on the number of records for each species; 5 folds were used for species with fewer than 500 records, while 10 folds were used for species with more than 500 records. Additionally, 15,000 background points were used for species with fewer than 500 records, and 20,000 points for species with more than 500 records [79].
Once the best model was identified based on AIC, its performance was further evaluated using multiple metrics, including the Area Under the Curve (AUC; [80]), Continuous Boyce Index (CBI; [81]), True Skill Statistic (TSS; [82]), Cohen’s Kappa (CK; [83]), and Somers’ D (D; [84]). AUC and CBI are threshold-independent metrics: AUC measures the model’s ability to discriminate between presence and background points [85], while CBI evaluates how predictions differ from a random distribution of presences across the prediction gradient [81]. On the other hand, TSS, CK, and D are threshold-dependent and require binary classification of suitable and unsuitable habitats. TSS accounts for both omission and commission errors and is not affected by prevalence [82], CK measures the agreement between observed and predicted occurrences beyond the level of agreement that could be expected by chance [83] and D assesses the strength and direction of asymmetric association between predicted probabilities and observed outcomes [84]. Therefore, the threshold that maximizes the sum of sensitivity and specificity (maxSSS) was chosen for calculating these threshold-dependent metrics, following the methodology of [86,87]. Using these metrics provided a robust and multidimensional evaluation of model performance, enhancing confidence in the reliability and predictive accuracy of the species distribution estimates.
Future climate scenarios
To model future habitat suitability, climate projections of the same environmental variables used for the present distributions were downloaded from the Bio-ORACLE database for two future time periods: 2040–2050 and 2090–2100. Near-future projections (e.g., centered around 2030) were not considered, as climatic changes are expected to be small relative to natural variability and subsequently limit the ability of SDMs to detect species-level responses. Topographic layers were assumed to remain constant, as future changes in these parameters are currently unknown. The three future climate scenarios used, known as Shared Socioeconomic Pathways (SSP1-1.9, SSP2-4.5 and SSP5-8.5), represent varying levels of socioeconomic development and emissions trajectories [88]. SSP1-1.9 assumes low emissions and high societal efforts for climate change mitigation, SSP2-4.5 represents a “middle-of-the-road” scenario with moderate emissions, while SSP-8.5 corresponds to high emissions driven by rapid economic growth and fossil fuel dependency [88].
Species distribution models were then examined under each of the three SSPs for both time periods to assess the potential impacts of different emission pathways on species distributions for the two projection periods. Binary habitat suitability maps were generated for each species by applying the maxSSS threshold to presence probabilities predicted by the SDMs. Based on these projections, the number of suitable habitat grids for each species was estimated under each scenario and time period and relative percentage changes compared to the present distribution were calculated to quantify potential shifts in habitat suitability.
Moreover, species-level binary suitability maps were aggregated by habitat type (demersal, pelagic, reef-associated and other; S1 Table), to construct overall habitat-based suitability maps. The “other” category includes species that could not be assigned to a single habitat type (e.g., bathypelagic, benthopelagic and bathydemersal species). Relative percentage changes in suitable cells between current and future scenarios were then estimated at species levels and then summarized descriptively by habitat type. In addition, a post hoc trait-based summary was conducted using trophic level (TL) information (FoodTroph) retrieved from FishBase [89]. Species were grouped into two trophic guilds based on their trophic level using a threshold of 3.5, separating mid-trophic (<3.5) from higher-trophic level species (≥3.5), as an operational classification to facilitate comparison of projected habitat changes between functional groups, given that trophic level is a continuous ecological trait. This threshold was selected as the midpoint within the observed trophic levels of the studied species (TL range = 3.0-4.5) and is consistent with empirical trophic structure studies in which mid-trophic level fishes typically occupy trophic positions between 3.0 and 3.5, while higher-trophic species are characterized by values ≥3.5 [90]. Projected habitat change was then summarized as the mean percentage change in suitable habitat per species within each guild. Trophic level data were available for 90 out of 103 species and subsequently the 13 species lacking this information were excluded from the trophic guild aggregation, while being retained in all SDM analyses. This classification resulted in two similarly sized groups (n = 43 for <3.5 and n = 47 for ≥3.5), allowing a balanced functional comparison between trophic guilds.
Furthermore, to assess spatial shifts in species distributions under climate change scenario SSP5-8.5 projected for the period 2090–2100, multiple spatial analyses performed. First, the geographic centroid of suitable habitats for each species was calculated under both present and future distribution, based on the binary maps on the maxSSS threshold. Then, Euclidean distances between present and future centroids were computed and expressed in kilometers to quantify the magnitude of distributional shifts, similar to [50]. Moreover, the dominant direction of centroid movement (e.g., northward, southward, eastward, or westward) was identified for each species. To further evaluate whether the observed shifts were primarily driven by colonization of new areas or by the decline of present suitable habitats, habitat gains, losses and net changes in the number of suitable grid cells were estimated for each species and subsequently aggregated by habitat type and scenario for descriptive comparison.
In addition to these comparisons based on the binary maps, shifts in predicted suitability were assessed using continuous probability maps. This was implemented by comparing the average predicted probability of occurrence across all species at each grid cell between the present and the future scenario, similar to [91]. This map highlights areas where mean suitability may increase or decrease, even if the species is not predicted to be fully absent or present.
Finally, to identify which areas gain or lose the most species and could be potentially at higher risk under climate change, species turnover (T) was calculated for each grid cell following the methodology of [92,93]. Turnover was estimated as , where L represents the number of species lost, G the number of species gained and SR the current species richness.
Results
Modeling performance and current habitat suitability
The mean AUC values of the predicted models ranged from 0.88 to 0.98, indicating a high level of discrimination power between presence locations and background points [85] (S2 Table). The CBI values for all models were similarly high, ranging from 0.48 to 0.97, with a mean value of 0.87, confirming that the predicted habitat suitability closely matches the observed distribution of species and differs significantly from random expectations [81]. Additionally, the threshold-dependent indices, estimated at the maxSSS threshold, further demonstrated the models’ performance. TSS values ranged from 0.7 to 0.93, demonstrating a strong balance between sensitivity and specificity, while Cohen’s Kappa (0.74-0.94) indicated a high level of agreement between observed and predicted occurrences beyond random chance [83]. Somers’ D ranged from 0.78 to 0.97, highlighting a strong positive relationship between predicted probabilities and the actual presence-absence classifications.
The permutation importance analysis of environmental layers incorporated into the models revealed that depth was the most significant predictor for all species, with an average contribution of approximately 74.4%. The second most influential layer was distance from the coast, with a mean contribution of 8.4%, followed by mean SBT at 5.5%. Mean CHL contributed 4.6%, while the range in SBT accounted for around 3.7%. Slope, mean SBS and mean DO each contributed approximately 1%. Layers with less than 1% contribution included only mean SWS (0.2%) indicating its minimal influence on the models. The specific permutation importance for each environmental layer and species is presented in detail in S3 Table.
