Benthic habitats of western Mediterranean seamounts: Spatial modelling and environmental drivers

doi:10.21203/rs.3.rs-7211889/v1

Download PDF

Research Article

Benthic habitats of western Mediterranean seamounts: Spatial modelling and environmental drivers

https://doi.org/10.21203/rs.3.rs-7211889/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Seamounts are considered biodiversity hotspots that play a relevant role in deep-sea ecosystems. The LIFE IP INTEMARES project aimed to enhance scientific knowledge of the biodiversity and distribution of benthic species and habitats in the Mallorca Channel seamounts (Balearic Islands, western Mediterranean): Ses Olives, Ausias March and Emile Baudot. The present study develops spatial distribution models, based on the presence/absence of species or assemblages and environmental and fishing variables, to predict and map 11 potential biogenic benthic habitats and pockmarks fields. Data were collected using ROV, photogrammetric sledge, beam trawl and rock dredge. Seafloor was characterized by acoustic and sediment data. Hydrodynamic variables were obtained from models, and the fishing footprint derived from Vessel Monitoring System signals. Maërl/rhodolith beds, mainly overlapping with sponge gardens, were found on the circalittoral detritic sandy bottoms of the summits of the shallowest seamounts Ausias March and Emile Baudot. Sponge and gorgonian gardens were found on coralligenous outcrops and rocky flanks, while smaller sponges and pockmarks were widespread on deeper adjacent bottoms. Notably, the bamboo coral Isidella elongata was recorded on bathyal muds between Ses Olives and Ausias March. These results emphasize the ecological importance of the area and provide cartographic information to support the proposal of the Mallorca Channel seamounts as a Site of Community Interest for their inclusion in the marine Natura 2000 network.

Benthic habitats

Habitat Suitability Models (HSM)

Mapping

Seamounts

Fishing pressure

Western Mediterranean

The conservation of marine species and ecosystems is a priority in the Mediterranean, a region recognized as one of the first 25 Global Biodiversity Hotspots (Myers et al., 2000). Seamounts are isolated underwater topographical highs present in both oceanic domains and continental margins, which are also considered as hotspots of biological activity and biodiversity (Clark et al., 2012). In the Mediterranean, seamounts span a broad depth range and comprise heterogeneous habitat types, some of them being internationally recognized as Vulnerable Marine Ecosystems (VME) (FAO, 2009). However, their scientific knowledge is marked by large gaps and an asymmetry between the number of geological studies and biological ones (Würtz & Rovere, 2015).

The Mallorca Channel is located in the Algerian sub-basin, south of the Balearic promontory serving as a seaway between the continental shelves of the Mallorca-Menorca and Ibiza-Formentera islands. Its hydrodynamics are mainly affected by density gradients and the influence of Atlantic waters, which are warmer and less saline than Mediterranean waters (Pinot et al., 1995, 2002). This area is distinguished by its remarkable geomorphological diversity, featuring seamounts, scarps, depressions, pockmarks and volcanic cones (Acosta et al., 2003; Vázquez et al., 2015).

The LIFE + INDEMARES project (https://www.indemares.es/), carried out in the Spanish marine waters between 2009 and 2014, targeted 10 large marine areas which were subsequently designated as Site of Community Importance (SCI). This commitment is currently reflected in the protection of almost the 13% of the marine environment in Spain, through the Spanish network of MPAs (UNEP-WCMC, 2025). The follow up project, LIFE IP INTEMARES (https://intemares.es/en/), developed between 2017 and 2024, aimed to complete these conservation efforts by enhancing the protection and management of those areas and incorporating new ones to the marine Natura 2000 network. Within this context, the seamounts of the Mallorca Channel Ses Olives (SO), Ausias March (AM) and Emile Baudot (EB), as well as their adjacent bottoms, have been studied. The first investigations on these seamounts analyzed their geomorphology and geodynamics (Acosta et al., 2001, 2004) and their benthic species and habitats (Marín et al., 2011; Aguilar et al., 2011; Maldonado et al., 2015; Mastrototaro et al., 2017), highlighting their high ecological value. In fact, the inclusion of this area within Natura 2000 network was proposed more than a decade ago by OCEANA, an international non-governmental organization dedicated to ocean conservation (https://oceana.org/) (Marín et al., 2011).

The seamounts SO, AM and EB rise 375 m, 264 m, and 600 m above the surrounding seafloor, with their summits at 225 m, 86 m, and 94 m, respectively, extending 10–17 km in length. SO and AM are of tectonic origin, being composed of carbonated materials, while EB is of volcanic origin, (Massutí et al., 2022a). The surrounding seafloor is part of the middle slope and corresponds to the underwater top of the Balearic Promontory (Fig. 1). The steep flanks of the seamounts reach slopes of up to 40°, while the eastern flank of EB transitions into the EB escarpment, characterized by small canyons with predominant slopes of up to 55°, which connects the area with the deepest western Mediterranean (> 2500 m). Predominating unconsolidated sedimentary substrates, vary from medium to coarse bioclastic sands at the summits of the seamounts to sandy-mud and fine sands enriched with bioclastic material at greater depths. Hard substrates are confined to specific geomorphological features, such as the summits and flanks of the seamounts, escarpments and structural ridges (Massutí et al., 2022a).

Massutí et al. (2022a) highlighted the biodiversity of these seamounts, as evidenced by an inventory of more than 740 different species or taxa, most of them sponges (26%), followed by mollusks (17%), crustaceans (16%), teleosts fishes (16%), echinoderms (9%) and cnidarians (6%). Some of these species are protected by international, European and/or Spanish regulations. Others were new to science or new records in the study area, the western basin or even the whole Mediterranean Sea (Ordines et al., 2019, 2024; Díaz et al., 2021, 2024a, 2024b, 2024c).

Up to 31 benthic habitats types were inventoried in the area (Massutí et al. 2022a), some of them corresponding to Habitats of Community Interest (HCI). These are characterized by sessile, long-lived, habitat-forming species that build three-dimensional structures on the seabed. Such structures increase habitat heterogeneity, generate new biotopes, providing shelter from predators or food resources, facilitate epifaunal colonization, and support higher species richness and biodiversity (Jones et al., 1994; Buhl-Mortensen et al., 2010; Victorero et al., 2018). They are also essential for enhancing ecosystem functioning and services, reproduction and recovery of fish stocks, as well as the persistence of rare and endangered species (Bologna, 2006; Buhl-Mortensen et al., 2010; Cerrano et al., 2010; de la Torriente et al., 2020).

Some demersal fisheries operate in the Mallorca Channel, mainly focused to deep-water decapod crustaceans of high economic value. Bottom trawling, developed at adjacent bottoms of SO and AM, is the most important fishery, being the red shrimp Aristeus antennatus its main target species (García-Rodríguez & Esteban, 1999). Trap fishery, mainly active at the summits of SO and the flanks of the three seamounts, targets the pandalid shrimp Plesionika edwardsii (García-Rodríguez et al., 2000). Small-scale fleet operates more sporadically using trammel nets at the summit of AM to capture the spiny lobster Palinurus elephas, while commercials and recreational fishing boats use bottom long-lines and hand-lines to capture large sparids and serranids at the summits of the three seamounts (Massutí et al., 2022a).

Habitat suitability models (HSM) (Rowden et al., 2017), are useful tools to predict the potential distribution of benthic habitats or species. They are based on the relationships between observed data and environmental variables, being widely used in the marine environment for conservation and management purposes (Reiss et al., 2015; Sánchez et al., 2017, de la Torriente et al., 2019). Such models are especially valuable in areas where direct sampling is logistically challenging, as they help to infer potential habitat distributions from environmental data. This is particularly relevant in complex environments such as rocky bottoms, where sampling is limited by accessibility (García-Alegre et al., 2014). The outputs of these models, combined with analysis of anthropogenic activities, such as fisheries, can provide essential information for decision-making. A well-implemented zoning approach should allow to combine the conservation of the biodiversity and ecosystems with the sustainability of their living resources and hence the services they provide, ultimately benefiting both biodiversity protection and human livelihoods.

The present study aims to identify the key environmental drivers influencing the distribution of benthic habitats in the Mallorca Channel and adjacent bottoms; and modelling their spatial distribution to produce high-resolution habitat maps. To achieve these objectives, the study integrates presence/absence data of habitat-forming species or assemblages, obtained from various sampling methods, with environmental and anthropogenic variables, to develop models for predicting the suitability of a location for a given benthic habitat. These results will improve the current knowledge on the habitat assemblages of the study area and contribute to the scientific assessment to designate this area as SCI for its inclusion and management within the Natura 2000 network.

2.1. Sampling

This study is mainly based on information acquired from multidisciplinary oceanographic surveys conducted between 2018 and 2020, focused on the geological and biological sampling of the Mallorca Channel. Geophysical techniques included multibeam and parametric echosounders, mapping 4506 km² of seafloor from 86 to 1720 m depth. Surface sediment sampling was mainly conducted using Shipek dredge at 122 stations (86–1062 m depth) and Box-Corer at 20 stations (285–926 m depth). Rock dredge at 54 stations (89–1191 m depth) provided additional geological sampling, as well as samples of benthic species on rocky bottoms. At soft bottoms, the samples of benthic communities were mainly obtained from 85 beam trawl stations at 99–764 m depth. At both rocky and soft bottoms, the sampling of benthic communities was completed with the acquisition of imagery from 48 video transects at 87–708 m depth (13 hours of recording, covering an area of 30066 m²) using the TASIFE photogrammetric sledge and 29 video transects at 89-1162 m depth (52 hours of recording, covering 17322 m²) using the ROV Liropus 2000. Figure 1 shows the location of sampling stations where we obtained data on benthic species. The sampling lines and stations to collect geomorphological and sedimentological data and more details about the sampling of benthic species are explained in Massutí et al. (2022a).

Additionally, information from OCEANA oceanographic surveys between 2006 and 2014 was incorporated into the dataset, including presence/absence data recorded for 20 species during 21 ROV transects (Marín et al., 2011; OCEANA, 2015). Moreover, 4 ROV transects (ROV Liropus 2000) conducted at the EB between 99 and 711 m depth during the LIROBAL_0516 survey were also included in the analysis (Guijarro, 2016).

2.2. Data collection and processing

2.2.1. Biological data: habitat-forming species and communities

For some habitats with more than one structuring species, like sponges and cnidarians, the "assemble first, predict later" approach was applied. This method involves two steps: (i) the identification of distinct species associations or communities; and (ii) the application of predictive models to estimate the spatial distribution of these communities (Moritz et al., 2013; de la Torriente et al., 2019). It was applied separately for sponges and cnidarians, due to their biological differences in structure, physiology and ecological needs. For cnidarians, we modelled two associations identified through cluster analysis (see Supplementary Material; Figure S1 and Table S1), whereas for sponge, we used the associations detected by Díaz et al. (2024a), who identified and characterized sponge assemblages in the study area.

The spatial distribution of 11 potential biogenic habitats based on the classification from the Reference List of Marine Habitats in Spain (Templado et al., 2012) and the corresponding habitat inventory guidelines (Table 1). Additionally, equivalences were established with the European Nature Information System (EUNIS, 2022) and the Habitats Directive (Table S2). Pockmarks fields were mapped based on their presence detected from multibeam and parametric echosounder sampling.

Table 1
Potential benthic habitats modeled in the seamounts of the Mallorca Channel (Balearic Islands, western Mediterranean) and their habitat-forming species (HFS). The code used by habitat is also shown.
Code	Habitat	HFS
H1	Maërl/rhodolith beds	Maërl/Rhodoliths species
H2	Rocky summits of seamounts in the circalittoral zone with Eunicella spp.	Eunicella spp.
H3	Infralittoral and circalittoral detritic platforms of seamounts with rhodoliths dominated by sponges	Foraminospongia balearica
H4	Rocky summits of seamounts in the circalittoral zone with Viminella flagellum	Viminella flagellum
H5	Deep rocky bottoms with aggregations of gorgonians	Cnidaria group E, with Eunicella spp., Acanthogorgia hirsuta, V. flagellum and B. mollis as main species. See Supplementary Material, Figure S1 and Table S1
H6	Bathyal rock with octocorals (Bebrice mollis and Callogorgia verticillata)	Cnidaria group A, with Bebryce mollis and Callogorgia verticillata as main species. See Supplementary Material, Figure S1 and Table S1
H7	Bathyal shelf-edge bottoms with Gryphus vitreus	Gryphus vitreus facies
H8	Cliffs, walls, and rocky slopes of submarine elevations and canyons with sponges	Porifera group sA from Díaz et al., (2024a), with Jaspis, Tretodictyum reiswigi, Heteroxya cf. beauforti, Hamacantha spp., Poecillastra compressa and Phakellia robusta as main species
H9	Bathyal detritic bottoms with sponges	Porifera group sF from Díaz et al., (2024a), with P. compressa, D. inornata, T. muricata, Hamacantha spp., Dragmatella aberrans and D. annexa as main species.
H10	Bathyal muds with Thenea muricata	Porifera group sE from Díaz et al., (2024a), primarily with Thenea muricata, accompanied by Desmacella innornata and Desmacella annexa
H11	Bathyal muds with Isidella elongata	Isidella elongata
H12	Pockmark fields	Pockmarks

2.2.2. Environmental and fishing variables

Up to 30 environmental variables were used to develop the HSM models (Table 2): 8 geomorphological, 7 sedimentological, 12 hydrodynamic and 3 fishing pressure variables.

