Rhodolith Beds Heterogeneity along the Apulian Continental Shelf (Mediterranean Sea)

Rhodolith beds represent a key habitat worldwide, from tropical to polar ecosystems. Despite this habitat is considered a hotspot of biodiversity, providing a suite of ecosystem goods and services, still scarce quantitative information is available thus far about rhodolith beds occurrence and ecological role, especially in the Mediterranean Sea. This study reports the composition and patterns of distribution of rhodolith assemblages found in four study areas along ca. 860 km of coast in the Central Mediterranean Sea. These rhodolith beds were studied for the first time and significant differences at all spatial scales have been highlighted, documenting the high variability of this habitat. Rhodolith species composition, morphology and distribution have been discussed considering the potential role of environmental factors in driving these patterns. The need for improving their protection is discussed to complement present conservation and management initiatives, particularly in the frame of the EU Marine Strategy Framework Directive.


Introduction
Human activities largely affect natural systems, but their cumulative effects on ecological communities remains largely unknown [1]. A low diversity of species can affect ecosystems sensitivity to multiple stressors and enhance the impacts on ecosystem functioning [2]. For this reason, understanding patterns of distribution of species, communities and habitats is a priority to tease apart natural patterns from those driven by human disturbances [3]. Biodiversity in marine ecosystems is related to habitat heterogeneity [4], which is enhanced by ecosystem engineers, i.e., organisms that modify, maintain or destroy habitats [5][6][7]. Rhodolith-forming algae grow as unattached nodules, often acting as ecosystem engineers and forming aggregations known as Rhodolith Beds (RBs) on flat or gently-sloped seabed [5,6,8]. RBs provide habitat for several associated species from the tropics to polar latitudes [6,[9][10][11][12], as well as a suite of ecosystem goods and services whose economic importance

Habitat Mapping
Mapping of the four study areas was carried out using Side-Scan Sonar (SSS; Klein 3000 and K3900) and MultiBeam EchoSounder (MBES; R2Sonic 2022 and RESON Seabat 8125) [8,17,32,67], for a total of 100 km 2 mapped. SSS data were acquired with a double frequency of 100 and 500 kHz, simultaneously, using a swath width of 100 m, with 50% of overlap between adjacent lines. SSS data were processed using the CARIS HIPS and SIPS software, according to the following steps: data conversion; navigation processing; slant range correction; time varying gain application; attitude data editing; data mosaicing; data output. Geo-referenced, grey-tone acoustic images of the seafloor with a 0.2 m cell resolution were produced. The four study areas have been chosen according to the above-mentioned old records, considering the poor predictability of RB occurrence [54] Figure 1d). In particular, Tremiti Archipelago is located 12 nautical miles north of the Gargano promontory (Adriatic Sea) and is characterized by cold, oxygenated and trophic-carrying water masses, coming from the north and proceeding southwards, that support rich marine communities including extensive coral forests on coralligenous bioconstructions [55][56][57]. The MPA shows a gradient of restrictions and is divided into three main zones: Zone A (no take, no entry zone), Zone B (highly protected zone; anchoring and recreational fishing are forbidden, while professional fishing is regulated), and Zone C (partially protected zone; all human activities are allowed and professional fishing is regulated). The area northwest Otranto is located in proximity of the Otranto Strait (Adriatic Sea), where the scarp that bounds the Salento Peninsula slopes eastward down to the sea and dips at relatively high angle down to about 50 m depth [53]. Patchy marine bioconstructions, mostly coralligenous, are present down to 100 m depth [58]. Off Gallipoli (Ionian Sea) the shelf break begins at ca. 120 m depth and is anticipated by three terraced surfaces located at 25-30, 50-60 and 110-120 m depth [59]. The seabed is characterized by coralligenous outcrops interspersed with coastal detritic and muddy bottoms [50,53]. Proceeding westwards in the Gulf of Taranto, the Armeleia shoal is characterized by a slope of rocks and coralligenous bioconstructions from 10 to 40 m depth. At the base of the shoal, large coralligenous outcrops are patchy present on a detritic bottom.
The mean surface circulation in the study areas is induced by north-coming cold winds in the Adriatic Sea, by fresh water input from the Po River in the northern sector, and by differences in water density between the Ionian and the Adriatic seas in the southern sector [53,60]. These surficial currents are active down to about 150 m depth [57,61] and consist of an elongated basin-wide cyclonic gyre, with southward currents near the Italian coasts [62,63], and sub-basin cyclonic-gyre cells [60]. In particular, they run along-shore north of the Gargano Promontory [64], leave the coast south of the same promontory and re-approach the coast parallel to the Murge Plateau near Bari [65], contour the Salento Peninsula and enter the Ionian Sea inducing an along-shore cyclonic gyre in the Taranto Gulf of a broadly lower intensity [66].