At the species level, foureyed sole Microchirus ocellatus showed the strongest response to mean depth, with a contribution of 93%. Bandtooth conger Ariosoma balearicum exhibited the largest contribution from distance to the coast (53%), while glacier lantern fish Benthosema glaciale had the highest contribution of mean CHL (26%). Mean SBT contributed most significantly to Cadenat’s rockfish Scorpaena loppei at 21%, while the range of SBT had its greatest impact on Facciola’s sorcerer Facciolella oxyrhynchus at 26%. Slope presented its greatest influence on slender snipe eel Nemichthys scolopaceus at 8% and both common Atlantic grenadier Nezumia aequalis and Ariosoma balearicum showed the greatest contribution of mean SBS (11%). Finally, mean DO had its greatest influence on armless snake eel Dalophis imberbis at 10%, while the highest contribution of SWS was presented by whiskered sole Monochirus hispidus (2%).
At the present-day scenario, suitable habitats ranged from 6,344 grid cells for crystal goby Crystallogobius linearis to 27,476 grid cells for European conger Conger conger, corresponding to approximately 131,320 km² and 568,749 km², respectively. For all modelled species, the estimated current suitable areas align with their known biogeographic distributions in Mediterranean Sea [89] (S1 Fig).
Future projections and habitat suitability changes
The projection of the future environmental variables under SSP5-8.5 scenario revealed an overall increase in mean SBS, mean and range of SBT, mean SBS and mean CHL, while mean DO and mean SWS decrease by the end of the century (Fig 2). These projected changes in environmental variables are closely linked to significant shifts in habitat suitability for demersal fish species. Notably, 97% of the modeled species are predicted to experience a decline in suitable habitat across all future climate scenarios, with the largest reductions under the high-emissions pathway SSP5-8.5 (Fig 3, S4 Table).
Topographic layers are omitted as they are assumed to remain constant in the future. Basemap derived from Natural Earth (https://www.naturalearthdata.com).
Comparisons are made between the present-day and two future time periods (2040–2050 and 2090–2100). The box represents the interquartile range (50% of values), the solid black line indicates the median, whiskers extend to values within 1.5 times the interquartile range, and points outside this range represent potential outliers.
Among the most affected species, Scorpaena lopei and Crystallogobius linearis are projected to face the largest relative declines with reductions of 46% and 49% by 2050 and further declines of 99.5% and 97.5% respectively, by 2100 under the SSP5-8.5 scenario (S2 Fig). Other species, such as Fries’s goby Lesueurigobius friesii, butterfly blenny Blennius ocellaris, starry weever Trachinus radiatus, snake blenny Ophidion barbatum, poor cod Trisopterus capelanus and stargazer Uranoscopus scaber are also expected to lose most of their suitable habitats, raising concerns about high risk of potential local extinctions. In contrast, only two species, spothead lantern fish Diaphus metopoclampus and Facciolella oxyrhynchus, showed an increase in suitable habitat by 2100, with a gain of 12 and 13% respectively. Meanwhile, species such as Nezumia sclerorhynchus are projected to maintain relatively stable suitable habitats across all climate scenarios (S2 Fig, S5 Table).
When species were grouped by habitat type, declines in habitat suitability were observed across all categories, although the magnitude of loss differed among habitat types. Reef-associated species showed the greatest relative declines, particularly under high-emission scenarios (SSP5-8.5), with an average habitat loss of 22% by 2050 and 73% by 2100 (Fig A in S3 Fig). Demersal (Fig 4) and pelagic (Fig B in S3 Fig) species followed similar patterns, with average relative declines of 21% and 16% by 2050, and 72% and 64% by 2100, respectively. Species classified under the “other” habitat category exhibited the lowest average declines, with losses of 15% by 2050 and 48% by 2100 (Fig C in S3 Fig). A complementary summary of projected gains, losses and stable suitable grid cells aggregated by habitat type and SSP scenario is provided in S6 Table, enhancing the clarity of habitat-level changes in suitable areas under future climate conditions. Additionally, the post hoc aggregation by trophic guild revealed consistent declines in suitable habitat across ecological groups. Species with higher-trophic levels (≥3.5) exhibited slightly greater mean habitat losses compared to mid-trophic species (<3.5) across all scenarios, with losses reaching up to 19% by 2050 and 66% by 2100 under SSP5-8.5, whereas mid-trophic species showed losses of up to 16% by 2050 and 56% by 2100 (S4 Fig).
Bar plots indicate the relative percentage differences between present situation and future projections. The left column of the plot represents habitat suitability for future period 2040–2050, while the right column represents habitat suitability for the period 2090–2100. Basemap derived from Natural Earth (https://www.naturalearthdata.com).
Regarding the spatial dynamics of species distributions under climate change, the direction of centroid displacement between the present-day and the SSP5-8.5 scenario projected for 2100 revealed clear trends. Specifically, 64% of the modelled species exhibited a westward shift in their distribution centroids, while 16% shifted eastward, 10% northward, and 11% southward (S5 Table). The magnitude of these shifts varied among species, ranging from 1 km for Sloane’s viperfish Chauliodus sloani to a maximum of 836 km for spotted dragonet Callionymus maculatus, indicating highly species-specific responses to future climate conditions.
In terms of mean habitat suitability difference in the prediction probability between the present-day and SSP5-8.5 by 2100, the map (Fig 5) showed that central and west basin of the Mediterranean Sea present the highest probability differences. On the other hand, species turnover illustrated its highest levels in areas located on the central and eastern basin of the Mediterranean, such as the continental shelf of Adriatic and Aegean Sea (Fig 6).
Basemap derived from Natural Earth (https://www.naturalearthdata.com).
Yellow indicates areas with high species turnover (T), while blue indicates areas with low turnover. Basemap derived from Natural Earth (https://www.naturalearthdata.com).
Discussion
This study employed species distribution models (SDMs) to assess the current and future distribution of 103 non-targeted demersal fish species in the Mediterranean Sea. The findings reveal significant spatial variation in the habitat suitability of these species, with many experiencing considerable distributional shifts under future climate change scenarios. These results align with patterns observed in marine ecosystems, where climate change is expected to shift species’ distributions [50,94,95] and potentially affect ecosystem functioning [51,56]. In some cases, the projected contractions of suitable habitats are so large that they may lead to local or even regional extinctions [96–98] as the relevant habitats become unsuitable for most of the species under study, especially for those with limited distribution ranges or narrow environmental tolerances [99–101]. For instance, demersal species inhabiting shallow coastal zones, such as Microchirus ocellatus, showed the highest projected losses, highlighting their limited ability to inhabit deeper waters or other areas that remain suitable under climate change [102,103].
The models identified water depth as the most significant environmental predictor for the majority of species, highlighting the importance of bathymetry in shaping species distribution patterns [104,105]. As deeper water layers warm and oxygen minimum zones expand, depth restricted species may face a vertical habitat contraction as it would be difficult to find and migrate to optimal habitats [106–108]. Similarly, distance from shore influenced the spatial distributions of the modelled species, since many species often inhabit areas close to the coast, similar to findings of other studies [91,109,110]. This distribution pattern likely reflects higher habitat heterogeneity [111,112], increased food availability and high primary productivity on continental shelves [113], which in combination support important ecological functions including feeding and reproduction [114], as well as act as nursery and refuge habitats for many species [115,116]. However, coastal and shelf habitats are also subject to intensified anthropogenic pressures, including fishing activity, habitat modification and coastal development [117–119], which may limit their capacity to use shallow refugia as climate buffers and increase climate-driven habitat loss [61].