Table 2
Geomorphological, sedimentological, hydrodynamics and fishing pressure variables used to develop predictive spatial distribution models of potential benthic habitats in the Mallorca Channel seamounts (Balearic Islands, western Mediterranean). D50: median grain size; SD: standard deviation; VMS: Vessel Monitoring by satellite System.
Geomorphology	Sedimentology	Hydrodynamics	Fishing pressure
Depth (m) Slope (º) Backscatter (dB) Rugosity Aspect Eastness Aspect Northness BPI (radii/cells)	Organic matter content (%) Inorganic carbonates (%) Gravel (%) Sand (%) Mud (%) (Silt + Clay) Texture D50 Mode	- Surface currents: Average speed (m/s) Max. speed (m/s) Variability (SD) Direction (º) - Bottom currents: Average speed (m/s) Max. speed (m/s) Variability (SD) Direction (º) - Surface temperature: Average (ºC) Variability (SD) - Bottom temperature: Average (ºC) Variability (SD)	Deep-water trap fleet targeted to Plesionika edwardsi: number of VMS records. Bottom longline fleet: number of VMS records. Bottom trawl fleet: Distance to the nearest trawling fishing ground (m).

The geomorphological and sedimentological variables of the seafloor were derived from the analysis of data and samples collected during the LIFE IP INTEMARES project surveys in the study area, as detailed in Massutí et al. (2022a). The processing of these variables in raster format was carried out using ArcMap v.10.8 from the ArcGIS software. Additionally, the external ArcGIS package Benthic Terrain Modeler (BTM) was utilized to calculate three of the geomorphological variables: rugosity, aspect and the Bathymetric Position Index (BPI). The aspect represents the orientation of the seabed at a given location. To avoid problems related to circular data, this variable was decomposed into two components: Eastness (sin(aspect)) and Northness (cos(aspect)). The BPI assess whether a specific area is topographically elevated or depressed in relation to its surroundings. Its calculation was performed using two distinct scales: (i) fine BPI, with inner and outer radii of 5 and 25 cells, respectively; and (ii) broad BPI, with inner and outer radii of 10 and 50 cells, respectively.

To characterize the hydrodynamic variables, ocean current and temperature data from the WMOP (Western Mediterranean OPerational forecasting system) model outputs were used. This information is available through the Balearic Islands Coastal Observing and Forecasting System services (Tintoré et al., 2013; https://www.socib.es/?seccion=modelling&facility=forecast_system_description). WMOP is a high-resolution (~ 2.6 km) 3D ROMS (Regional Ocean Modelling System) model implemented in the western Mediterranean, with a daily temporal resolution (Juza et al., 2016; Mourre et al., 2018). Values of average and maximum current speed, current variability and direction, as well as average and variability of temperature, were obtained both at the surface and near the seafloor for the period 2016–2023. The area considered covers the Balearic Islands and the nearest coast of the Iberian Peninsula, including not only the Mallorca Channel but also part of the Algerian sub-basin. WMOP model outputs were processed using MATLAB R2016a. The data were interpolated to a grid resolution of 100x100 m, consistent with the model’s grid resolution using the Nearest Neighbour interpolation.

The fishing footprint was estimated from the Vessel Monitoring by satellite System (VMS) during the period 2010–2019 (Massutí et al., 2022b). The VMS database includes records of geographic location, date, time and vessel speed, recorded at two-hour intervals. The main demersal fishing fleets operating on the seamounts were those using deep-water traps and bottom long-lines (Figure S2). To limit the VMS positions to those most likely representing fishing activity, only signals with velocities ≤2 knots were filtered for both traps and long-line fleets, after analyzing their velocity frequency distributions using R software version 4.0.5. After filtering, a total of 123326 VMS signals were retained, 53523 from the deep-water traps fleet and 69803 from the bottom long-line fleet. Final values were derived by summing the VMS signals per grid cell over the entire time period. The resulting dataset was then processed to generate a raster layer with the highest possible precision at a 100 × 100 m resolution.

Although bottom trawling is not conducted on the seamounts themselves, it occurs at the base of AM and in adjacent areas of SO. Trawling on muddy bottoms induces sediment resuspension and increases water turbidity (e.g. Palanques et al., 2001), generating nepheloid layers that can be transported by marine currents, potentially impacting the biogeochemical cycles and ecosystems far from the fishing grounds (Martín et al., 2014). The sediment resuspended in these layers can have a variety of effects, including smothering of feeding and respiratory organs of sessile species, such as sponges and cnidarians (e.g. Bilan, 2023). For that reason, the variable considered for this fishery was the distance to the nearest trawling ground. This variable was calculated using the Euclidean Distance tool from the Spatial Analyst module in ArcGIS, based on a map of trawl fishing grounds around the Balearic Islands developed by Guijarro et al. (2020). In that map, three different fishing grounds currently exploited, had been identified in the vicinity of SO and AM (Massutí et al., 2022a): (i) one located to the east and northeast of Ibiza Island, with its southern part including upper and middle slope bottoms adjacent to SO and AM; and (ii) two located east of Formentera, one including upper slope bottoms and the other middle slope bottoms, both adjacent to AM (Fig. 1).

A consistent spatial resolution of 100 x 100 m was applied to all variables, based on their original resolutions and the sampling methods. For layers with finer resolution, values were averaged using the bilinear interpolation method in ArcGIS 10.8’s Resample tool, while coarser resolution layers were resampled accordingly. All layers were standardized to the same spatial extent and coordinates projection. The resulting raster layers were then converted to ASCII format for integration into the modelling process.

Prior to model development, a correlation analysis was conducted among all explanatory variables to prevent redundancy and collinearity. Both the Spearman’s correlation coefficient and the Variance Inflation Factor (VIF, Zuur et al., 2009) were calculated (Table S3). Variables with a correlation coefficient > 0.7 or a VIF > 5 were excluded. As a result, the following 14 variables were selected: depth, slope, rugosity, backscatter, Eastness, Northness, broad BPI, percentage of organic matter (OM) and percentage of sand in the sediment, maximum bottom current speed, mean bottom temperature, VMS data for trap and bottom longline fleets, and distance to the nearest trawl fishing ground.

2.2.3. Sampling unit

For the photogrammetric sledge and the beam trawl, the sampling unit corresponds to the entire transects, which were conducted on flat and homogeneous sedimentary bottoms, with a narrow depth range. In contrast, for the longer ROV transects, conducted on more heterogeneous bottoms with a broader depth range, a larger variation in environmental conditions and a greater diversity of species and communities, the sampling unit consisted of individual video scenes. These were segments of the ROV footage where a homogeneous habitat type and substrate were observed, characterized by a stable community composition and a consistent presence of species. The rock dredge was the most restrictive sampling method, due to the irregularity of transects, which could introduce bathymetric and geolocation errors. Consequently, transects with a depth range exceeding 50 m were excluded from the analysis.

Presence/absence of habitat-forming species or assemblages was recorded at the midpoint of each sampling units and environmental variables were extracted, with an estimated geolocation error of ± 100 m. The following data have been used for modelling 553 ROV scenes and 48, 85 and 46 transects from photogrammetric sledge, beam trawl and rock dredge, respectively. Additionally, 21 ROV transects were provided by OCEANA.

2.3. Predictive modelling and mapping

Habitat Suitability Models (HSM) were developed to predict the mean presence probability of the habitats. To enhance prediction reliability, two methodologies were applied: Generalized Additive Models (GAMs; Hastie & Tibshirani, 1990) and Maximum Entropy Modelling (MaxEnt; Phillips et al., 2006). They present specific advantages and limitations. GAMs require a larger number of presence records (response variable) and the selection of a subset of the most relevant explanatory variables for each species or group of species. In contrast, MaxEnt can operate with a smaller number of presence records and allows the inclusion of all explanatory variables in the model.

GAM modelling was performed using the functions of the package mgcv in R (Wood, 2011), employing the logit function to fit a binomial distribution. The selection of variables for each GAM model followed a backward selection procedure, eliminating those that were not statistically significant (p < 0.05), based on Akaike's Information Criterion (Akaike, 1973). For MaxEnt models, all non-correlated explanatory variables were included and fitted using the maxnet package in R, which enables flexible feature selection while optimizing model performance (Phillips, 2017).

Model evaluation was carried out within the dismo package in R, applying cross-validation by randomly dividing the presence and absence data into two subgroups: one for training the model (66.7%) and the other for its evaluation (33.3%). This process was repeated 10 times and the Area Under the Curve (AUC) of the Receiver Operating Characteristics (ROC) was calculated to assess model performance (Fielding & Bell, 1997), based on its ability to correctly distinguish between presence and absence.

The relative importance of explanatory variables was assessed for both GAM and MaxEnt models by calculating the contribution of each variable to the total explained deviance by comparing models with and without the variable. To ensure comparability between the two methodologies, the explained deviance in the GAM model was set to 100%.

Each modelling approach generated a probability matrix for the occurrence of habitat-associated species or assemblages per grid cell, which were then converted into binomial maps of presence/absence by applying a threshold determined during the modeling process. To maximize predictive reliability, the threshold that maximized the Kappa statistic (Fielding & Bell, 1997) was selected, which reduces the likelihood of overestimating habitat presence, compared to other thresholds like Prevalence, Specificity and Sensitivity. According to Landis & Koch (1977), Kappa values are interpreted as acceptable (0.2–0.4) to excellent (> 0.75). Finally, spatial autocorrelation of the residuals was analysed using variograms.

In order to maximize predictive reliability, we created an ensemble of GAM and MaxEnt models by overlapping the prediction results from both, considering habitat suitability only in those grid cells where both models agreed. To assess model accuracy, two metrics were calculated: (1) Overlap %, defined as the percentage of the predicted habitat area relative to the total area where both models have predicted habitat presence; and (2) Conformity %, defined as the percentage of overlap between grid cells with concordant results, including cells where both models predicted absence (zeros). The area of each habitat was calculated using both R and the Calculate Geometry Area tool in ArcGIS 10.8.

The identification, mapping, and characterization of pockmarks in the study area is based on the analysis of bathymetric data through geospatial analyses using ArcGIS tools. The mapping process consists of three main stages: (1) data smoothing, (2) pockmark delineation, and (3) extraction of key morphological parameters. To define the pockmarks, the BPI layer was used, applying ring radii values of 5 and 15 in this study. These values allowed the identification of depressions associated with pockmarks based on their relative elevation in comparison to the surrounding seafloor. Following the application of these semi-automated methods, a manual interpretation was conducted to refine the mapping of these morphological features that were not well defined by automatic methods. This process integrated bathymetric data, derived layers, seafloor reflectivity, and sub-bottom acoustic profiles to ensure accurate delineation and characterization of pockmarks.

2.4. Relationships with environmental parameters

A distance-based-Redundancy Analysis (dbRDA) method and stepwise selection was used for modelling the effect of environmental and fishing variables on the distribution of habitats. Unlike multivariate indirect gradient analyses, canonical analyses such as RDA provide the means for conducting direct explanatory analyses in which the association among species can be studied with respect to their common and unique relationships with environmental variables (Peres-Neto et al., 2006).

A presence/absence matrix of habitats per sample was employed. The explanatory variables used were the same than those selected for habitat modelling. Bray-Curtis dissimilarity was used to compute pairwise differences among habitat assemblages across samples. This dissimilarity matrix was subjected to a Principal Coordinates Analysis (PCoA), and the resulting axes were constrained using environmental predictors within a canonical redundancy framework. The analysis was conducted using the CANOCO 5.1 software package (ter Braak & Smilauer, 2018). The significance of the model was assessed using the Monte Carlo permutation-based test (Manly, 1991).