Habitat Mapping
Mapping of the four study areas was carried out using Side-Scan Sonar (SSS; Klein 3000 and K3900) and MultiBeam EchoSounder (MBES; R2Sonic 2022 and RESON Seabat 8125) [8,17,32,67], for a total of 100 km 2 mapped. SSS data were acquired with a double frequency of 100 and 500 kHz, simultaneously, using a swath width of 100 m, with 50% of overlap between adjacent lines. SSS data were processed using the CARIS HIPS and SIPS software, according to the following steps: data conversion; navigation processing; slant range correction; time varying gain application; attitude data editing; data mosaicing; data output. Geo-referenced, grey-tone acoustic images of the seafloor with a 0.2 m cell resolution were produced.
MBES data were acquired with the echo sounder mounted on a side pole solidly installed on vessel side. A SV probe was installed close to MBES transducer and connected to the system in order to measure sound speed in salt water to a real time correction of every acquired sounding. MBES data were acquired with a maximum swath angle of 120 • and a frequency from 450 kHz to 210 kHz depending on water depth. Data were processed using the Teledyne PDS2000 and CARIS HIPS and SIPS software to produce a digital terrain model (DTM) and a backscatter mosaic of the seafloor that was mapped and described based on morphometric characterization. The DTM was characterized by: 0.2 m cell resolution up to 10 m depth; 0.5 m cell resolution at 10-20 m depth; 1.0 m cell resolution for depth >20 m. The acoustic data were calibrated, interpreted and properly validated based on 20 stations of ground-truthing for each study area, using a drop camera system (steel slide with high-definition camera and led light of 6000 lumen). Videos were analyzed using Adobe Premiere Pro software. The combination of SSS and MBES data was used to identify and map the RBs, as well as to evaluate their possible extent.

Data Collection
Three sites, distant about 500-1000 m one from the others, were randomly selected within the RBs identified and mapped at each study area with the acoustic surveys (Figure 1a-d). Three linear transects of 200 m each, distant about 50-100 m one from the others, were carried out at each site by using Remotely Operated Vehicle (ROV) ( Table S1). Surveys were carried out on board the vessel Issel (property of CoNISMa-Consorzio Nazionale Interuniversitario per le Scienze del Mare), using the ROV Prometeo. The ROV was equipped with a low-definition camera for navigation and a 4 K video camera for the detailed observation of the rhodoliths, as well as a depth sensor, a compass for navigation, led lights with a maximum scene illumination of 13,000 lumen, a geographic positioning system and two laser beams for size and area measurements [68].
ROV videos were analysed using Adobe Premiere Pro software. Sequences with bad visibility, due to water turbidity or distance from the seabed, were discarded. A total of 20 frames were randomly extracted from each video transect. Each frame defined an area of 2.5 ± 0.2 m 2 , according to the minimal area proposed by Weinberg [69] for the study of Mediterranean benthic communities and recently used for ROV imaging [57,68,70]. The rhodolith cover percentage and the proportion of living thalli were estimated for each frame by visual inspection as an approximate measure of the algal growth rate and vitality [37]. We considered as dead thalli those rhodoliths that where white or whitish in colour, based on expert judgement.
Visual inspections were validated through one quantitative sample per transect (3 samples per site), collected using a Van Veen grab (0.1 m 2 of surface and 0.02 m 3 of volume), according to the monitoring protocol for Mediterranean RBs developed within the MFSD [27]. Samples were sieved through a 0.5 mm mesh size. Living rhodolith thalli were sorted, dried with absorbent paper and preserved on-board in silica gel [38]. The taxa composing the algal nodules were identified at the lowest possible taxonomic level (i.e., species or genus).
Qualitative sub-samples of few rhodoliths were collected along each transect using Van Venn grab in order to perform algae taxonomic identification. Based on this data and on the general morphology of the rhodoliths samples, the overall contribution of each algal taxon to the total rhodolith cover percentage was estimated analysing the frames extracted from each video (20 frames per transect, 60 frames per site, 180 frames per area).