Moreover, sea bottom temperature was found as a major variable affecting species distribution, especially under climate change [61,96]. In particular, it has been shown that temperature may affect the abundance, survival, migration and phenology of marine species [99,107,120]. The rest of the environmental factors also affected species distribution, but to a lesser extent. This suggests that non-targeted species in the Mediterranean may be vulnerable to climate-induced changes in ocean stratification, which could alter the depth at which suitable conditions occur [106]. These findings are consistent with previous research emphasizing the role of abiotic factors, such as depth and temperature, in determining species fundamental niche in marine ecosystems [120–122].
Certain species showed a stronger response to specific environmental variables, reflecting their unique ecological requirements and vulnerability to climate change. For instance, shallow-water demersal species like Microchirus ocellatus were strongly influenced by mean depth, which aligns with its benthic ecology. Typically, this species is found on the continental shelves at moderate depths (up to 300 m) [89] and thus its’ distribution depends on depth related factors, such as substrate type, prey availability and temperature gradients, which may change under climate change conditions, limiting its’ suitable habitats [101]. On the other hand, shallow coastal flatfish like Bothus podas which are usually inhabiting shallow waters on the continental shelves [123] were more affected by distance to the coast, which suggests that any loss of shallow coastal habitat will likely have a direct impact on their persistence [115].
Projections under future climate scenarios varied across species and revealed a general decline in habitat suitability for the majority of species across all habitat types (demersal, pelagic, reef-associated, or other) and especially under higher emissions scenario (SSP5-8.5). Most of the species across all habitats are expected to experience a decrease in suitable habitats by more than 20% compared to present conditions by 2040–2050, with reductions reaching up to 99% by 2100. These patterns support previous research showing that climate change may decrease habitat availability for marine species [23,24,61] and could even drive native species to regional extinction in the Mediterranean [61,96,97].
Within demersal communities, species with wider depth ranges, such as Nezumia sclerorhynchus, or generalist thermal tolerances, such as Gaidropsarus mediterraneus, are predicted to maintain relatively stable number of suitable habitats across all scenarios. In contrast, pelagic species with narrow thermal ranges, or reef-associated species dependent on shallow habitats, are projected to undergo severe reductions in habitat suitability, reflecting their limited capacity to shift vertically or geographically. These interspecific differences may indicate that species capable of adapting to climate change [124], as well as eurytherms and depth generalists may maintain their distribution by moving deeper to escape warming [94,106,125]. On the other hand, stenothermal species and depth specialists may be severely impacted, experiencing reduced distributions that could potentially change community structure and predator-prey interactions [53,106].
From a functional perspective, the post hoc trophic guild aggregation indicated that higher-trophic level species (≥3.5) experienced slightly greater mean habitat losses compared to mid-trophic species (<3.5), particularly under high emissions (SSP5-8.5) and by the end of the century. This trend is consistent with the trophic amplification of climate change impacts, where climatic niche contraction and subsequent biomass declines are increased at higher-trophic levels due to their narrower dietary breadths and dependence on the stability of lower-level prey [126,127]. In contrast, species occupying intermediate trophic positions may benefit from greater ecological plasticity and the ability to switch resources, a characteristic of generalist species that often makes them more resilient to the biotic homogenization observed under global environmental change [128,129]. Consequently, climate-driven habitat contraction at higher-trophic levels may lead to cascading effects on community structure and food-web dynamics [54,130–132], particularly in Mediterranean marine ecosystems which already face the impacts of climate change [61,97].
While climate-driven poleward range shifts are commonly reported in marine environments [24,102], the findings of this study indicate a predominantly westward shift in species distribution centroids across the Mediterranean (64% of species) [94]. This pattern may be shaped by the unique oceanographic conditions of the Mediterranean. The semi-enclosed nature of the basin, combined with accelerated warming in the eastern Mediterranean [99,133] and limited dispersal connectivity, may constrain eastward expansion and instead drive westward displacement [96,97]. Furthermore, ecological constraints, such as salinity, productivity gradients, and bathymetry, could act as barriers creating “cul-de-sacs” [97,102] and restricting northward shifts [99,134], potentially trapping species in the northern areas of Mediterranean and leading to local extinctions as species cannot tolerate such conditions and reproduce [97]. This pattern also highlights that for many species range loss will outweigh any expansion and concentrate biodiversity in areas already intensively pressured from fisheries [100].
These findings highlight the complexity of region-specific responses of marine species to climate change and underscore the importance of incorporating regional oceanography and species-specific traits when projecting climate-driven range shifts [49,50]. Given that climate change impacts vary across regions, evaluating changes at a regional scale is necessary to develop effective mitigation strategies [50,100]. In most cases, projected distributional shifts appear to be driven more by the loss of currently suitable habitats than by expansion into newly suitable areas, a pattern consistent with findings from other studies [48]. This contraction of suitable ranges combined with the existing pressures from overfishing [135], habitat degradation [117,119], expansion of invasive species [134,136] and other anthropogenic factors may increase the vulnerability of species, supporting the importance of considering climate change impacts in conservation and management efforts [137,138]. Given that many non-targeted species are bycatch in multi-species Mediterranean fisheries [17], climate driven distributional shifts may also affect the discard rates, catch potential [139,140] and subsequently ecosystem balance, factors that should be considered in future policies.
One of the key insights from this study is the need to prioritize the conservation of species that are most vulnerable to climate change, such as Scorpaena loppei, Crystallogobius linearis and Lesueurigobius friesii. The significant habitat declines predicted for the majority of species highlight the importance of identifying potential areas where species may persist despite changing environmental conditions. Areas on the continental shelf of eastern and central basin of the Mediterranean; including the Aegean, Levantine, Adriatic, and Ionian Seas, as well as the Cretan Passage and the Gulf of Sirte, which are characterized by high levels of species turnover, could be potentially at risk as diversity shifts could lead to cascade effects impacting the whole food web of these areas [53,141]. Similar findings were reported by [97], who supported that 14 of the 75 endemic Mediterranean species studied would become fully extinct from these eastern and central areas by the end of the century. On the contrary, areas that remain more suitable under most climate scenarios, such as the west basin of Mediterranean, could act as climate refugia and be considered for enhanced protection of species at risk [138,142]. The establishment of protected areas in refugia areas and the development of strategies to mitigate the impacts of climate change on biodiversity is of high priority [136,142,143].
Despite the insights of this study, there are several limitations and uncertainties that should be addressed in future research. While the MaxEnt algorithm has been proved to produce reliable distribution predictions, its performance depends on user defined settings (e.g., features and regularization parameters) and the quality of presence-only data [144,145]. Consequently, interpretations should be made with caution, especially under future climate conditions. Additionally, this study did not account for potential species interactions (e.g., predator-prey dynamics and competition) [146,147], ontogenetic shifts [146] or evolutionary responses such as genetic or phenotypic plasticity [148,149], which may affect future species distributions. Moreover, the ongoing increased spread of invasive species in the Mediterranean may further modify biotic interactions and community structure, potentially reshaping species distribution patterns under climate change [134]. Future research should incorporate these factors, as well as apply alternative modelling methods, to enhance the understanding of the factors driving species distributions.
To conclude, this study shows that climate change will likely induce major shifts in distributions of non-targeted demersal fish species in the Mediterranean, with varying impacts depending on the species and emission scenarios and independent of habitat type. These findings highlight the importance of considering climate change impacts in marine conservation and fisheries management, particularly for often understudied non-targeted species [9,10,13]. By integrating these climate projections into management strategies, this study provides valuable insights and contributes to the development of more effective strategies for the protection of Mediterranean marine ecosystems in the face of ongoing environmental change.