3.1. Predictive models and habitat mapping

All GAM models demonstrated high levels of explained deviance (> 24%) (Table 3). Predictive accuracy, as measured by AUC values, ranged from ‘good’ (0.8–0.89) to ‘excellent’ (0.9-1.0), while Kappa values indicated performance from ‘acceptable-to-good’ (0.4–0.75) to ‘excellent’ (> 0.75), with the exception of habitats H6 and H7. The overlap percentages of habitat distribution areas predicted from GAMs and MaxEnt varied across habitats, with the highest values in H1 (62.1%), H3 (60.8%) and H10 (61.4%). In contrast, H2 (14.5%) and H7 (12.7%) showed the lowest overlap percentages. Conformity percentages were generally high, exceeding 90% for most habitats, with the exception of H7 (86.4%) and H10 (65.9%) (Table 3). These results suggest a strong agreement between models for most habitats, particularly regarding absence areas and general distribution patterns.

Table 3
Summary results of the Generalized Additive Models (GAM) and Maximum entropy models (MaxEnt) applied to develop predictive spatial distribution models of potential benthic habitats in the Mallorca Channel seamounts (Balearic Islands, western Mediterranean). Area under the curve (AUC) and Kappa values are provided for both models, while the explained deviance is reported only for GAM models. Expl. Dev.: explained deviance. Variables acronyms: D (depth), SLP (slope), BKS (backscatter), RUG (rugosity), NOR (*Northness*), EST (*Eastness*), BPI_b (broad BPI), OM (organic matter content), SAN (sand), BC_max (bottom current maximum speed), BT_av (bottom temperature average), BLL (bottom longline fleet), DWT (deep-water trap fleet), DTFG (distance to the nearest trawling fishing ground). The codes of the habitats are shown in Table 1.
Habitat	Depth (m)	Overlap %	Conformity %	MaxEnt				GAM
Habitat	Depth (m)	Overlap %	Conformity %	Variables	AUC	Kappa	Kappa Threshold	Variables	AUC	Kappa	Kappa Threshold	Expl. Dev.
H1	87–148	62.05	99.03	D, BT_av, BKS, SLP, SAN	0.8224946	0.6016176	0.4717684	D, SAN, BKS	0.8772279	0.674012	0.7031042	45.4
H2	89–153	14.45	99.52	D, BLL, SAN, BC_max, SLP	0.8920095	0.6571893	0.6945839	D, BT_av, BPI_b, EST	0.9071619	0.4925104	0.3094611	38.6
H3	89–349	60.77	98.68	D, BLL, OM, BT_av, SLP, BC_max, SAN, DWT	0.9475944	0.7211773	0.4095136	D, BC_max, OM	0.9259565	0.6614428	0.37148	55.2
H4	120–277	34.10	99.68	BLL, D, BT_av, EST, NOR, DWT, SAN	0.9729278	0.6865377	0.3790905	BT_av, D	0.9446993	0.4656679	0.2756664	44.5
H5	89–382	36.17	97.12	D, BT_av, SLP, DWT, SAN	0.8396227	0.4981502	0.3706874	D, SLP, SAN	0.7883651	0.4396282	0.20004	33.1
H6	116–433	21.05	92.81	D, BLL, BC_max, SLP, NOR, EST	0.8707679	0.5291508	0.1308004	D, SAN, EST	0.8313717	0.3280027	0.1645969	27.9
H7	129–718	12.65	86.35	BLL, DWT, SLP, D, BT_av, SAN, EST, BPI_b, NOR	0.7896631	0.3914522	0.4871142	D, SAN, OM, SLP	0.7628749	0.3213679	0.193622	24.9
H8	146–726	20.36	97.60	DWT, SLP, D, BPI_b, SAN, BT_av, NOR, BC_max	0.853332	0.5041225	0.4439022	RUG, OM, BKS, SAN	0.7743568	0.4454822	0.2697871	26.5
H9	223–507	42.40	93.29	SAN, DWT, BLL, SLP, D, BPI_b, NOR	0.8656966	0.6199734	0.3579906	D, SAN, BC_max	0.8923913	0.6392074	0.1653567	60.2
H10	283–755	61.37	65.94	DWT, BT_av, D, BC_max, NOR, SLP, EST	0.8679395	0.5987731	0.4180019	D, SLP	0.8481595	0.6085676	0.2678874	50.8
H11	464–719	48.00	92.78	D, BT_av, BC_max, BPI_b, SLP, RUG, DWT, DTFG, OM, NOR, BLL	0.9131386	0.6688879	0.2337632	D, BC_max, DTFG, SLP	0.9459592	0.7354621	0.4508966	57.6

The benthic habitats of the Mallorca Channel seamounts follow distinct spatial patterns, influenced mainly by depth, followed by substrate type, characteristics of sediments and hydrodynamic conditions (Table 4). The predictive models delineated the main groups of habitats across the seamounts: (i) photic summits hosting structurally complex habitats which are benefit by higher light availability and moderate hydrodynamic conditions; (ii) flanks as transitional zones, influenced by hydrodynamics and substrate heterogeneity; and (iii) adjacent muddy bottoms with deep sedimentary habitats, under stable and low-energy conditions.

Table 4
Relative importance of predictors in each model developed to predict the spatial distribution of potential benthic habitats in the Mallorca Channel seamounts (Balearic Islands, western Mediterranean). The significant response variables and the percentage contribution of each variable in both MaxEnt and GAM models is shown. Variables acronyms: D (depth), SLP (slope), BKS (backscatter), RUG (rugosity), NOR (*Northness*), EST (*Eastness*), BPI_b (broad BPI), OM (organic matter content), SAN (% sand), BC_max (bottom current maximum speed), BT_av (bottom temperature average), BLL (bottom longline fishery), DWT (deep-water trap fishery), DTFG (distance to the nearest trawling fishing ground). The codes of the habitats are shown in Table 1.
Habitat	Depth (m)	Model	D	SLP	BKS	RUG	EST	NOR	BPI_b	OM	SAN	BC_max	BT_av	DWT	BLL	DTFG
H1	87–148	MaxEnt	91.1	0.6	2.0						0.5		4.6
H1	87–148	GAM	65.1		4.5						30.5
H2	89–153	MaxEnt	82.1	5.9							0.8		9.6	1.0
H2	89–153	GAM	51.9	33.6							14.5
H3	89–349	MaxEnt	83.8	1.4						2.7	1.2	1.3	1.9	1.0	4.9
H3	89–349	GAM	51.8							16.1		32.1
H4	120–277	MaxEnt	37.9				5.5	0.9			0.7		7.1	0.8	46.7
H4	120–277	GAM	43.1										56.9
H5	89–382	MaxEnt	72.1	1.0							4.7	2.8			18.5
H5	89–382	GAM	72.4				7.4		9.4				10.8
H6	116–433	MaxEnt	71.1	2.7			1.4	1.5				4.5			17.9
H6	116–433	GAM	43.4				14.9				41.7
H7	129–718	MaxEnt	8.6	11.9			1.6	1.5	1.5		3.7		3.7	32.4	35.3
H7	129–718	GAM	72.4				7.4		9.4				10.8
H8	146–726	MaxEnt	16.7	31.1				1.1	3.5		2.2	0.8	1.4	42.6
H8	146–726	GAM			15.4	49.0				26.4	9.2
H9	223–507	MaxEnt	6.1	7.4				2.0	3.5		34.3			33.7	11.9
H9	223–507	GAM	55.8								24.6	19.6
H10	283–755	MaxEnt	24.4	2.7			2.5	2.9				6.1	28.2	31.9
H10	283–755	GAM	73.1	26.9
H11	464–719	MaxEnt	38.2	3.8		3.6		1.3	4.3	1.7		16.5	23.5	2.8	1.6	2.1
H11	464–719	GAM	47.8	12.8								25.2				14.2

The main results obtained for each of the modelled potential habitats are described below:

H1: Maërl/rhodoliths beds

Composed of free-living calcareous algae from the order Corallinales, this habitat was widely distributed in the shallowest areas across AM and EB summits, between 87 and 148 m depth (Fig. 2). At SO, where the summit is located at a greater depth (225 m), the models did not predict the presence of the habitat. This habitat is restricted to a narrow and shallow bathymetric range, suggesting a preference for stable depth conditions that ensure sufficient light availability for photosynthetic organisms. Depth was identified as the most significant variable in both GAM and Maxent models, while high reflectivity and elevated sandy sediment composition were also significant predictors, although their contribution to the models was notably lower compared to depth (Table 4).

H2: Rocky summits of seamounts in the circalittoral zone with Eunicella spp.

It was distributed between 89 and 153 m depth, specifically on the rocky bottoms of southwestern and northwestern AM and EB summits edges (Fig. 2). It was characterized by aggregations of the octocorals Eunicella cavolini and Eunicella verrucosa. It was mainly influenced by depth and moderate slopes, being these two variables the most significant in both models (Table 4). Other factors such as substrate type and bottom temperature also contribute to the distribution of this habitat. Substrate type showed a clear association with coarse, sand-dominated sediments (60–80% sand content), indicative of mixed hard-bottom environments.

H3: Infralittoral and circalittoral detritic platforms of seamounts with rhodoliths dominated by sponges

It is also mainly distributed across the summits of AM and EB, although its bathymetric range is deeper (89 to 349 m) (Fig. 2). In some areas, sedimentary substrates are interspersed with coralligenous rocky outcrops, contributing to the structural complexity of the habitat, which is also supported by high species richness and benthic biomass. Depth was identified as the most significant variable, followed by OM content in the sediment and maximum bottom current velocity (Table 4). The combination of these factors creates optimal conditions for sponge communities, which dominate the habitat.

H4: Rocky summits of seamounts in the circalittoral zone with Viminella flagellum

It was found exclusively on the circalittoral rocky bottoms of EB, specifically at the southern part of its summit and the shallowest southeastern upper flank, with a depth range of 120 m to 277 m (Fig. 2). Depth and bottom water temperature were the most significant variables, followed mainly by the fishing footprint of the bottom longline fleet and Eastness (Table 4).

H5: Deep rocky bottoms with aggregations of gorgonians

It was mainly distributed across the rocky bottoms of the upper flanks of AM and EB, within a depth range of 89 to 382 m (Fig. 2). On the EB summit, it occupies transition zones between maërl/rhodolith beds and coralligenous bottoms, while on the AM summit it extends along its southern and eastern edges. This habitat is characterized by the co-occurrence of several habitat-forming cnidarians, particularly gorgonians such as Eunicella spp., Acanthogorgia hirsuta, V. flagellum and B. mollis (Table S1). Depth, the fishing footprint of the bottom longline fleet and bottom temperature and BPI were the most influential variables in the distribution models of this habitat.

H6: Bathyal rock with octocorals (Bebryce mollis and Callogorgia verticillata)

It was distributed across rocky outcrops on the summits of the three seamounts, especially extending along their summit borders and the upper flanks (Fig. 2). These substrates, generally covered by a thin layer of sediment, provide a suitable environment for the development of octocoral communities. The habitat was found within the bathymetric range 116–433 m, with C. verticillata dominating on summit margins and B. mollis becoming more abundant at greater depth. Depth was the most significant variable driving its distribution, followed by sandy sediment composition, seafloor orientation and fishing footprint of the longline fleet (Table 4).

H7: Bathyal shelf-edge bottoms with Gryphus vitreus

The facies of the brachiopod G. vitreus predominantly occurred on mixed detritic sedimentary bottoms in the southeastern SO and around EB (Fig. 2). This habitat was distributed along a wide bathymetric range, between 129 and 718 m, with depth emerging as the main driver of its distribution. The fishing footprint of bottom longline and deep-water trap fleets were the most relevant indirect factors, followed by other relevant predictors including bottom temperature, BPI slope gradient and seafloor orientation.

H8: Cliffs, walls, and rocky slopes of submarine elevations and canyons with sponges

Primarily associated with the rocky margins and flanks of the three seamounts (Fig. 2), where sponge communities thrive, this habitat occupied bottoms between 146 and 726 m depth, with steep slopes, high rugosity, fishing footprint of the deep-water trap fisheries, OM, and high reflectivity of the seafloor, being these variables the most significant influencing its distribution (Table 4). The only variable that consistently aligns across both models was the percentage of sand in the substrate.