Taxonomic Identification
Calcareous algae were analysed by scanning electron microscopy (SEM) for detailed anatomical observation. Fragments of samples were sonicated using a Vitec sonicator (Carlsbad, CA, USA) to remove sediments, mounted on aluminium stubs with acrylic adhesive and coated with gold/palladium with a S150 Sputter Coater (Edwards, Crawley, UK) [71]. Then, observation was carried out using a LEICA Steroscan 430i scanning electron microscope (Leica, Cambridge, UK). The identification of non-geniculate red algae follows the most updated scientific literature, and algal taxonomy follows Algaebase [72].

Statistical Analysis
Statistical analyses were carried out using PRIMER v.6 software [73]. Permutational multivariate analysis of variance (PERMANOVA) [74] was used to test differences among algal assemblages at different spatial scales, based on the cover percentage of the species observed and identified, using the following hierarchical sampling design: Area (Ar, as random factor with 3 levels), Site (Si, as random factor with 3 levels) and Transect (Tr, as random factor with 3 levels), with n = 20 frames by each ROV video transect.
Data were analysed using the Bray-Curtis dissimilarity measure on untransformed data through 9999 permutations of residuals under a reduced model [75], while univariate analyses based on Euclidean distances [74] were performed on the rhodolith cover percentage and the proportion of living thalli. Mean squares calculated by PERMANOVA were used to estimate multivariate variance components associated at each spatial scale [76]. Non-metric multidimensional scaling (MDS) ordination plot was produced in order to visualize multivariate patterns in assemblages across scales. Furthermore, permutational analyses of multivariate dispersion (PERMDISP) were also performed to test the heterogeneity of multivariate dispersions among sites and among transects [77,78]. In order to detect which algal species contributed most to dissimilarity among the areas, a similarity percentage (SIMPER) routine was performed [79]. The assemblage structure of the different sites was then compared by plotting the cumulative dominance vs species rank in order to obtain k-dominance curves based on cover percentage [80].