Supporting information
S1 Table. Selected species for habitat suitability modeling, along with their taxonomy, habitat, importance in fisheries (discard or by-catch), trophic level (FoodTroph retrieved from FishBase), trophic guild (defined using a threshold of 3.5) and the number of presences used for the analysis after the thinning procedure.
https://doi.org/10.1371/journal.pclm.0000838.s001
(DOCX)
S2 Table. Model evaluation metrics for each species, grouped by taxonomic order- suborder and family, along with the feature class and regularization multiplier used for MaxEnt model fitting.
In the column of feature class L stands for linear, H for hinge, Q for quadratic, LQ for linear-quadratic, LQH linear-quadratic-hinge and LQHP for linear-quadratic-hinge-product features.
https://doi.org/10.1371/journal.pclm.0000838.s002
(DOCX)
S3 Table. Permutation importance of environmental layers for each species’ habitat suitability distribution model, presented by their taxonomic classification (order-suborder-family).
BWS refers to bottom water speed, SBS to sea bottom salinity, CHL to the concentration of chlorophyll a on the surface, DO to concentration of dissolved oxygen at the bottom, and SBT to sea bottom temperature. Minor deviations from 100% (e.g., 99.98–100.02) are due to rounding of percentage values for presentation and do not affect the underlying model outputs.
https://doi.org/10.1371/journal.pclm.0000838.s003
(DOCX)
S4 Table. Relative percentage changes in predicted habitat suitability for each species under climate change scenarios (SSP1-1.9, SSP2-4.5 and SSP5-8.5) for the periods 2040–2050 and 2090–2100, compared with present day suitability, presented together with their taxonomic classification (order-suborder-family).
https://doi.org/10.1371/journal.pclm.0000838.s004
(DOCX)
S5 Table. Predicted distribution changes for species between present-day and the period 2090–2100 under SSP5-8.5 scenario, presented together with their taxonomic classification (order-suborder-family).
Loss cells express the number of grid cells lost under climate change scenario, gain cells the number of cells gained, stable cells the number of cells that remained stable and net change the difference between loss and gain cells. Distance, expressed in km, indicates the extent of the displacement of the centroid at the future scenario, compared to the present-day and direction presents the dominant direction of the centroid.
https://doi.org/10.1371/journal.pclm.0000838.s005
(DOCX)
S6 Table. Summary of projected gains, losses and stable suitable grid cells by habitat type under each SSP scenario.
Loss, gain and stable suitable cells were calculated from species-level binary suitability maps and aggregated across species within each habitat type. Percentage change is expressed relative to the total number of present suitable cells.
https://doi.org/10.1371/journal.pclm.0000838.s006
(DOCX)
S1 Fig. Maps presenting the observed and the predicted distributions of each species.
The left map presents the observed records of each species (red points), and the right map presents the predicted distribution of each species at present-day based on MaxEnt modelling (color scales indicate the predicted probabilities of occurence). Basemap derived from Natural Earth (https://www.naturalearthdata.com).
https://doi.org/10.1371/journal.pclm.0000838.s007
(PDF)
S2 Fig. Future species distribution projections under climate change scenarios (SSP1-1.9, SSP2-4.5 and SSP5-8.5) for each species.
Bar plots indicate the relative percentage differences between the present-day situation and future scenarios at the threshold of maxSSS. The left column of the plot represents habitat suitability for future period 2040–2050, while the right column represents habitat suitability for the period 2090–2100. Blue color refers to SSP1-1.9, green to SSP2-4.5 and red to SSP5-8.5 climate change scenario. Color scales on the maps represent the future predicted probabilities from the MaxEnt modelling. Basemap derived from Natural Earth (https://www.naturalearthdata.com).
https://doi.org/10.1371/journal.pclm.0000838.s008
(PDF)
S3 Fig. Overall predicted possibility of habitat suitability for each habitat type under three climate change scenarios (SSP1-1.9, SSP2-4.5 and SSP5-8.5).
Panels: (A) reef-associated, (B) pelagic and (C) other habitat types. Bar plots indicate the relative percentage differences between present situation and future projections. The left column of the plot represents habitat suitability for future period 2040–2050, while the right column represents habitat suitability for the period 2090–2100. Basemap derived from Natural Earth (https://www.naturalearthdata.com).
https://doi.org/10.1371/journal.pclm.0000838.s009
(PDF)
S4 Fig. Changes in the number of suitable habitat grids across trophic guilds under three climate change scenarios (SSP1-1.9, SSP2-4.5 and SSP5-8.5).
Comparisons are made between the present-day and two future time periods (2040–2050 and 2090–2100). Trophic guilds were defined using FishBase trophic level (FoodTroph) with a threshold of 3.5 (<3.5 and ≥3.5). The box represents the interquartile range (50% of values), the solid black line indicates the median, whiskers extend to values within 1.5 times the interquartile range, and points outside this range represent potential outliers. Only species with available trophic information (n = 90; < 3.5: n = 43; ≥ 3.5: n = 47) were included in the post hoc aggregation.
https://doi.org/10.1371/journal.pclm.0000838.s010
(TIF)
References
- 1. Valente S, Colloca F. A dataset of thermal preferences for Mediterranean demersal and benthic macrofauna. Sci Data. 2024;11(1):314. pmid:38538659
- 2. Coll M, Piroddi C, Steenbeek J, Kaschner K, Ben Rais Lasram F, Aguzzi J, et al. The biodiversity of the Mediterranean Sea: estimates, patterns, and threats. PLoS One. 2010;5(8):e11842. pmid:20689844
- 3. Osio GC, Orio A, Millar CP. Assessing the vulnerability of Mediterranean demersal stocks and predicting exploitation status of un-assessed stocks. Fish Res. 2015;171:110–21.
- 4. Kaiser M, Austen MC, Ojaveer H. European biodiversity action plan for fisheries: issues for non-target species. Fish Res. 2004;69:1–6.
- 5. Stergiou K, Tsikliras A. Market and regulation in American urban sprawl: a different view. Ethics Sci Environ Polit. 2006;6:15–7.
- 6. Dimarchopoulou D, Stergiou KI, Tsikliras AC. Gap analysis on the biology of Mediterranean marine fishes. PLoS One. 2017;12(4):e0175949. pmid:28407022
- 7.
Revill A. Survival of discarded fish, a rapid review of studies on discard survival rates. 615631. 2012.
- 8. Tsagarakis K, Nikolioudakis N, Papandroulakis N, Vassilopoulou V, Machias A. Preliminary assessment of discards survival in a multi-species Mediterranean bottom trawl fishery. J Appl Ichthyol. 2018;34:842–9.
- 9. Mace GM, Norris K, Fitter AH. Biodiversity and ecosystem services: a multilayered relationship. Trends Ecol Evol. 2012;27(1):19–26. pmid:21943703
- 10. Young HS, McCauley DJ, Galetti M, Dirzo R. Patterns, causes, and consequences of anthropocene defaunation. Annu Rev Ecol Evol Syst. 2016;47:333–58.
- 11. de Mazancourt C, Isbell F, Larocque A, Berendse F, De Luca E, Grace JB, et al. Predicting ecosystem stability from community composition and biodiversity. Ecol Lett. 2013;16(5):617–25. pmid:23438189
- 12. Tilman D, Isbell F, Cowles JM. Biodiversity and Ecosystem Functioning. Annu Rev Ecol Evol Syst. 2014;45:471–93.