H9: Bathyal detritic bottoms with sponges

It was found between 223 m to 507 m, occupying the summit of SO and especially the sedimentary bottoms of the base of AM and EB (Fig. 2). The main influencing factors were depth, sand percentage in the sediment, maximum bottom current velocity and slope, jointly with the fishing foot-print of the deep-water trap and the bottom longline fleets (Table 4).

H10: Bathyal muds with Thenea muricata

It corresponds to a facies characterized by the sponge Thenea muricata, which was distributed from 283 to 755 m depth, covering large areas adjacent to the seamounts, predominantly bathyal muddy bottoms (Fig. 2). Its distribution was primarily determined by depth, which shows a strong positive correlation, and also exhibits a negative correlation with slope, bottom mean temperature and maximum bottom current velocity, suggesting that the species thrive in areas with low hydrodynamics, where sedimentation is higher, favoring the accumulation of fine sediments.

H11: Bathyal muds with Isidella elongata

It was found western SO and in the channel between SO and AM, at depths between 464 m and 719 m (Fig. 2). Its distribution was strongly influenced by depth, in both GAM and MaxEnt models, and the most important variables included a negative correlation to bottom current velocity, slope, and proximity to the nearest trawl fishing ground.

H12: Pockmark fields

Up to 3593 pockmarks were identified at depths ranging from 301 to 1178 m. They were located predominantly along the bathyal bottoms adjacent to the three seamounts, although they were also found at the summit of Bel Guyot. These fields were distributed in three distinct clusters, with densities ranging from 1 to 6.8 pockmarks per km². Most pockmarks had circular shapes with U to V-shaped cross sections. Their length ranged from 10 to 500 m and they were up to 20 m in deep, covering a surface of up to 0.18 km². Although most of them appeared randomly distributed, some were aligned, forming strings in mainly northwest-southeast, northeast-southwest and north-south trends. No evidence of ongoing fluid expulsion in these pockmarks has been observed, suggesting that they are relict features formed by ancient fluid migration processes.

3.2. Habitats relationship with environmental variables

The dbRDA showed the environmental variables significantly explained 42.76% of the total variance in presence-absence data matrix of habitats (adjusted R² = 40.86%). The global model was highly significant (pseudo-F = 22.5, p = 0.001), as was the first canonical axis (pseudo-F = 8.8, p = 0.001). Forward selection identified Depth (variance explained: 18.7%), Slope (variance explained: 11.6%), Backscatter (variance explained: 4.5%), Deep-water trap fishery (variance explained: 1.4%), OM content (variance explained: 1.2%), bottom current maximum speed (variance explained: 1.3%), and bottom longline fishery (variance explained: 1.3%) as the most influential variables in the model. Additionally, the model included categorical predictors such as sediment type (Mud% and Sand%) and geographic zones (SO, EB and AM). All selected variables showed significant effects (p = 0.001), based on permutation tests.

The ordination diagram showed three distinct groups of habitats, based on the variables (Fig. 3). A shallow group was associated to the summits of AM and EB. These habitats, hosting H1 maërl/rhodolith beds and associated communities, especially H3 dominated by sponges and to a minor extent H2, H4 and H5, structured by gorgonian-type octocorals, were primarily influenced by high backscatter and a higher percentage of sandy sediments. A second group, including T. muricata and I. elongata grounds (H10 and H11, respectively), was found at deeper areas adjacent to seamounts, specially related to SO. It was associated to sedimentary bottoms containing higher proportions of mud and OM. A third group, characterized by large brachiopod grounds (H7) and sponge communities (H8), was associated to the flanks of the seamounts and occurred on steeper slopes, exposed to bottom currents.

Regarding fisheries, the deep-water traps fishing footprint was linked to the structurally complex habitats on the flanks of the seamounts, which provide niches suitable for Plesionika edwardsii, the target species of this fishery. This relationship is primarily spatial, as the fishing activity overlaps with areas where this species finds shelter and feeding grounds within sponge- and coral-dominated habitats. Bottom longline fishing fleet also showed some spatial relation with these habitats; however, it was more strongly associated with the shallow habitats on the summits of seamounts, especially EB, structured by red algae, sponges and cnidarians.

Given the complexity of deep-sea ecosystems and the lesser accessibility of this environment, robust modelling approaches are essential to improve scientific knowledge and advice for their management and conservation. The predictive models developed in this study generated high-resolution maps of suitability of benthic habitats across the Mallorca Channel seamounts, by integrating biological data with geomorphological, sedimentological, hydrodynamic and anthropogenic variables, providing key insights into the environmental drivers shaping benthic communities.

The integration of GAM and MaxEnt models provided a complementary framework, according to previous research highlighting the advantages of combining multiple modeling techniques to improve spatial predictions in deep-sea ecosystems (Misiuk & Brown, 2024). While MaxEnt was particularly effective in data-limited scenarios, GAM allowed for a more interpretable assessment of environmental drivers, reinforcing their complementarity in habitat suitability modelling. The predictive models demonstrated high reliability and robustness in mapping potential presence of benthic habitats. GAMs explained a high percentage of deviance (> 25%) and most models from both approaches showed reliable AUC (> 0.8) and Kappa values (> 0.4). However, for certain habitats, like H6 (Bathyal rock with octocorals) and H7 (Bathyal shelf-edge bottoms with Gryphus vitreus), the models exhibited lower predictive power compared to the rest. The reduced performance of H6 is likely due to its multispecific nature, as it includes species with varied ecological responses and the limited number of presence records. Both factors have been previously identified as contributors to model uncertainty (Zimmermann & Kienast, 1999; Reiss et al., 2015; de la Torriente et al., 2019). In the case of H7, habitat complexity may have contributed to increase the model uncertainty, as ROV video transects revealed that G. vitreus inhabits both sedimentary and rocky substrates, often interspersed in space. Similarly, the low overlap observed between GAM and MaxEnt predictions for habitats H2 (Rocky summits of seamounts in the circalittoral zone with Eunicella spp.) and H7 (14.5% and 12.7%, respectively) likely stems from the structural and environmental heterogeneity of these habitats. Both occur in mixed or topographically complex substrates and are influenced by a wide array of environmental and anthropogenic variables, which are differently prioritized in each modeling approach.

Although both predictive models identified key environmental drivers of benthic habitat distribution, some differences were observed in the relative importance of variables. This is expected given the distinct methodological approaches: while GAM and MaxEnt predict species occurrence based on individual environmental correlations to spatially delineate habitat distributions, RDA provides a complementary multivariate perspective by jointly analyzing all habitats and explanatory variables to explore their overall relationships. Such methodological differences can lead to variations in identified predictors, a common challenge in ecological modelling (Elith & Leathwick, 2009; Misiuk & Brown, 2024).

4.1. Habitat zonation and spatial distribution

4.1.1. Summits

Depth differences strongly influence the benthic habitats of the three summits. In the mesophotic AM and EB summits, whose minimum depths are 86 and 94 m, respectively, coralline algal beds is the most characteristic habitat. By contrast, the deepest summit of SO (225–290 m) neither has algae nor fauna characteristic of the mesophotic zone. Therefore, the habitats of SO summits are more similar to those distributed on the flanks of the AM and EB seamounts. This is similar to what was found by Diaz et al., (2024a) in the same area for sponge communities, who differentiated communities based on light availability, with mesophotic assemblages at the summits of AM and EB, and aphotic ones on the flanks of all three seamounts and at the deeper summit of SO.

The habitat “Maërl/rhodolith beds” (H1) is distributed along the sedimentary bottoms of the shallower circalittoral zones of AM and EB. As in other Mediterranean regions, the maërl/rhodoliths beds of AM and EB summits thrive at bottoms with sufficient light availability and moderate water currents that help oxygenation and prevents burial, with stable sedimentary conditions (Ballesteros, 1988; Foster, 2001). The habitat “Infralittoral and circalittoral detritic platforms of seamounts dominated by sponges” (H3) is located in a transitional zone between rhodolith beds and deeper rocky environments. The most characteristic species of this habitat, Foraminospongia balearica, was described only a few years ago in the area by Díaz et al. (2021), and soon latter reported from the Tyrrhenian and the Ligurian seas (Toma et al., 2024). This habitat is structured by sponges that grow on accumulations of dead rhodoliths and biogenic gravels, as well as on coralligenous rocky outcrops. These substrates, found on deeper mesophotic detritic bottoms, provide a substrate for sponge development (Díaz et al., 2024a). The dominance of F. balearica in these habitats further highlights the ecological uniqueness of these seamount summits, enhancing their structural complexity and biodiversity (Díaz et al., 2021, 2024a).

Unlike the previous ones, the other two habitats of the seamount’s summits were those restricted to rocky bottoms and were characterized by cnidarians: H2 “Circalittoral rocky seamount summits with Eunicella spp.” and H4 “Circalittoral rocky seamount summits with Viminella flagellum”. Both habitats overlapped with H3, dominated by sponges and, in their deeper range, with other habitats also dominated by octocorals, mainly distributed on the flanks of the seamounts. A similar distribution pattern of these habitats has been documented in the Menorca Channel, where Eunicella spp. communities occur along the shelf break and canyon heads, while V. flagellum dominates in vertical rocky outcrops of the upper slope (Grinyó et al., 2016, 2018). Viminella flagellum is primarily found on rocky outcrops with gentle slopes and on slightly elevated rocky formations adjacent to vertical walls, while only isolated colonies have been recorded directly on steep cliffs (Requena & Gili, 2014; de la Torriente et al., 2018).

4.1.2. Flanks

The flanks of the three seamounts extend to around 720 m depth and are characterized by steep bathymetric gradients, escarpments, and heterogeneous substrates, and are dominated by sessile filter-feeders such as corals and sponges, which benefit from increased current flow and reduced sedimentation (Brooke et al., 2009; Larsson & Purser, 2011; Du Preez et al., 2016; Serrano et al., 2017).

The habitat H5 “Deep rocky bottoms with aggregations of gorgonians” is distributed along the rocky upper slopes of AM and EB. It occurs on semi-buried rocky and coralligenous substrates, often transitioning from maërl/rhodolith beds and the habitat of the mesophotic zone dominated by sponges to the deeper H6 “Bathyal rock with octocorals (Bebryce mollis and Callogorgia verticillata)”. Acanthogorgia hirsuta, one of the main structural species of habitat H5, can form monospecific facies on bathyal rocks (Michez et al., 2014). In the study area it co-occurs with Eunicella spp. and, to a lesser extent, with V. flagellum on shallower margins, while B. mollis, a smaller species, predominates in deeper areas. Similar communities have been reported in the Menorca Channel, the Seco de los Olivos seamount and Sardinia, in the central Mediterranean (Requena & Gili, 2014; de la Torriente et al., 2018; Cau et al., 2017). H6 is distributed primarily on irregular substrates, such as semi-buried rocky formations, including promontories, escarpments and canyon walls, which provide both anchorage and exposure to bottom currents, essential for the settlement and feeding of suspension feeders (Bo et al., 2012; de la Torriente et al., 2020). This habitat is structured by the octocorals C. verticillata, a tall and delicate gorgonian that can reach over 1 m in height and B. mollis, more common on deeper slopes. Callogorgia verticillata has been recorded with B. mollis in the deep gorgonian communities of the Menorca Channel (40–360 m depth; Grinyó et al., 2016) and distributed on rocky outcrops associated with V. flagellum in the Seco de los Olivos (de la Torriente et al., 2018). In the Tyrrhenian Sea, it is often found on hard substrates with weak currents and covered by a thin sediment layer, likely due to its fragility and limited tolerance to strong water movement (Bo et al., 2014).

Rocky outcrops, escarpments and steep slopes across the flanks of the three seamounts, were mainly structured by sponges assigned to H8 “Cliffs, walls, and rocky slopes of submarine elevations and canyons with sponges”. The diverse morphology of species characterizing it, including massive (Jaspis sp.), encrusting (Hamacantha spp.) and erect forms (Phakellia aff. robusta), with Tretodictyum reiswigi representing one of its most distinctive species (Díaz et al., 2024), provides to this habitat a moderate three-dimensional complexity, although less pronounced than that of cnidarian-dominated habitats. The transition between T. reiswigi dominance, one of the most distinctive species of H8 and the increment of F. balearica, the most characteristic species of H3, also dominated by sponges, marks the shift from aphotic to mesophotic zones, respectively. By contrast, the habitat H7 “Bathyal shelf-edge bottoms with Gryphus vitreus” is structured by the large brachiopod G. vitreus, a filter-feeding species widely distributed along the outer continental shelf and upper slope throughout the Mediterranean. Although G. vitreus does not build three-dimensional structures, it can constitute a dominant biotic element that defines a distinct biotope (Toma et al., 2022).