Mapping and Description of Rhodolith Beds
Habitat mapping activities allowed to identify areas characterized by rhodolith-forming algae, providing a reliable indication for the allocation of ROV transects and grab samplings. RBs showed a typical backscatter response characterized by a texture of bands of high and low signal, mostly due to the general patchiness of the habitat (Figure 2a). Areas characterized by higher cover of rhodoliths where identifiable by a higher backscatter, while low-cover RBs were not always distinguishable from detritic seabed without rhodoliths, such as the coastal detritic biocoenosis [49]. Based on acoustic data validated by visual surveys, we identified high-density RBs as those with a higher backscatter, corresponding to a cover broadly higher than 20-30%. On the contrary, areas with a lower backscatter and mostly characterized by a cover of 10% or less, often interspersed with high-density RBs, were considered low-density RBs. Coralligenous bioconstructions and muddy bottoms were also easily distinguishable based on the backscatter (Figure 2a).  Table 1).
The rhodolith cover percentage in Armeleia was larger than the coverages found in Tremiti and in Gallipoli (Table 1), while the proportion of living thalli was higher in Tremiti than Gallipoli and Armeleia (Table 2). Both the rhodolith cover percentage and the proportion of living thalli showed significant differences at the scales of transects and sites (Table 2). PERMDISP analyses revealed significant differences in multivariate dispersion across scales, both across transects (rhodolith cover percentage: F= 6.765 (p = 0.0001); proportion of living thalli: F: 6.650 (p = 0.0001)) and sites (rhodolith cover percentage F= 6.891 (p = 0.0001); proportion of living thalli: F= 14.842 (p = 0.0001)).  Consistently with both SSS and MBES backscatter, rhodoliths were observed in all the ROV transects and grab samples, within the four study areas. RBs (sensu [6,17]) were observed at Tremiti Islands MPA, off Gallipoli and at Armeleia shoal. On the contrary, only occasional and densely-branched nodules of Lithophyllum racemus, characterized by short ramifications, were found northwest Otranto from 38 to 44 m depth (Table 1). For this reason, this latter area was not included in the analysis. Table 1. Mean cover percentage and standard deviation of each rhodolith-forming species observed in the four study areas, with indication of total cover percentage and the proportion of living thalli for each area. The RB at Tremiti Islands MPA was patchy distributed around the archipelago as a mosaic of rhodoliths and coralligenous, and as RB sensu stricto from 15 to 48 m depth. The RB covered a total surface of 2.47 km 2 (1.70 km 2 of high-density RB) and was present in both the B and C zones of the MPA, such as the highly protected and the partially protected zone, respectively (      The RB off Gallipoli was found between 36-45 m depth, mainly as a mosaic of rhodoliths and coralligenous habitat (i.e., RB and patchy coralligenous outcrops). The RB covered an area of 4.04 km 2 (2.90 km 2 of high-density RB) ( Figure 2c) and was dominated by unattached branches morphotype ( Figure 3). Rhodoliths were characterized by short and numerous branches (mostly L. racemus), or by few, long and thin ramifications (e.g., Lithothamnion corallioides). The dominant species was L. racemus (Figure 4i-l), while L. corallioides, L. crispatum and T. pustulatum were less abundant in terms of cover percentage (Table 1).
The rhodolith cover percentage in Armeleia was larger than the coverages found in Tremiti and in Gallipoli (Table 1), while the proportion of living thalli was higher in Tremiti than Gallipoli and Armeleia (Table 2). Both the rhodolith cover percentage and the proportion of living thalli showed significant differences at the scales of transects and sites (Table 2). PERMDISP analyses revealed significant differences in multivariate dispersion across scales, both across transects (rhodolith cover percentage: F = 6.765 (p = 0.0001); proportion of living thalli: F: 6.650 (p = 0.0001)) and sites (rhodolith cover percentage F = 6.891 (p = 0.0001); proportion of living thalli: F = 14.842 (p = 0.0001)). Table 2. Results of PERMANOVAs testing for spatial differences in the proportion of living thalli and the rhodolith cover percentage at scales of area, site and transect. Estimates of multivariate variation are given for each spatial scale. df = degrees of freedom; MS = mean sum of squares; Pseudo-F = F value by permutation; p (perm) = p-values based on 9999 random permutations of appropriate units.

Proportion of Living Thalli
Rhodolith Cover

Heterogeneity of Rhodolith Beds
The PERMANOVA analysis highlighted significant differences in the algal cover percentage from the three RBs studied, at all the considered scales (Table 3). In addition, estimates of multivariate variation showed the largest variation at area scale, followed by residual variation, indicating multivariate heterogeneity among sampling units, whereas the contributions of site and transect scales resulted marginal (Table 3; Figure 5). Table 3. Results of PERMANOVAs testing for spatial differences in structure of algal assemblages at area (Ar), site (Si) and transect (Tr) scales, based on the algal cover percentage from ROV imaging. Analyses based on Bray-Curtis dissimilarities from untransformed data. Estimates of multivariate variation are given for each spatial scale. df = degrees of freedom; MS = mean sum of squares; Pseudo-F = F value by permutation; p (perm) = p-values based on 9999 random permutations of appropriate units. * p < 0.05; *** p < 0.001. MDS plot revealed important differences between the assemblage of Armeleia shoal and the other two areas. Pearson correlations coefficients (>0.6) indicated that the observed differences were driven by L. minervae and L. corallioides, which mainly characterized the assemblages from Armeleia shoal, whereas RBs from Tremiti Islands and Gallipoli were mainly characterized by L. racemus ( Figure 6). Higher variability among replicates was also observed at Tremiti Islands MPA and Gallipoli, compared to Armeleia shoal ( Figure 6).