- 13. Duffy JE, Lefcheck JS, Stuart-Smith RD, Navarrete SA, Edgar GJ. Biodiversity enhances reef fish biomass and resistance to climate change. Proc Natl Acad Sci U S A. 2016;113(22):6230–5. pmid:27185921
- 14. Horn M. Diversity and ecological roles of noncommercial fishes in California marine habitats. Calif Coop Ocean Fish Investig Reports. 1980;21.
- 15. Cömert N, Deniz T, Göktürk D. Sustainable Fisheries and Non-Target Species Management: A Seasonal and Depth-Based Study in the Deep-Sea Fisheries of Antalya Bay. Sustainability. 2025;17(11):5040.
- 16.
Baran E. The importance of non-commercial fish [Internet]. Safran P, editor. Fish. Aquac. Towar. Sustain. Aquat. Living Resour. Manag. Oxford, UK: UNESCO Encyclopedia of Life Support Systems; 2002. https://hdl.handle.net/20.500.12348/2266
- 17. Tsagarakis K, Palialexis A, Vassilopoulou V. Mediterranean fishery discards: review of the existing knowledge. ICES Journal of Marine Science. 2013;71(5):1219–34.
- 18. Gilman E, Perez Roda A, Huntington T, Kennelly SJ, Suuronen P, Chaloupka M, et al. Benchmarking global fisheries discards. Sci Rep. 2020;10(1):14017. pmid:32820238
- 19. Säterberg T, Sellman S, Ebenman B. High frequency of functional extinctions in ecological networks. Nature. 2013;499(7459):468–70. pmid:23831648
- 20. Tsikliras AC, Touloumis K, Pardalou A, Adamidou A, Keramidas I, Orfanidis GA. Status and exploitation of 74 un-assessed demersal fish and invertebrate stocks in the Aegean Sea (Greece) using abundance and resilience. Frontiers in Marine Science. 2021;7.
- 21. Stergiou K. Selectivity experiments in the NE Mediterranean: the effect of trawl codend mesh size on species diversity and discards. ICES Journal of Marine Science. 1997;54(5):774–86.
- 22. Adamidou A, Touloumis K, Koutrakis M, Tsikliras AC. Estimation of selectivity parameters for target and bycatch fishes of the trammel net fisheries in the northern Aegean Sea (eastern Mediterranean Sea). AIeP. 2023;53:65–80.
- 23. Brander K. Climate and current anthropogenic impacts on fisheries. Climatic Change. 2012;119(1):9–21.
- 24. Cheung WWL, Lam VWY, Sarmiento JL, Kearney K, Watson R, Pauly D. Projecting global marine biodiversity impacts under climate change scenarios. Fish and Fisheries. 2009;10(3):235–51.
- 25. Perry AL, Low PJ, Ellis JR, Reynolds JD. Climate change and distribution shifts in marine fishes. Science. 2005;308(5730):1912–5. pmid:15890845
- 26. Elith J, Leathwick JR. Species distribution models: ecological explanation and prediction across space and time. Annu Rev Ecol Evol Syst. 2009;40:677–97.
- 27. Guisan A, Thuiller W. Predicting species distribution: offering more than simple habitat models. Ecol Lett. 2005;8(9):993–1009. pmid:34517687
- 28. Damalas D, Maravelias CD, Katsanevakis S, Karageorgis AP, Papaconstantinou C. Seasonal abundance of non-commercial demersal fish in the eastern Mediterranean Sea in relation to hydrographic and sediment characteristics. Estuarine, Coastal and Shelf Science. 2010;89(1):107–18.
- 29. Despoti S, Stergiou KI, Machias A, Vassilopoulou V, Tsagarakis K, Valavanis V. Assessing the spatial distribution of five non-commercial fish species in the Aegean Sea (Greece, eastern Mediterranean Sea) based on discards data. Reg Stud Mar Sci. 2021;44:101736.
- 30. Lek S, Guégan JF. Artificial neural networks as a tool in ecological modelling, an introduction. Ecol Modell. 1999;120:65–73.
- 31. Elith J, Leathwick JR, Hastie T. A working guide to boosted regression trees. J Anim Ecol. 2008;77(4):802–13. pmid:18397250
- 32. Cutler DR, Edwards TC Jr, Beard KH, Cutler A, Hess KT, Gibson J, et al. Random forests for classification in ecology. Ecology. 2007;88(11):2783–92. pmid:18051647
- 33. Phillips SJ, Anderson RP, Schapire RE. Maximum entropy modeling of species geographic distributions. Ecol Modell. 2006;190:231–59.
- 34. Ben Lamine E, Schickele A, Goberville E, Beaugrand G, Allemand D, Raybaud V. Expected contraction in the distribution ranges of demersal fish of high economic value in the Mediterranean and European Seas. Sci Rep. 2022;12(1):10150. pmid:35710852
- 35. Coro G, Bove P, Kesner-Reyes K. Ensemble Ecological Niche Models and Biodiversity Index for 2019 of 96 European Marine Species based on Ecological Niche Models developed with Artificial Neural Networks, Maximum Entropy, and Support Vector Machines at 0.1° Resolution. 2022.
- 36. Effrosynidis D, Tsikliras A, Arampatzis A, Sylaios G. Species distribution modelling via feature engineering and machine learning for pelagic fishes in the Mediterranean Sea. Applied Sciences. 2020;10:8900.
- 37. Panzeri D, Russo T, Arneri E, Carlucci R, Cossarini G, Isajlović I, et al. Identifying priority areas for spatial management of mixed fisheries using ensemble of multi‐species distribution models. Fish and Fisheries. 2023;25(2):187–204.
- 38. Elith J, Graham H, Anderson PF, Dudík M, Ferrier S, Guisan A. Novel methods improve prediction of species’ distributions from occurrence data. Ecography. 2006;29:129–51.
- 39. Monk J, Ierodiaconou D, Versace V, Bellgrove A, Harvey E, Rattray A. Habitat suitability for marine fishes using presence-only modelling and multibeam sonar. Mar Ecol Prog Ser. 2010;420:157–74.
- 40. Pearson RG, Raxworthy CJ, Nakamura M, Townsend Peterson A. Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. J Biogeogr. 2007;34:102–17.
- 41. Radomski T, Beamer D, Babineau A, Wilson C, Pechmann J, Kozak KH. Finding what you don’t know: Testing SDM methods for poorly known species. Diversity and Distributions. 2022;28:1769–80.
- 42. Sheth SN, Jiménez I, Angert AL. Identifying the paths leading to variation in geographical range size in western North American monkeyflowers. J Biogeogr. 2014;41:2344–56.
- 43. Jones MC, Dye SR, Pinnegar JK, Warren R, Cheung WWL. Modelling commercial fish distributions: Prediction and assessment using different approaches. Ecol Modell. 2012;225:133–45.
- 44. Melo-Merino SM, Reyes-Bonilla H, Lira-Noriega A. Ecological niche models and species distribution models in marine environments: A literature review and spatial analysis of evidence. Ecol Modell. 2020;415:108837.
- 45. Silva C, Leiva F, Lastra J. Predicting the current and future suitable habitat distributions of the anchovy (Engraulis ringens) using the Maxent model in the coastal areas off central‐northern Chile. Fish Oceanogr. 2019;28:171–82.