4.1.3. Adjacent bottoms

The adjacent seabed of the Mallorca Channel seamounts extends from the flanks into the bathyal zone, reaching depths of up to 1720 m. It is characterized by sedimentary bottoms that create a more homogeneous environment compared to the flanks and summits. At these bottoms, two sponge-dominated communities are widespread, while a facies of deep-water coral has also been located. This homogeneity is disrupted by pockmarks, geomorphological erosive and depositional structures that are very numerous and widely distributed in the muddy substrates of the area.

A structurally complex sponge-dominated community, composed by both large (e.g. P. compressa) and small sponges (e.g. D. inornata), was found not only at the base of AM and EB but also at the deep summit of SO. It could be assigned to the habitat H9 “Bathyal detritic bottoms with sponges”, although it exhibits much lower species richness and biomass than shallower mesophotic habitats dominated by sponges, its richness is still remarkable (Díaz et al., 2024a). In some areas, it overlaps with facies of the brachiopod G. vitreus and, at their deeper range, transitions into bathyal muds dominated by T. muricata. These muds located in bathyal bottoms composed by a small, semi-buried demosponges could be assigned to the habitat H10 “Bathyal Muds with Thenea muricata”. They are widely distributed across the deep-sea sedimentary plains surrounding the three seamounts, making it one of the most extensive and deepest potential habitats in the area. T. muricata has been frequently recorded in sedimentary trawling grounds off Balearic Islands, where it reaches high densities (Massutí & Reñones, 2005; Ramon et al., 2014). The bathyal muds with T. muricata can be considered a continuation of the bathyal detritic bottoms with sponges (Templado et al., 2012) and overlap with both I. elongata and pockmark fields.

The habitat H11 “Bathyal Muds with Isidella elongata” was distributed western off SO and within the channel between SO and AM. Our results partially align with those obtained by other authors (Mastrototaro et al., 2017; González-Irusta, et al., 2022; Standaert et al., 2023), who also reported the presence of this habitat in the southeastern and southern areas of AM. The observed discrepancies may stem from differences in the environmental variables considered, spatial resolution and modelling approaches employed in both studies. The octocoral I. elongata is the only bamboo coral species in the Mediterranean, where it has its primary distribution (Dieuzeide, 1960; Maynou & Cartes, 2012; Michez et al., 2014; Lauria et al., 2017; Carbonara et al., 2020). It structures bathyal soft-bottom habitats, particularly in gently sloping muddy areas (Cartes et al., 2013), where it enhances three-dimensional complexity and contributes to ecosystem productivity (Cartes et al., 2022). Their facies support high diversity of fish, decapod crustaceans and other invertebrates (Maynou & Cartes, 2012), including Aristeus antennatus and Aristaeomorpha foliacea, the main target species of deep-water trawling fishery in the Mediterranean (Sardà et al., 2004).

Covering a surface of 2583 km² (60% of the modelled area), the habitat H12 “Pockmark fields” was the most widespread habitat distributed in the Mallorca channel seamounts. The formation and persistence of these geomorphological structures are likely driven by moderate-intensity bottom currents, which influence sediment transport and OM accumulation (Judd and Hovland, 2007). In fact, in these areas the highest mud and OM contents encountered at their sediments, as well as the thicker sedimentary cover within the Mallorca Channel, support that these fields constitute potential areas for deep fluid generation, OM degradation as well as fluid uplifting throughout these pockmarks. The location of pockmarks on sedimentary bottoms adjacent to seamounts and other types of structural morphologies linked to sedimentary instabilities suggests their association with the presence of deep faults or slide surfaces, which facilitate the expulsion of fluids under pressure (gas or water), which lead to the formation of these depressions (Iglesias et al., 2010; Acosta et al., 2013). However, the absence of evidence of active emissions observed during data collection and sampling process suggests that their activity is episodic. These pockmarks are likely relict features that have maintained their morphology over time due to the action of prevailing bottom currents in the study area and can be considered inactive fossil depressions. Despite their potential ecological significance, no chemosynthetic communities or carbonate crusts typically associated with methane fluid seepage have been identified in the study area. Instead, the epibenthic and nectobenthic communities observed in pockmark fields resemble those found in other soft-sediment deep-sea habitats, consisting of sponges, bivalves, decapod crustaceans, echinoderms and fishes (Massutí et al., 2022a), although some structuring species like F. quadrangularis and I. elongata have also been observed at low densities inside them.

4.2. Environmental drivers influencing benthic habitats

Depth, slope, backscatter, substrate type (soft vs. hard substrates) and bottom currents were identified as key drivers influencing habitat formation in the seamounts of the Mallorca Channel. These findings align with previous studies that emphasize the role of geomorphological and oceanographic factors in structuring benthic communities (e.g. Reiss et al., 2015; Howell et al., 2016; Misiuk & Brown, 2024).

Depth is widely recognized as a fundamental driver of habitat zonation in deep-sea ecosystems (e.g. McClain & Lundsten, 2015; Du Preez et al., 2016), due to its influence on multiple environmental factors, including temperature, light and nutrient availability (Buhl-Mortensen et al., 2010; Serrano et al., 2017; de la Torriente et al., 2019). This is confirmed from our results, showing the shallowest habitats (H1 and H3) strongly associated to the AM and EB summits, with optimal conditions for the development of rhodolith beds, which thrive in shallow, dim-light environments with limited sediment deposition (Basso et al., 2017). The summits of AM and EB also host coralligenous gorgonian-dominated habitats, such as H2, H4 and H5 which is consistent with studies highlighting the role of depth and topographic features in promoting gorgonian aggregations at seamounts (Clark et al., 2010), and sponges, which can benefit from light due to its associated photosynthetic symbionts (Keesing et al., 2012; Harris, 2022).

Backscatter intensity, a proxy for seafloor substrate type which reflects variations in sediment grain size, hardness and biogenic structures (e.g., Fish & Carr, 1990; Blondel & Murton, 1997), affects the distribution of H1, H2 and H3, which are associated with biogenic hard-bottom communities like rhodoliths and coralligenous assemblages and also other gorgonian-dominated habitats such as H4 and H5 which are distributed deeper, along the flanks and upper slopes of the seamounts. Consolidated substrates or dense biogenic structures provide stable environment for gorgonians, sponges and cold-water corals (Miller et al., 2012). Conversely, sedimentary habitats, such as H10 and H11, were associated with lower backscatter values, consistent with their unconsolidated nature. Despite this general pattern, some inconsistencies were observed on backscatter data, particularly in areas where thin sediment deposits over hard substrates may attenuate acoustic reflectivity. Sañé et al. (2021) demonstrated that rhodolith beds exhibit heterogeneous backscatter signatures, influenced by density, size, and burial degree of rhodoliths. These challenges highlight the importance of integrating acoustic data with direct observations, such as ROV imagery, for accurate habitat characterization.

Slope gradient is another habitat structuring factor, as it affects substrate stability and local hydrodynamics, which in turn shape species composition and biodiversity (Brown et al., 2011). High slope values are typically associated with rocky outcrops and escarpments, where currents limit sediment deposition and enhance the availability of suspended OM, benefiting filter-feeding organisms (McArthur et al., 2010; Gori et al., 2011). In our study, the habitat H8, characterized by diverse sponge morphotypes of both encrusting and massive/erect forms, corresponds with high-slope areas, where steep terrain facilitates increased current flow and enhances larval recruitment of sessile filter-feeders. These findings align with studies showing that steep slopes and escarpments promote the development of structured benthic communities reliant on strong water flow for feeding (Du Preez et al., 2016; Serrano et al., 2017).

The distribution of H7 highlights the complexity of mixed-substrate habitats, where slope gradient appears to be more influential than backscatter. Although G. vitreus is considered a stenotypic species, which ideal conditions are constant bottom currents (Emig, 1987), its presence in both rocky and sedimentary substrates (Toma et al., 2022) suggests that slope-driven variations in sediment dynamics and hydrodynamics are key factors in defining habitat boundaries.

Hydrodynamics is a driver of habitat distribution, as shown by the influence of bottom current maximum speed on both H7 and H8. Our results align with studies showing that high-energy environments promote the development of suspension-feeding communities, such as gorgonians and sponges, by enhancing the delivery of food particles (McArthur et al., 2010; Gori et al., 2011; Templado et al., 2012). Similar patterns have been observed in other Mediterranean seamounts, where current acceleration around topographic features supports diverse benthic assemblages (Mohn et al., 2014). These hydrodynamic conditions reduce sediment deposition, providing stable substrates for habitat-forming species (Würtz & Rovere, 2015) and maintain habitat heterogeneity across seamount ecosystems (Clark et al., 2010).

In contrast, deeper environments in low-slope areas characterized by higher sedimentation rates and lower hydrodynamic energy, were predominantly associated to soft bottoms habitats such as H10 and H11. In these habitats, finer sediments accumulate, providing stable conditions for settlement and anchorage of deposit feeders and burrowing fauna. Furthermore, sedimentological variables such as mud and OM contents, influenced both habitats. The OM content is linked to substrate characteristics and nutrient availability, determining the occurrence and composition of benthic assemblages (Wei et al., 2010). Soft environments tend to exhibit lower structural complexity, often linked to reduced biodiversity, due to fewer microhabitats and shelter opportunities offered by sedimentary plains compared to rocky environments (McArthur et al., 2010).

The impacts of fishing activities in marine ecosystems are very diverse (e.g. Gislason et al., 1994; Dayton et al., 1995; Goñi, 1998; Jennings & Kaiser, 1998). Despite being considered a less destructive gear than bottom trawls (e.g. Pham et al., 2014), bottom longlines can still damage fragile benthic organisms through entanglement or breakage, especially gorgonians, corals and large sponges with branching morphologies (Bo et al., 2014; de la Torriente et al., 2018). There is very little information on the impact of deep-water traps, and their impact is supposed to be low (Kopp et al., 2020). However, our results also suggested the footprints of bottom longline and deep-water trap fleets as potential drivers for some of those habitats distributed in summits (H4, H6) and flanks of the seamounts (H7, H8). However, the positive relationships among them suggest that these fisheries are more likely responding to habitat presence, targeting areas where commercially valuable species associated with sponges and octocorals occur (García-Rodríguez et al., 2000; Massutí et al., 2022a). During the INTEMARES research surveys developed in the Mallorca Channel seamounts, we observed some longlines and traps lost on the seabed, suggesting potential threats such as ghost fishing or physical abrasion on benthic species.

4.3. Towards protection and management

The habitat modelling and mapping conducted in this study provide the first comprehensive spatial framework supporting the designation of the Mallorca Channel seamounts and their adjacent areas as a SCI within the Natura 2000 network under the EU Habitats Directive. According to this directive, the HCI are defined as those that are endangered and/or in decline across the European Union and for which Natura 2000 serves as the main instrument for protection and conservation.

The potential designation as HCI of the 12 habitats identified and modelled across the study area should be justified based on the criteria of the EU Interpretation Manual of Habitats, the Barcelona Convention, the EUNIS classification (2022) and the Standard List of Marine Habitats in Spain (Templado et al., 2012). Six of them (H2, H4, H5, H6, H8 and H11) could potentially be included under the category 1170 (Reefs); and two others (H1 and H3) could be assigned within category 1110 (Sandbanks permanently covered by seawater). The most spatially extensive habitat, Pockmark fields (H12), aligns with the category 1180 (Submarine structures made by leaking gases), although no recent fluid emissions or associated chemosynthetic communities have been detected. The remaining H8, H10 and H11 have a limited structural complexity or lack of clear alignment with existing conservation typologies. While these habitats may share certain features with HCI category 1110, such as their sedimentary nature, they generally exhibit lower biodiversity, coverage and abundance of habitat-forming species, though further ecological assessment would be required to support their potential inclusion.

The spatial modelling has also allowed to detect overlapping between potential habitats and fishing activities, which emphasizes the need for a further multiple-use zonation strategy, wherein core conservation zones should be protected, while in other buffer zones fishing activities could be allowed and regulated. Such an approach would reconcile biodiversity conservation with traditional fisheries, in line with the Marine Strategy Framework Directive (2008/56/CE), the Habitats Directive (92/43/CEE) and the Maritime Spatial Planning Directive (2014/89/EU). This methodological approach, similar to the previously applied at the Seco de los Olivos seamount already located within a SCI of the Natura 2000 network (de la Torriente et al., 2019, 2022), offer a replicable strategy for the protection and management of other Mediterranean deep-sea ecosystems.