- 46. Basen T, Ros A, Chucholl C, Oexle S, Brinker A. Who will be where: Climate driven redistribution of fish habitat in southern Germany. PLOS Clim. 2022;1(5):e0000006.
- 47. Masoumi AH, Esmaeili HR, Khosravi R, Gholamhosseini A, Korkmaz M, Jeppesen E. Species on the move: Impacts of climate change on the spatial range of endemic fishes of the eco-sensitive semi-arid area of the Arabian Peninsula. Sci Total Environ. 2024;947:174095. pmid:38914333
- 48. Tims AR, Saupe EE. Forecasting climate‐driven habitat changes for Australian freshwater fishes. Divers Distrib. 2023;29:641–53.
- 49. Poloczanska ES, Brown CJ, Sydeman WJ, Kiessling W, Schoeman DS, Moore PJ. Nature Climate Change. 2013;3:919–25.
- 50. Sailley SF, Catalan IA, Batsleer J, Bossier S, Damalas D, Hansen C, et al. Multiple Models of European Marine Fish Stocks: Regional Winners and Losers in a Future Climate. Glob Chang Biol. 2025;31(4):e70149. pmid:40177781
- 51. Hillebrand H, Brey T, Gutt J, Hagen W, Metfies K, Meyer B, et al. Climate change: warming impacts on marine biodiversity. Handbook of Marine Environmental Protection. Cham: Springer International Publishing. 2018. p. 353–73.
- 52. Worm B, Lotze HK. Marine Biodiversity and Climate Change. Climate Change. Elsevier. 2016. p. 195–212.
- 53. Beaugrand G, Edwards M, Raybaud V, Goberville E, Kirby RR. Future vulnerability of marine biodiversity compared with contemporary and past changes. Nature Climate Change. 2015;5:695–701.
- 54. Keramidas I, Dimarchopoulou D, Kokkos N, Sylaios G, Tsikliras A. Temporal ecotrophic impacts of fisheries and climate change in the Aegean Sea. Mar Ecol Prog Ser. 2024;749:19–45.
- 55. Stock CA, Dunne JP, John JG. Drivers of trophic amplification of ocean productivity trends in a changing climate. Biogeosciences. 2014;11(24):7125–35.
- 56. Blanchard JL, Jennings S, Holmes R, Harle J, Merino G, Allen JI, et al. Potential consequences of climate change for primary production and fish production in large marine ecosystems. Philos Trans R Soc Lond B Biol Sci. 2012;367(1605):2979–89. pmid:23007086
- 57. Cheung WWL, Jones MC, Reygondeau G, Stock CA, Lam VWY, Frölicher TL. Structural uncertainty in projecting global fisheries catches under climate change. Ecol Modell. 2016;325:57–66.
- 58. Marshall CE, Glegg GA, Howell KL. Species distribution modelling to support marine conservation planning: The next steps. Marine Policy. 2014;45:330–2.
- 59. Poursanidis D, Kougioumoutzis K, Minasidis V, Chartosia N, Kletou D, Kalogirou S. Uncertainty in Marine Species Distribution Modelling: Trying to Locate Invasion Hotspots for Pterois miles in the Eastern Mediterranean Sea. J Mar Sci Eng. 2022;10:729.
- 60. Solanou M, Valavanis VD, Karachle PK, Giannoulaki M. Looking at the Expansion of Three Demersal Lessepsian Fish Immigrants in the Greek Seas: What Can We Get from Spatial Distribution Modeling?. Diversity. 2023;15(6):776.
- 61. Albouy C, Guilhaumon F, Leprieur F, Lasram FB, Somot S, Aznar R. Journal of Biogeography. 2013;40:534–47.
- 62. Bertrand JA, Gil de Sola L, Papaconstantinou C, Relini G, Souplet A. The general specifications of the MEDITS surveys. Sci Mar. 2002;66:9–17.
- 63.
MEDITS Working Group. MEDITS Survey – Instruction Manual. 2017.
- 64. Machias A, Vassilopoulou V, Vatsos D, Bekas P, Kallianiotis A, Papaconstantinou C. Bottom trawl discards in the northeastern Mediterranean Sea. Fish Res. 2001;53:181–95.
- 65. Aiello‐Lammens ME, Boria RA, Radosavljevic A, Vilela B, Anderson RP. spThin: an R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography. 2015;38:541–5.
- 66. Freitas C, Villegas-Ríos D, Moland E, Olsen EM. Sea temperature effects on depth use and habitat selection in a marine fish community. J Anim Ecol. 2021;90(7):1787–800. pmid:33844859
- 67. Drazen JC, Seibel BA. Depth‐related trends in metabolism of benthic and benthopelagic deep‐sea fishes. Limnol Oceanogr. 2007;52:2306–16.
- 68. Gallo N, Beckwith M, Wei C, Levin L, Kuhnz L, Barry J. Dissolved oxygen and temperature best predict deep-sea fish community structure in the Gulf of California with climate change implications. Mar Ecol Prog Ser. 2020;637:159–80.
- 69. Lehtonen TK, Wong BBM, Kvarnemo C. Effects of salinity on nest-building behaviour in a marine fish. BMC Ecol. 2016;16:7. pmid:26928449
- 70. Holliday FGT. The effects of salinity on the eggs and larvae of teleosts. 1969. p. 293–311.
- 71. Ware DM, Thomson RE. Bottom-up ecosystem trophic dynamics determine fish production in the Northeast Pacific. Science. 2005;308(5726):1280–4. pmid:15845876
- 72. Kostylev V, Todd B, Fader G, Courtney R, Cameron G, Pickrill R. Benthic habitat mapping on the Scotian Shelf based on multibeam bathymetry, surficial geology and sea floor photographs. Mar Ecol Prog Ser. 2001;219:121–37.
- 73. Reiss H, Greenstreet S, Sieben K, Ehrich S, Piet G, Quirijns F, et al. Effects of fishing disturbance on benthic communities and secondary production within an intensively fished area. Mar Ecol Prog Ser. 2009;394:201–13.
- 74. Assis J, Fernández Bejarano SJ, Salazar VW, Schepers L, Gouvêa L, Fragkopoulou E. Bio‐ORACLE v3.0. Pushing marine data layers to the CMIP6 Earth System Models of climate change research. Global Ecology and Biogeography. 2024;33.
- 75. Tyberghein L, Verbruggen H, Pauly K, Troupin C, Mineur F, De Clerck O. Bio‐ORACLE: a global environmental dataset for marine species distribution modelling. Global Ecology and Biogeography. 2012;21:272–81.
- 76.
R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. 2022.
- 77.
Hijmans RJ, Phillips S, Leathwick J, Elith J. dismo: Species Distribution Modeling. CRAN Contrib. Packag. 2024.
- 78. Kass JM, Muscarella R, Galante PJ, Bohl CL, Pinilla‐Buitrago GE, Boria RA. ENMeval 2.0: Redesigned for customizable and reproducible modeling of species’ niches and distributions. Methods Ecol Evol. 2021;12:1602–8.
- 79. Phillips SJ, Dudík M. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography. 2008;31:161–75.
- 80. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143(1):29–36. pmid:7063747
- 81. Boyce MS, Vernier PR, Nielsen SE, Schmiegelow FK. Evaluating resource selection functions. Ecol Modell. 2002;157:281–300.