Funding

Open Access funding was provided thanks to the CRUE-CSIC agreement with Springer Nature. This research was performed in the framework of the LIFE IP INTEMARES project, coordinated by the Biodiversity Foundation of the Ministry for the Ecological Transition and the Demographic Challenge, with financial support from the European Union’s LIFE program (LIFE15 IPE ES012). AF’s contribution to this study was supported by the BIODIV project: "Scientific and technical advice for the monitoring of marine biodiversity: protected marine areas and species of State competence (2022–2025)", funded by the European Union – NextGenerationEU through the Recovery, Transformation and Resilience Plan, and promoted by the Directorate General for Biodiversity, Forests and Desertification of the Ministry for Ecological Transition and the Demographic Challenge and the Spanish National Research Council (CSIC), through the Spanish Institute of Oceanography (IEO).

Author Contribution

All authors contributed to the study conception and design. A. Frank (AF), O. Sanchez-Guillamón, S. Keller and D. Mata analyzed the data. The first draft of the manuscript was written by AF, E. Massutí and M.T. Farriols and all authors commented on previous versions of the manuscript. All authors contributed to the article and approved the submitted version.

Acknowledgement

The authors thank the European Union’s LIFE program for funding the LIFE IP INTEMARES project and the Biodiversity Foundation of the Spanish Ministry for the Ecological Transition and the Demographic Challenge for the coordination of this project. We also thank the crew of R/V Ángeles Alvariño and the R/V Sarmiento de Gamboa and the participants who took part in the INTEMARES_A22B_0718, INTEMARES_A22B_1019, INTEMARES_A22B_0720, INTEMARES_A22B_0820 and LIROBAL_0516 research surveys.

Data Availability

The datasets used and analyzed in the current study are available from the corresponding author upon reasonable request, subject to the data-sharing policies and limitations of the projects.

Acosta J, Muñoz A, Herranz P, Palomo C, Ballesteros M, Vaquero M, Uchupi E (2001) Geodynamics of the Emile Baudot Escarpment and the Balearic Promontory, western Mediterranean. Mar Pet Geol 18(3):349–369. https://doi.org/10.1016/S0264-8172(01)00003-4
Acosta J, Canals M, López-Martínez J, Muñoz A, Herranz P, Urgeles R, Palomo C, Casamor JL (2003) The Balearic Promontory geomorphology (western Mediterranean): Morphostructure and active processes. Geomorphology 49(3–4):177–204. https://doi.org/10.1016/S0169-555X(02)00168-X
Acosta J, Canals M, Carbó A, Muñoz A, Urgeles R, Muñoz-Martín A, Uchupi E (2004) Sea floor morphology and Plio-Quaternary sedimentary cover of the Mallorca Channel, Balearic Islands, western Mediterranean. Mar Geol 206(1–4):165–179. https://doi.org/10.1016/j.margeo.2004.02.008
Acosta J, Fontán A, Muñoz A, Muñoz-Martín A, Rivera J, Uchupi E (2013) The morpho-tectonic setting of the Southeast margin of Iberia and the adjacent oceanic Algero-Balearic Basin. Mar Pet Geol 45:17–41. https://doi.org/10.1016/j.marpetgeo.2013.04.005
Aguilar R, López Correa M, Calcinai B, Pastor X, de la Torriente A, García S (2011) First records of Asbestopluma hypogea Vacelet and Boury-Esnault, 1996 (Porifera, Demospongiae, Cladorhizidae) on seamounts and in bathyal settings of the Mediterranean Sea. Zootaxa 2925:33–40. https://doi.org/10.11646/zootaxa.2925.1.2
Akaike H (1998) Information theory and an extension of the maximum likelihood principle. In E. Parzen, K. Tanabe, & G. Kitagawa (Eds.), Selected papers of Hirotugu Akaike (pp. 199–213). Springer. https://doi.org/10.1007/978-1-4612-1694-0_15
Ballesteros E (1989) Composición y estructura de los fondos de maërl de Tossa de Mar (Gerona, España). Collectanea Bot 17(2):161–182. https://doi.org/10.3989/collectbot.1989.v17.137
Basso D, Babbini L, Ramos-Esplá AA, Salomidi M (2017) Mediterranean rhodolith beds. In R. Riosmena-Rodríguez, W. Nelson, & J. Aguirre (Eds.), Rhodolith/Maërl Beds: A Global Perspective (pp. 281–298). Springer. https://doi.org/10.1007/978-3-319-29315-8_11
Bilan M, Grinyó J, Puig P, Gori A (2023) Vulnerability of six cold-water corals to sediment resuspension from bottom trawling fishing. Mar Pollut Bull 196:115423. https://doi.org/10.1016/j.marpolbul.2023.115423
Blondel P, Murton BJ (1997) Handbook of seafloor sonar imagery. Wiley/Praxis Series in Remote Sensing
Bo M, Canese S, Spaggiari C, Pusceddu A, Bertolino M, Angiolillo M, Giusti M, Loreto MF, Salvati E, Greco S (2012) Deep Coral Oases in the South Tyrrhenian Sea. PLoS ONE 7(11):e49870. https://doi.org/10.1371/journal.pone.0049870
Bo M, Bava S, Canese S, Angiolillo M, Cattaneo-Vietti R, Bavestrello G (2014) Fishing impact on deep Mediterranean rocky habitats as revealed by ROV investigation. Biol Conserv 171:167–176. https://doi.org/10.1016/j.biocon.2014.01.011
Bologna PAX (2006) Assessing within habitat variability in plant demography, faunal density, and secondary production in an eelgrass (Zostera marina L.) bed. J Exp Mar Biol Ecol 329(1):122–134. https://doi.org/10.1016/j.jembe.2005.08.004
Brooke SD, Holmes MW, Young CM (2009) Sediment tolerance of two different morphotypes of the deep-sea coral Lophelia pertusa from the Gulf of Mexico. Mar Ecol Prog Ser 390:137–144. https://doi.org/10.3354/meps08191
Brown CJ, Smith SJ, Lawton P, Anderson JT (2011) Benthic habitat mapping: A review of progress towards improved understanding of the spatial ecology of the seafloor using acoustic techniques. Estuar Coast Shelf Sci 92(3):502–520. https://doi.org/10.1016/j.ecss.2011.02.007
Bruno JF, Kennedy CW (2000) Patch-size dependent habitat modification and facilitation on New England cobble beaches by Spartina alterniflora. Oecologia 122(1):98–108. https://doi.org/10.1007/PL00008841
Buhl-Mortensen L, Vanreusel A, Gooday AJ, Levin LA, Priede IG, Buhl-Mortensen P, Gheerardyn H, King NJ, Raes M (2010) Biological structures as a source of habitat heterogeneity and biodiversity on the deep ocean margins. Mar Ecol 31(1):21–50. https://doi.org/10.1111/j.1439-0485.2010.00359.x
Carbonara P, Zupa W, Follesa MC, Cau A, Capezzuto F, Chimienti G, Maiorano P (2020) Exploring a deep-sea vulnerable marine ecosystem: Isidella elongata (Esper, 1788) species assemblages in the Western and Central Mediterranean. Deep Sea Res Part I 166:103406. https://doi.org/10.1016/j.dsr.2020.103406
Carpine C, Grasshoff M (1975) Les gorgonaires de la Méditerranée. Bull de l'Institut Océanographique Monaco 7(430):1–140. http://paleopolis.rediris.es/benthos/TaP/Carpine-1975.pdf
Cartes JE, Iacono L, Mamouridis C, López-Pérez V, C., Rodríguez P (2013) Geomorphological, trophic and human influences on the bamboo coral Isidella elongata assemblages in the deep Mediterranean: To what extent does Isidella form habitat for fish and invertebrates? Deep Sea Res Part I 76:52–65. https://doi.org/10.1016/j.dsr.2013.01.006
Cartes JE, Díaz-Viñolas D, González-Irusta JM, Serrano A, Mohamed S, Lombarte A (2022) The macrofauna associated to the bamboo coral Isidella elongata: To what extent the impact on Isididae affects diversification of deep-sea fauna. Coral Reefs 41(4):1273–1284. https://doi.org/10.1007/s00338-022-02243-w
Cau A, Follesa MC, Moccia D, Bellodi A, Mulas A, Bo M, Canese S, Angiolillo M, Cannas R (2017) Leiopathes glaberrima millennial forest from SW Sardinia as nursery ground for the small spotted catshark Scyliorhinus canicula. Aquatic Conservation: Marine and Freshwater Ecosystems, 27(3), 731–735. https://doi.org/10.1002/aqc.2717
Cerrano C, Danovaro R, Gambi C, Pusceddu A, Riva A, Schiaparelli S (2010) Gold coral (Savalia savaglia) and gorgonian forests enhance benthic biodiversity and ecosystem functioning in the mesophotic zone. Biodivers Conserv 19(1):153–167. https://doi.org/10.1007/s10531-009-9712-5
Clark MR, Rowden AA, Schlacher T, Williams A, Consalvey M, Stocks KI, Rogers AD, O’Hara TD, White M, Shank TM, Hall-Spencer JM (2010) The ecology of seamounts: Structure, function, and human impacts. Annual Rev Mar Sci 2:253–278. https://doi.org/10.1146/annurev-marine-120308-081109
Clark MR, Schlacher TA, Rowden AA, Stocks KI, Consalvey M (2012) Science priorities for seamounts: Research links to conservation and management. PLoS ONE 7(1):e29232. https://doi.org/10.1371/journal.pone.0029232
Clarke KR, Warwick RM (1994) Change in marine communities: An approach to statistical analysis and interpretation, 1st edn. Plymouth Marine Laboratory
Dayton PK, Thrush SF, Agardy MT, Hofman RJ (1995) Environmental effects of marine fishing. Aquat Conservation: Mar Freshw Ecosyst 5(3):205–232. https://doi.org/10.1002/aqc.3270050305
De la Torriente A, Serrano A, Fernández-Salas LM, García M, Aguilar R (2018) Identifying epibenthic habitats on the Seco de los Olivos Seamount: Species assemblages and environmental characteristics. Deep Sea Res Part I 135:9–22. https://doi.org/10.1016/j.dsr.2018.03.015
De la Torriente A, González-Irusta JM, Aguilar R, Fernández-Salas LM, Punzón A, Serrano A (2019) Benthic habitat modelling and mapping as a conservation tool for marine protected areas: A seamount in the western Mediterranean. Aquat Conservation: Mar Freshw Ecosyst 29(5):732–750. https://doi.org/10.1002/aqc.3075
De la Torriente A, Aguilar R, González-Irusta JM, Blanco M, Serrano A (2020) Habitat forming species explain taxonomic and functional diversities in a Mediterranean seamount. Ecol Ind 118:106747. https://doi.org/10.1016/j.ecolind.2020.106747
De la Torriente Diez A, González-Irusta JM, Serrano A, Aguilar R, Sánchez F, Blanco M, Punzón A (2022) Spatial assessment of benthic habitats vulnerability to bottom fishing in a Mediterranean seamount. Mar Policy 135:104850. https://doi.org/10.1016/j.marpol.2021.104850
Dieuzeide R (1960) Les fonds chalutables à 600 m. par le travers de Castiglione. Recherches sur la faune à Isidella elongata. Bull des Travaux de la Stn d’Aquiculture et de Pêche de Castiglione 2:9–86
Díaz JA, Ramírez-Amaro S, Ordines F (2021) Sponges of Western Mediterranean seamounts: New genera, new species, and new records. PeerJ 9:e11879. https://doi.org/10.7717/peerj.11879
Díaz JA, Ordines F, Farriols MT, Melo-Aguilar C, Massutí E (2024a) Sponge assemblages in fishing grounds and seamounts of the Balearic Islands (western Mediterranean). Deep Sea Res Part I 203:104211. https://doi.org/10.1016/j.dsr.2023.104211
Díaz JA, Ordines F, Massutí E, Cárdenas P (2024b) From caves to seamounts: The hidden diversity of tetractinellid sponges from the Balearic Islands, with the description of eight new species. PeerJ 12:e16584. https://doi.org/10.7717/peerj.16584
Díaz JA, Ordines F, Massutí E (2024c) First record of the recently described Axinella venusta Idan, Shefer, Feldstein & Ilan, 2021 (Demospongiae: Axinellidae) in the western Mediterranean. Cahiers de biologie marine 65:87–91. https://doi.org/10.21411/CBM.A.B2FE3AF5
Du Preez C, Curtis JMR, Clarke ME (2016) The structure and distribution of benthic communities on a shallow seamount (Cobb Seamount, Northeast Pacific Ocean). PLoS ONE 11(11):e0165513. https://doi.org/10.1371/journal.pone.0165513
Elith J, Leathwick JR (2009) Species distribution models: Ecological explanation and prediction across space and time. Annu Rev Ecol Evol Syst 40:677–697. https://doi.org/10.1146/annurev.ecolsys.110308.120159
Emig CC (1987) Offshore brachiopods investigated by submersible. J Exp Mar Biol Ecol 108(3):261–273. https://doi.org/10.1016/0022-0981(87)90089-X
Farriols MT, Ordines F, Somerfield PJ, Pasqual C, Hidalgo M, Guijarro B, Massutí E (2017, April) Bottom trawl impacts on Mediterranean demersal fish diversity: Not so obvious or are we too late? Cont Shelf Res 137:11–22 Elsevier BV. https://doi.org/10.1016/j.csr.2016.11.011
Food and Agriculture Organization of the United Nations (FAO) (2009) Report of the Technical Consultation on International Guidelines for the Management of Deep-Sea Fisheries in the High Seas. FAO Fisheries and Aquaculture Report No. 881. FAO, Rome, Italy. https://www.fao.org/4/i0605t/i0605t00.pdf
Fielding AH, Bell JF (1997) A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ Conserv 24(1):38–49. https://doi.org/10.1017/S0376892997000088
Fish JP, Carr HA (1990) Sound underwater images: A guide to the generation and interpretation of side scan sonar data. Lower Cape Publishing, Orleans, MA
Foster MS (2001) Rhodoliths: Between rocks and soft places. J Phycol 37(5):659–667. https://doi.org/10.1046/j.1529-8817.2001.00195.x
García-Alegre A, Sánchez F, Gómez-Ballesteros M, Hilmar H, Serrano A (2014) Modelling and mapping the local distribution of representative species on the Le Danois Bank, El Cachucho Marine Protected Area (Cantabrian Sea). Deep Sea Res Part II 106:151–164. https://doi.org/10.1016/j.dsr2.2013.12.012
García-Rodríguez M, Esteban A, Pérez Gil JL (2000) Considerations on the biology of Plesionika edwardsii (Brandt, 1851) (Decapoda, Caridea, Pandalidae) from experimental trap catches in the Spanish western Mediterranean Sea. Scientia Marina, 64, 369–379. https://doi.org/10.3989/scimar.2000.64n4369
García-Rodríguez M, Esteban A (1999) On the biology and fishery of Aristeus antennatus (Risso, 1816), (Decapoda, Dendrobranchiata) in the Ibiza Channel (Balearic Islands, Spain). Scientia Marina, 63(1), 27–37. https://doi.org/10.3989/scimar.1999.63n127
Gislason H (1994) Ecosystem effects of fishing activities in the North Sea. Mar Pollut Bull 29(6–12):520–527. https://doi.org/10.1016/0025-326X(94)90680-7
Goñi R (1998) Ecosystem effects of marine fisheries: An overview. Ocean Coastal Manage 40(1):37–64. https://doi.org/10.1016/S0964-5691(98)00037-4
González-Irusta JM, Cartes JE, Punzón A, Díaz D, de Sola G, L., Serrano A (2022) Mapping habitat loss in the deep-sea using current and past presences of Isidella elongata (Cnidaria: Alcyonacea). ICES J Mar Sci 79(6):1888–1901. https://doi.org/10.1093/icesjms/fsac123
Gori A, Rossi S, Linares C, Berganzo E, Orejas C, Dale MRT, Gili JM (2011) Size and spatial structure in deep versus shallow populations of the Mediterranean gorgonian Eunicella singularis (Cap de Creus, northwestern Mediterranean Sea). Mar Biol 158(8):1721–1732. https://doi.org/10.1007/s00227-011-1686-7
Grinyó J, Gori A, Ambroso S, Purroy A, Calatayud C, Dominguez-Carrió C, Coppari M, Lo Iacono C, López-González PJ, Gili JM (2016) Diversity, distribution and population size structure of deep Mediterranean gorgonian assemblages (Menorca Channel, Western Mediterranean Sea). Prog Oceanogr 145:42–56. https://doi.org/10.1016/j.pocean.2016.05.001
Grinyó J, Gori A, Greenacre M, Requena S, Canepa A, Lo Iacono C, Ambroso S, Purroy A, Gili JM (2018) Megabenthic assemblages in the continental shelf edge and upper slope of the Menorca Channel, Western Mediterranean Sea. Prog Oceanogr 162:40–51. https://doi.org/10.1016/j.pocean.2018.02.002
Guijarro B, Ordines F, Pasqual C, Valls M, Quetglas A, Massutí E (2020) La pesca de ròssec al voltant de l’arxipèlag de Cabrera. En A. M. Grau, J. J. Fornós, G. Mateu, P. Oliver & B. Terrasa (Eds.), Arxipèlag de Cabrera: Història Natural (pp. 375–391). Monografies de la Societat d’Història Natural de les Balears, 30. http://hdl.handle.net/10261/311597
Guijarro B (2016) Informe Ejecutivo Campaña LIROBAL_0516: B/O Ángeles Alvariño, 23–26 Mayo 2016. Instituto Español de Oceanografía. [Unpublished report, 6 pp]
Harris BJ (2022) The distribution and feeding ecology of temperate marine sponges through shallow and mesophotic habitats (Doctoral dissertation, Victoria University of Wellington, New Zealand). Open Access Te Herenga Waka–Victoria University of Wellington. https://openaccess.wgtn.ac.nz/articles/thesis/The_distribution_and_feeding_ecology_of_temperate_marine_sponges_through_shallow_and_mesophotic_habitats/19669398
Hastie T, Tibshirani R (1990) Generalized additive models. Chapman and Hall
Howell KL, Piechaud N, Downie AL, Kenny A (2016) The distribution of deep-sea sponge aggregations in the North Atlantic and implications for their effective spatial management. Deep Sea Res Part I 115:309–320. https://doi.org/10.1016/j.dsr.2016.07.005
Iglesias J, Ercilla G, García-Gil S, Judd AG (2010) Pockforms: an evaluation of pockmark-like seabed features on the landes plateau, Bay of Biscay. Geo Mar Lett 30(3–4):207–219. https://doi.org/10.1007/s00367-009-0182-2
Jennings S, Lancaster JE, Woolmer A, Cotter AJ (1999) Distribution, diversity and abundance of epibenthic fauna in the North Sea. J Mar Biol Association United Kingd 79(3):385–399. https://doi.org/10.1017/S0025315498000581
Jones CG, Lawton JH, Shachak M (1994) Organisms as ecosystem engineers. Oikos 69:373–386. https://doi.org/10.2307/3545850
Judd AG, Hovland M (2007) Seabed Fluid Flows, the Impact on Geology, Biology and the Marine Environment. Cambridge University Press, p. 475
Juza M, Mourre B, Renault L, Gómara S, Sebastián K, Lora S, Beltran JP, Frontera B, Garau B, Troupin C, Torner M, Heslop E, Casas B, Escudier R, Vizoso G, Tintoré J (2016) SOCIB operational ocean forecasting system and multi-platform validation in the Western Mediterranean Sea. J Oper Oceanogr 9(sup1):s155–s166. https://doi.org/10.1080/1755876X.2015.1117764
Keesing JK, Usher KM, Fromont J (2012) First record of photosynthetic cyanobacterial symbionts from mesophotic temperate sponges. Mar Freshw Res 63(5):403–408. https://doi.org/10.1071/MF11216
Kopp D, Coupeau Y, Vincent B, Méhault S, Morandeau F, Simon J (2020) The low impact of fish traps on the seabed makes it an eco-friendly fishing technique. PLoS ONE 15(8):e0237819. https://doi.org/10.1371/journal.pone.0237819
Larsson AI, Purser A (2011) Sedimentation on the cold-water coral Lophelia pertusa: Cleaning efficiency from natural sediments and drill cuttings. Mar Pollut Bull 62(6):1159–1168. https://doi.org/10.1016/j.marpolbul.2011.03.041
Lauria V, Garofalo G, Fiorentino F, Massi D, Milisenda G, Piraino S, Russo T, Gristina M (2017) Species distribution models of two critically endangered deep-sea octocorals reveal fishing impacts on vulnerable marine ecosystems in central Mediterranean Sea. Sci Rep 7:1. https://doi.org/10.1038/s41598-017-08386-z
Maldonado M, Aguilar R, Blanco J, García S, Serrano A, Punzón A (2015) Aggregated clumps of lithistid sponges: A singular, reef-like bathyal habitat with relevant paleontological connections. PLoS ONE 10(5):e0125378. https://doi.org/10.1371/journal.pone.0125378
Manly BFJ (1991) Randomization and Monte Carlo methods in biology. Chapman & Hall
Marín P, Aguilar R, García S, Pardo E (2011) Seamounts of the Balearic Islands: Mallorca Channel. Protection proposal for Ausias March, Emile Baudot and Ses Olives. Oceana. https://oceana.org/wp-content/uploads/sites/18/OCEANA_Seamounts_Balearic_Islands_ENG.pdf
Martín J, Puig P, Masqué P, Palanques A, Sánchez-Gómez A (2014) Impact of bottom trawling on deep-sea sediment properties along the flanks of a submarine canyon. PLoS ONE 9(8):e104536. https://doi.org/10.1371/journal.pone.0104536
Massutí E, Reñones O (2005) Demersal resource assemblages in the trawl fishing grounds off the Balearic Islands (western Mediterranean). Scientia Mar 69(1):167–181. https://doi.org/10.3989/scimar.2005.69n1167
Massutí E, Sánchez-Guillamón O, Farriols MT, Palomino D, Frank A, Bárcenas P, Rincón B, Martínez-Carreño N, Keller S, López-Rodríguez C Díaz, J. A.
López-González N, Marco-Herrero E, Fernández-Arcaya U, Valls M, Ramírez-Amaro S, Ferragut F, Joher S, Ordines F, Vázquez J-T (2022) Improving scientific knowledge of Mallorca Channel seamounts (Western Mediterranean) within the framework of Natura 2000 Network. Diversity 14(1):4. https://doi.org/10.3390/d14010004
Massutí E, Keller S, Frank A, Farriols MT, Ferragut F, Joher S, Ordines F (2022) Huella pesquera en los montes submarinos del Canal de Mallorca (Islas Baleares, Mediterráneo occidental). Instituto Español de Oceanografía. [Unpublished report, 66 pp.]. https://intemares.es/wp-content/uploads/2023/11/HP_A2.2_Canal-Mallorca_2022.pdf
Mastrototaro F, Chimienti G, Acosta J, Blanco J, Garcia S, Rivera J, Aguilar R (2017) Isidella elongata (Cnidaria: Alcyonacea) facies in the western Mediterranean Sea: Visual surveys and descriptions of its ecological role. Eur Zoological J 84(1):209–225. https://doi.org/10.1080/24750263.2017.1315745
Maynou F, Cartes JE (2012) Effects of trawling on fish and invertebrates from deep-sea coral facies of Isidella elongata in the western Mediterranean. J Mar Biol Association United Kingd 92(7):1501–1507. https://doi.org/10.1017/S0025315411001603
McArthur MA, Brooke BP, Przeslawski R, Ryan DA, Lucieer VL, Nichol S, McCallum AW, Mellin C, Cresswell ID, Radke LC (2010) On the use of abiotic surrogates to describe marine benthic biodiversity. Estuar Coast Shelf Sci 88(1):21–32. https://doi.org/10.1016/j.ecss.2010.03.003
McClain CR, Lundsten L (2015) Assemblage structure is related to slope and depth on a deep offshore Pacific seamount chain. Mar Ecol 36(2):210–220. https://doi.org/10.1111/maec.12136
Michez N, Fourt M, Aish A, Bellan G, Bellan-Santini D, Chevaldonné P, Fabri M-C, Goujard A, Harmelin J-G, Labrune C, Pergent G, Sartoretto S, Vacelet J, Verlaque M (2014) Typologie des biocénoses benthiques de Méditerranée. Version 2 (Rapport SPN 2014-33, 26 pp.). Muséum National d’Histoire Naturelle. https://archimer.ifremer.fr/doc/00272/38315/36556.pdf
Misiuk B, Brown CJ (2024) Benthic habitat mapping: A review of three decades of mapping biological patterns on the seafloor. Estuar Coast Shelf Sci 296:108599. https://doi.org/10.1016/j.ecss.2023.108599
Miller RJ, Hocevar J, Stone RP, Fedorov DV (2012) Structure-forming corals and sponges and their use as fish habitat in Bering Sea submarine canyons. PLoS ONE 7(3):e33885. https://doi.org/10.1371/journal.pone.0033885
Mohn C, Rengstorf A, White M, Duineveld G, Mienis F, Soetaert K, Grehan A (2014) Linking benthic hydrodynamics and cold water coral occurrences: A high-resolution model study at three cold-water coral provinces in the NE Atlantic. Prog Oceanogr 122:92–104. https://doi.org/10.1016/j.pocean.2013.12.003
Moritz C, Lévesque M, Gravel D, Vaz S, Archambault D, Archambault P (2013) Modelling spatial distribution of epibenthic communities in the Gulf of St. Lawrence (Canada). J Sea Res 78:75–84. https://doi.org/10.1016/j.seares.2012.10.009
Mourre B, Aguiar E, Juza M, Hernandez-Lasheras J, Reyes E, Heslop E, Escudier R, Cutolo E, Ruiz S, Pascual A, Tintoré J (2018) Assessment of high-resolution regional ocean prediction systems using multi-platform observations: Illustrations in the Western Mediterranean Sea. In F. Chassignet, A. Pascual, J. Tintoré, & J. Verron (Eds.), New Frontiers in Operational Oceanography (pp. 663–694). GODAE Ocean View. https://doi.org/10.17125/gov2018.ch24
Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403(6772):853–858. https://doi.org/10.1038/35002501
OCEANA (2015) Expedition 2014 Balearic Islands: Cabrera National Park and Mallorca Channel Seamounts (21 pp.). Oceana, Washington (USA). https://europe.oceana.org/wp-content/uploads/sites/26/oceana_expedition2014_balearic_islands_eng_11.pdf
Ordines F, Ramírez-Amaro S, Fernandez-Arcaya U, Marco-Herrero E, Massutí E (2019) First occurrence of an Ophiohelidae species in the Mediterranean: The high abundances of Ophiomyces grandis from the Mallorca Channel seamounts. J Mar Biol Association United Kingd 99(8):1817–1823. https://doi.org/10.1017/S0025315419000808
Ordines F, Massutí E (2009) Relationships between macro-epibenthic communities and fish on the shelf grounds of the western Mediterranean. Aquat Conservation: Mar Freshw Ecosyst 19(4):370–383. https://doi.org/10.1002/aqc.969
Ordines F, Ramírez-Amaro S, Calero B, Farriols MT, Massutí E (2024) A new species of the genus Ophiomyxa (Echinodermata: Ophiuroidea: Ophiomyxidae) from the Mallorca Channel seamounts in the western Mediterranean. Zootaxa 5418(2):159–171. https://doi.org/10.11646/zootaxa.5418.2.3
Palanques A, Guillén J, Puig P (2001) Impact of bottom trawling on water turbidity and muddy sediment of an unfished continental shelf. Limnol Oceanogr 46(5):1100–1110. https://doi.org/10.4319/lo.2001.46.5.1100
Peres-Neto PR, Legendre P, Dray S, Borcard D (2006) Variation partitioning of species data matrices: Estimation and comparison of fractions. Ecology 87(10):2614–2625. https://doi.org/10.1890/0012-9658(2006)87[2614:VPOSDM]2.0.CO;2
Pérès JM (1985) History of the Mediterranean biota and the colonization of the depths. In: Margalef R (ed) Western Mediterranean. Pergamon, Oxford, pp 198–232
Pham CK, Diogo H, Menezes G, Porteiro F, Braga-Henriques A, Vandeperre F, Morato T (2014) Deep-water longline fishing has reduced impact on vulnerable marine ecosystems. Sci Rep 4:4837. https://doi.org/10.1038/srep04837
Phillips SJ, Dudík M, Schapire RE (2006) Maximum entropy modeling of species geographic distributions. Ecol Model 190(3–4):231–259. https://doi.org/10.1016/j.ecolmodel.2005.03.026
Phillips SJ, Anderson RP, Dudík M, Schapire RE, Blair ME (2017) Opening the black box: an open-source release of Maxent. Ecography 40(7):887–893. https://doi.org/10.1111/ecog.03049
Pinot JM, Tintoré J, Gomis D (1995) Multivariate analysis of the surface circulation in the Balearic Sea. Prog Oceanogr 36(4):343–376. https://doi.org/10.1016/0079-6611(96)00003-1
Pinot JM, López-Jurado JL, Riera M (2002) The CANALES experiment (1996–1998): Interannual, seasonal, and mesoscale variability of the circulation in the Balearic Channels. Prog Oceanogr 55(3–4):335–370. https://doi.org/10.1016/S0079-6611(02)00139-8
Ramón M, Abelló P, Ordines F, Massutí E (2014) Deep epibenthic communities in two contrasting areas of the Balearic Islands (western Mediterranean). J Mar Syst 132:54–65. https://doi.org/10.1016/j.jmarsys.2014.01.002
Reiss H, Birchenough S, Borja A, Buhl-Mortensen L, Craeymeersch J, Dannheim J, Darr A, Galparsoro I, Gogina M, Neumann H, Populus J, Rengstorf AM, Valle M, van Hoey G, Zettler ML, Degraer S (2015) Benthos distribution modelling and its relevance for marine ecosystem management. ICES J Mar Sci 72(2):297–315. https://doi.org/10.1093/icesjms/fsu107
Requena S, Gili JM (eds) (2014) Caracterización ecológica del área marina del Canal de Menorca: Zonas profundas y semiprofundas (100–400 m). Informe final área LIFE + INDEMARES (LIFE07/NAT/E/000732). Instituto de Ciencias del Mar – CSIC. Coordinación: Fundación Biodiversidad. 167 pp. https://digital.csic.es/handle/10261/171168
Rowden AA, Pearman TRR, Bowden DA, Anderson OF, Clark MR (2020) Determining coral density thresholds for identifying structurally complex Vulnerable Marine Ecosystems in the deep sea. Front Mar Sci 7:95. https://doi.org/10.3389/fmars.2020.00095
Sánchez F, Rodríguez Basalo A, García-Alegre A, Gómez-Ballesteros M (2017) Hard-bottom bathyal habitats and keystone epibenthic species on Le Danois Bank (Cantabrian Sea). J Sea Res 130:134–153. https://doi.org/10.1016/j.seares.2017.09.005
Sañé E, Ingrassia M, Chiocci FL, Argenti L, Martorelli E (2021) Characterization of rhodolith beds-related backscatter facies from the western Pontine Archipelago (Mediterranean Sea). Mar Environ Res 169:105339. https://doi.org/10.1016/j.marenvres.2021.105339
Sardà F, Canals M, Tselepides A, Calafat A, Flexas MM, Espino M, Tursi A (2004) An introduction to Mediterranean deep-sea biology. Scientia Mar 68(3):7–38. https://doi.org/10.3989/scimar.2004.68s37
Serrano A, González-Irusta JM, Punzón A, García-Alegre A, Lourido A, Ríos P, Blanco M, Gómez-Ballesteros M, Druet M, Cristobo J, Cartes JE (2017) Deep-sea benthic habitats modelling and mapping in a NE Atlantic seamount (Galicia Bank). Deep Sea Res Part I 126:115–127. https://doi.org/10.1016/j.dsr.2017.06.003
Standaert W, Puerta P, Mastrototaro F, Palomino D, Aguilar R, Ramiro-Sánchez B, Vázquez JT, Sánchez-Guillamón O, Marín P, Blanco J, Orejas C (2023) Habitat suitability models of a critically endangered cold-water coral, Isidella elongata, in the Mallorca Channel. Thalassas: Int J Mar Sci 39:587–600. https://doi.org/10.1007/s41208-023-00531-y
Templado J, Ballesteros E, Galparsoro I, Borja A, Serrano A, Martín L, Brito A (2012) Guía interpretativa. Inventario Español de Hábitats Marinos. Ministerio de Agricultura, Alimentación y Medio Ambiente
ter Braak CJF, Šmilauer P (2018) Canoco reference manual and user's guide: Software for ordination (version 5.10). Wageningen University & Research
Tintoré J et al (2013) SOCIB: The Balearic Islands observing and forecasting system responding to science, technology and society needs. Mar Technol Soc J 47(1):101–117. https://doi.org/10.4031/MTSJ.47.1.10
Toma M, Enrichetti F, Bavestrello G, Canese S, Cau A, Andaloro F, Angiolillo M, Greco S, Bo M (2022) Brachiopod fauna from the deep Mediterranean Sea: Distribution patterns and ecological preferences. Diversity 14(9):753. https://doi.org/10.3390/d14090753
Toma M, Bo M, Bertolino M, Canessa M, Angiolillo M, Cau A, Andaloro F, Canese S, Greco S, Bavestrello G (2024) Shedding Light on the Italian Mesophotic Spongofauna. J Mar Sci Eng 12(11):2110. https://doi.org/10.3390/jmse12112110
UNEP-WCMC (2025) Protected Area Profile for Spain from the World Database on Protected Areas (May 2025). UN Environment Programme World Conservation Monitoring Centre
Vázquez JT, Alonso B, Fernández-Puga MC, Gómez-Ballesteros M, Iglesias J, Palomino D, Roque C, Ercilla G (2015) Seamounts along the Iberian continental margins. Boletín Geológico y Min 126(2–3):483–514 & Díaz del Río
Victorero L, Robert K, Robinson LF, Taylor ML, Huvenne VAI (2018) Species replacement dominates megabenthos beta diversity in a remote seamount setting. Sci Rep 8:4152. https://doi.org/10.1038/s41598-018-22296-8
Wei C-L, Rowe GT, Escobar-Briones E, Boetius A, Soltwedel T, Caley MJ, Soliman Y, Huettmann F, Qu F, Yu Z, Pitcher CR, Haedrich RL, Wicksten MK, Rex MA, Baguley JG, Sharma J, Danovaro R, MacDonald IR, Nunnally CC, Deming JW, Montagna P, Levesque M, Weslawski JM, Wlodarska-Kowalczuk M, Ingole BS, Bett BJ, Billett DSM, Yool A, Bluhm BA, Iken K, Narayanaswamy BE (2010) Global patterns and predictions of seafloor biomass using random forests. PLoS ONE 5(12):e15323. https://doi.org/10.1371/journal.pone.0015323
Wood SN (2011) Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J Royal Stat Society: Ser B (Statistical Methodology) 73(1):3–36. https://doi.org/10.1111/j.1467-9868.2010.00749.x
Würtz M, Rovere M (eds) (2015) Introducing the Mediterranean Seamount Atlas: General aspects. In Atlas of the Mediterranean Seamounts and Seamount-like Structures (pp. 11–19). IUCN, Gland (Switzerland) and Málaga (Spain)
Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM Mixed effects models and extensions in ecology with R., Springer (2009) https://doi.org/10.1007/978-0-387-87458-6
Zimmermann NE, Kienast F (1999) Predictive mapping of alpine grasslands in Switzerland: Species versus community approach. J Veg Sci 10(4):469–482. https://doi.org/10.2307/3237182