- 82. Allouche O, Tsoar A, Kadmon R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J Appl Ecol. 2006;43:1223–32.
- 83. Cohen J. A Coefficient of Agreement for Nominal Scales. Educ Psychol Meas. 1960;20:37–46.
- 84. Somers RH. A new asymmetric measure of association for ordinal variables. Am Sociol Rev. 1962;27:799.
- 85.
Anderson RP, Martínez-Meyer E, Nakamura M, Araújo MB, Peterson AT, Soberón J. Ecological niches and geographic distributions. Princeton University Press. 2011.
- 86. Liu C, White M, Newell G. Selecting thresholds for the prediction of species occurrence with presence‐only data. J Biogeogr. 2013;40:778–89.
- 87. Liu C, Newell G, White M. On the selection of thresholds for predicting species occurrence with presence-only data. Ecol Evol. 2015;6(1):337–48. pmid:26811797
- 88. Riahi K, van Vuuren DP, Kriegler E, Edmonds J, O’Neill BC, Fujimori S, et al. The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Glob Environ Chang. 2017;42:153–68.
- 89.
Froese R, Pauly D. FishBase 2000: Concepts Designs and Data Sources. ICLARM. 2000.
- 90. Graham NAJ, McClanahan TR, MacNeil MA, Wilson SK, Cinner JE, Huchery C. Human Disruption of Coral Reef Trophic Structure. Current Biology. 2017;27:231–6.
- 91. Lima ARA, Baltazar-Soares M, Garrido S, Riveiro I, Carrera P, Piecho-Santos AM, et al. Forecasting shifts in habitat suitability across the distribution range of a temperate small pelagic fish under different scenarios of climate change. Sci Total Environ. 2022;804:150167. pmid:34798731
- 92. Peterson AT, Ortega-Huerta MA, Bartley J, Sánchez-Cordero V, Soberón J, Buddemeier RH, et al. Future projections for Mexican faunas under global climate change scenarios. Nature. 2002;416(6881):626–9. pmid:11948349
- 93. Thuiller W, Midgley GF, Hughes GO, Bomhard B, Drew G, Rutherford MC, et al. Endemic species and ecosystem sensitivity to climate change in Namibia. Global Change Biology. 2006;12:759–76.
- 94. Moullec F, Velez L, Verley P, Barrier N, Ulses C, Carbonara P. Capturing the big picture of Mediterranean marine biodiversity with an end-to-end model of climate and fishing impacts. Prog Oceanogr. 2019;178:102179.
- 95. Jennings S, Brander K. Predicting the effects of climate change on marine communities and the consequences for fisheries. J Mar Syst. 2010;79:418–26.
- 96. Ben Rais Lasram F, Guilhaumon F, Mouillot D. Fish diversity patterns in the Mediterranean Sea: deviations from a mid-domain model. Mar Ecol Prog Ser. 2009;376:253–67.
- 97. Ben Rais Lasram F, Guilhaumon F, Albouy C, Somot S, Thuiller W, Mouillot D. The Mediterranean Sea as a “cul-de-sac” for endemic fishes facing climate change. Global Change Biology. 2010;16:3233–45.
- 98. Urban MC. Accelerating extinction risk from climate change. Science. 2015;348(6234):571–3.
- 99. Marbà N, Jordà G, Agustí S, Girard C, Duarte CM. Footprints of climate change on Mediterranean Sea biota. Front Mar Sci. 2015;2.
- 100.
Peck M, Pinnegar JK. Climate change impacts, vulnerabilities and adaptations: North Atlantic and Atlantic Arctic marine fisheries. In: Barange M, Bahri T, Beveridge MCM, Cochrane KL, Funge-Smith S, Poulain F, editors. Impacts Clim Chang Fish Aquac. Rome, Italy: FAO. 2019. p. 87–111.
- 101. Pinsky ML, Eikeset AM, McCauley DJ, Payne JL, Sunday JM. Greater vulnerability to warming of marine versus terrestrial ectotherms. Nature. 2019;569(7754):108–11. pmid:31019302
- 102. Burrows MT, Schoeman DS, Buckley LB, Moore P, Poloczanska ES, Brander KM. The pace of shifting climate in marine and terrestrial ecosystems. Science. 2011;334:652–5.
- 103. Pecl GT, Araújo MB, Bell JD, Blanchard J, Bonebrake TC, Chen I-C, et al. Biodiversity redistribution under climate change: Impacts on ecosystems and human well-being. Science. 2017;355(6332):eaai9214. pmid:28360268
- 104. Myers EMV, Anderson MJ, Eme D, Liggins L, Roberts CD. Changes in key traits versus depth and latitude suggest energy-efficient locomotion, opportunistic feeding and light lead to adaptive morphologies of marine fishes. J Anim Ecol. 2020;89(2):309–22. pmid:31646627
- 105. Orfanidis GA, Touloumis K, Koutrakis E, Tsikliras AC. Fish assemblages along a bathymetric gradient in the northern Aegean Sea: An ecomorphological approach. Deep Sea Research Part I: Oceanographic Research Papers. 2024;207:104223.
- 106. Chaikin S, Dubiner S, Belmaker J. Cold‐water species deepen to escape warm water temperatures. Global Ecology and Biogeography. 2022;31:75–88.
- 107. Deutsch C, Ferrel A, Seibel B, Pörtner H-O, Huey RB. Ecophysiology. Climate change tightens a metabolic constraint on marine habitats. Science. 2015;348(6239):1132–5. pmid:26045435
- 108. Penn JL, Deutsch C, Payne JL, Sperling EA. Temperature-dependent hypoxia explains biogeography and severity of end-Permian marine mass extinction. Science. 2018;362(6419):eaat1327. pmid:30523082
- 109. Giannoulaki M, Schismenou E, Pyrounaki M-M, Tsagarakis K. Habitat Characterization and Migrations. Biology and Ecology of Sardines and Anchovies. CRC Press. 2014. p. 191–241.
- 110. Schickele A, Goberville E, Leroy B, Beaugrand G, Hattab T, Francour P, et al. European small pelagic fish distribution under global change scenarios. Fish and Fisheries. 2020;22(1):212–25.
- 111. Beck MW, Heck KL, Able KW, Childers DL, Eggleston DB, Gillanders BM, et al. The Identification, Conservation, and Management of Estuarine and Marine Nurseries for Fish and Invertebrates. BioScience. 2001;51(8):633.
- 112. Monfort T, Cheminée A, Bianchimani O, Drap P, Puzenat A, Thibaut T. The three-dimensional structure of Mediterranean shallow rocky reefs: use of photogrammetry-based descriptors to assess its influence on associated teleost assemblages. Frontiers in Marine Science. 2021;8.
- 113. Finkel ZV. Global Environmental Change. Springer Netherlands. 2014. p. 117–24.
- 114. Friedland KD, Langan JA, Large SI, Selden RL, Link JS, Watson RA, et al. Changes in higher trophic level productivity, diversity and niche space in a rapidly warming continental shelf ecosystem. Sci Total Environ. 2020;704:135270. pmid:31818590
- 115. Sheaves M, Baker R, Nagelkerken I, Connolly RM. True Value of Estuarine and Coastal Nurseries for Fish: Incorporating Complexity and Dynamics. Estuaries and Coasts. 2014;38(2):401–14.