No competing interests reported.

FranketalSuppMat.docx

Download PDF

Editorial decision: Revision requested
15 Dec, 2025
Reviews received at journal
09 Dec, 2025
Reviewers agreed at journal
12 Nov, 2025
Reviewers invited by journal
12 Nov, 2025
Editor assigned by journal
29 Jul, 2025
Submission checks completed at journal
29 Jul, 2025
First submitted to journal
25 Jul, 2025

You are reading this latest preprint version

Benthic habitats of western Mediterranean seamounts: Spatial modelling and environmental drivers

Status:

Version 1

Abstract

Figures

1. INTRODUCTION

2. MATERIAL AND METHODS

2.1. Sampling

2.2. Data collection and processing

2.2.1. Biological data: habitat-forming species and communities

2.2.2. Environmental and fishing variables

2.2.3. Sampling unit

2.3. Predictive modelling and mapping

2.4. Relationships with environmental parameters

3. RESULTS

3.1. Predictive models and habitat mapping

3.2. Habitats relationship with environmental variables

4. DISCUSSION

4.1. Habitat zonation and spatial distribution

4.1.1. Summits

4.1.2. Flanks

4.1.3. Adjacent bottoms

4.2. Environmental drivers influencing benthic habitats

4.3. Towards protection and management

Declarations

Funding

Author Contribution

Acknowledgement

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1