- 116. Cheminée A, Le Direach L, Rouanet E, Astruch P, Goujard A, Blanfuné A, et al. All shallow coastal habitats matter as nurseries for Mediterranean juvenile fish. Sci Rep. 2021;11(1):14631. pmid:34272431
- 117. Lotze HK, Lenihan HS, Bourque BJ, Bradbury RH, Cooke RG, Kay MC, et al. Depletion, degradation, and recovery potential of estuaries and coastal seas. Science. 2006;312:1806–9.
- 118. Halpern BS, Frazier M, Afflerbach J, Lowndes JS, Micheli F, O’Hara C, et al. Recent pace of change in human impact on the world’s ocean. Sci Rep. 2019;9(1):11609. pmid:31406130
- 119. Halpern BS, Walbridge S, Selkoe KA, Kappel CV, Micheli F, D’Agrosa C. A global map of human impact on marine ecosystems. Science. 2008;319:948–52.
- 120. Sunday JM, Bates AE, Dulvy NK. Thermal tolerance and the global redistribution of animals. Nat Clim Chang. 2012;2:686–90.
- 121.
Moyle PB, Cech JJ. Fishes: an introduction to ichthyology [Internet]. 5th ed. Pearson Prentice Hall; 2004. https://books.google.gr/books?id=sZYWAQAAIAAJ
- 122. Brown A, Thatje S. The effects of changing climate on faunal depth distributions determine winners and losers. Glob Chang Biol. 2015;21(1):173–80. pmid:25044552
- 123. Nash RDM, Geffen AJ, Santos RS. The wide-eyed flounder, Bothus podas delaroche, a singular flatfish in varied shallow-water habitats of the azores. Netherlands Journal of Sea Research. 1991;27(3–4):367–73.
- 124. Munday PL, Jones GP, Pratchett MS, Williams AJ. Climate change and the future for coral reef fishes. Fish and Fisheries. 2008;9(3):261–85.
- 125. Pinsky ML, Worm B, Fogarty MJ, Sarmiento JL, Levin SA. Marine Taxa Track Local Climate Velocities. Science. 2013;341:1239–42.
- 126. du Pontavice H, Gascuel D, Reygondeau G, Maureaud A, Cheung WWL. Climate change undermines the global functioning of marine food webs. Glob Chang Biol. 2020;26(3):1306–18. pmid:31802576
- 127. Lotze HK, Tittensor DP, Bryndum-Buchholz A, Eddy TD, Cheung WWL, Galbraith ED, et al. Global ensemble projections reveal trophic amplification of ocean biomass declines with climate change. Proc Natl Acad Sci U S A. 2019;116(26):12907–12. pmid:31186360
- 128. Clavel J, Julliard R, Devictor V. Worldwide decline of specialist species: toward a global functional homogenization?. Front Ecol Environ. 2011;9:222–8.
- 129. Magurran AE, Dornelas M, Moyes F, Gotelli NJ, McGill B. Rapid biotic homogenization of marine fish assemblages. Nat Commun. 2015;6:8405. pmid:26400102
- 130. Pauly D, Christensen V. Primary production required to sustain global fisheries. Nature. 1995;374(6519):255–7.
- 131. Marguin A, Bussotti S, Guidetti P, Rossi F. A systematic review of fishing impacts on the trophic level of fish populations and assemblages in the Mediterranean Sea. Front Mar Sci. 2025;12.
- 132. Touloumis K, Tsikliras AC, Dimarchopoulou D. The cascading effect of fishing down marine top predators in the northeastern Mediterranean Sea. Discov Oceans. 2025;2(1).
- 133. Darmaraki S, Somot S, Sevault F, Nabat P, Cabos Narvaez WD, Cavicchia L. Future evolution of marine heatwaves in the Mediterranean Sea. Clim Dyn. 2019;53:1371–92.
- 134. Azzurro E, Sbragaglia V, Cerri J, Bariche M, Bolognini L, Ben Souissi J, et al. Climate change, biological invasions, and the shifting distribution of Mediterranean fishes: A large-scale survey based on local ecological knowledge. Glob Chang Biol. 2019;25(8):2779–92. pmid:31111639
- 135. Dulvy NK, Sadovy Y, Reynolds JD. Extinction vulnerability in marine populations. Fish and Fisheries. 2003;4(1):25–64.
- 136. Rilov G, Fraschetti S, Gissi E, Pipitone C, Badalamenti F, Tamburello L, et al. A fast-moving target: achieving marine conservation goals under shifting climate and policies. Ecol Appl. 2020;30(1):e02009. pmid:31549453
- 137. Lawler JJ. Climate change adaptation strategies for resource management and conservation planning. Ann N Y Acad Sci. 2009;1162:79–98.
- 138. Wilson KL, Tittensor DP, Worm B, Lotze HK. Incorporating climate change adaptation into marine protected area planning. Glob Chang Biol. 2020;26(6):3251–67. pmid:32222010
- 139. Ben Lamine E, Schickele A, Guidetti P, Allemand D, Hilmi N, Raybaud V. Redistribution of fisheries catch potential in Mediterranean and North European waters under climate change scenarios. Sci Total Environ. 2023;879:163055. pmid:36972882
- 140. Cheung WWL, Lam VWY, Sarmiento JL, Kearney K, Watson R, Zeller D. Large‐scale redistribution of maximum fisheries catch potential in the global ocean under climate change. Global Change Biology. 2010;16:24–35.
- 141. Guidetti P. Marine reserves reestablish lost predatory interactions and cause community changes in rocky reefs. Ecol Appl. 2006;16(3):963–76. pmid:16826995
- 142. Keppel G, Van Niel KP, Wardell‐Johnson GW, Yates CJ, Byrne M, Mucina L, et al. Refugia: identifying and understanding safe havens for biodiversity under climate change. Global Ecology and Biogeography. 2011;21(4):393–404.
- 143. Ban SS, Alidina HM, Okey TA, Gregg RM, Ban NC. Identifying potential marine climate change refugia: A case study in Canada’s Pacific marine ecosystems. Glob Ecol Conserv. 2016;8:41–54.
- 144. Elith J, Phillips SJ, Hastie T, Dudík M, Chee YE, Yates CJ. A statistical explanation of MaxEnt for ecologists. Diversity and Distributions. 2011;17:43–57.
- 145. Merow C, Smith MJ, Silander JA. A practical guide to MaxEnt for modeling species’ distributions: what it does, and why inputs and settings matter. Ecogr. 2013;36:1058–69.
- 146. Robinson LM, Elith J, Hobday AJ, Pearson RG, Kendall BE, Possingham HP. Pushing the limits in marine species distribution modelling: lessons from the land present challenges and opportunities. Global Ecology and Biogeography. 2011;20:789–802.
- 147. Zlateva I, Raykov V, Slabakova V, Stefanova E, Stefanova K. Habitat suitability models of five keynote Bulgarian Black Sea fish species relative to specific abiotic and biotic factors. Oceanologia. 2022;64(4):665–74.
- 148. Anderson JJ, Gurarie E, Bracis C, Burke BJ, Laidre KL. Modeling climate change impacts on phenology and population dynamics of migratory marine species. Ecol Modell. 2013;264:83–97.
- 149. Asch RG, Sobolewska J, Chan K. Assessing the reliability of species distribution models in the face of climate and ecosystem regime shifts: Small pelagic fishes in the California Current System. Front Mar Sci. 2022;9.