Source
Results of PERMDISP showed significant differences in multivariate dispersion across scales. More specifically, significant differences in the spatial heterogeneity of assemblages across transects (F = 11.396 [p = 0.0001]) and sites (F = 26.588 [p = 0.0001]) were observed (Figure 7a,b). MDS plot revealed important differences between the assemblage of Armeleia shoal and the other two areas. Pearson correlations coefficients (>0.6) indicated that the observed differences were driven by L. minervae and L. corallioides, which mainly characterized the assemblages from Armeleia shoal, whereas RBs from Tremiti Islands and Gallipoli were mainly characterized by L. racemus ( Figure 6). Higher variability among replicates was also observed at Tremiti Islands MPA and Gallipoli, compared to Armeleia shoal ( Figure 6).  Results of PERMDISP showed significant differences in multivariate dispersion across scales. More specifically, significant differences in the spatial heterogeneity of assemblages across transects (F = 11.396 [p = 0.0001]) and sites (F = 26.588 [p = 0.0001]) were observed (Figure 7a,b).
The SIMPER analysis revealed a high dissimilarity in the algal assemblages among the study areas, reaching the 97% between Gallipoli vs Armeleia (Table 4). RBs from Armeleia shoal and Tremiti Islands MPA were also characterized by the patchy presence of green seaweeds belonging to the family Caulerpaceae, settled on the rhodoliths and locally abundant. For this reason, Caulerpaceae have been included in the SIMPER analysis. In particular, Caulerpa prolifera was found only at Tremiti Islands MPA (0.4%), while C. cylindracea both at Armeleia shoal (4.58%) and Tremiti Islands MPA (8.12%) ( Table 4).    K-dominance curves showed that rhodolith taxa dominance at Gallipoli and Armeleia markedly differs from curves of Tremiti Islands MPA, this latter being characterized by less dominant taxa (Figure 8). At Armeleia shoal the most abundant species in terms of cover was L. corallioides, with a dominance of 90% approximately. On the contrary, L. racemus resulted the dominant species at Gallipoli and Tremiti Islands MPA, with a dominance of 90% and 40-60%, respectively. K-dominance curves showed that rhodolith taxa dominance at Gallipoli and Armeleia markedly differs from curves of Tremiti Islands MPA, this latter being characterized by less dominant taxa (Figure 8). At Armeleia shoal the most abundant species in terms of cover was L. corallioides, with a dominance of 90% approximately. On the contrary, L. racemus resulted the dominant species at Gallipoli and Tremiti Islands MPA, with a dominance of 90% and 40-60%, respectively.

Rhodolith Beds Variability from Small to Large Scale
The combination of visual surveys and acoustic data represented the best approach for the identification of RBs, and the high backscatter registered in association with this habitat was in

Rhodolith Beds Variability from Small to Large Scale
The combination of visual surveys and acoustic data represented the best approach for the identification of RBs, and the high backscatter registered in association with this habitat was in accordance with recent literature (e.g., [8,67,81]). However, backscatter can vary according to many factors such as, among the others, incident angle, sediment type, grain size, seabed slope and roughness [81,82], thus ground truth is essential for a proper interpretation.
The adoption of a hierarchical nested design for the analysis of spatial patterns based on ROV imaging and samples over small (~10 −1 m) to large (~10 6 m) spatial scales allowed to investigate the composition of the newly-identified rhodolith assemblages across the coast of Apulia (Central Mediterranean Sea). Multivariate analyses highlighted significant differences in the rhodolith assemblages at all the considered scales, with the largest variation among areas. These differences are driven by four main species, such as L. minervae, L. corallioides, L. racemus and N. brassica-florida ( Figure 3; Table 4). The analysis of multivariate variation showed that areas added a significant contribution of variance above that of sampling units that are, however, also contributing to the overall variability. This variability at area level is likely due to the distance and the environmental differences among each area, underlining a strong heterogeneity and the presence of specific rhodolith communities in the different areas. Rhodoliths were also characterized by significant differences at the smaller scales in both rhodolith cover percentage and proportion of living thalli (Table 2). In addition, wave-induced turbulence, light intensity and sedimentation together with depth range, can represent environmental drivers determining the presence, structure and functioning of rhodolith beds [83]. Moreover, fishing activities can also influence the size and the morphology of the rhodoliths.
Our result further reinforces the idea that small-scale processes are as important as large-scale ones in generating patterns in rhodolith assemblages, as also observed in other marine habitats such as intertidal and subtidal turfing algal assemblages, soft-sediment macrofauna and rocky shore communities [42,43,84,85]. Moreover, our results confirm that Mediterranean RBs are featured by complex spatial patterns and a high diversity of shapes and morphologies, determining a substantial heterogeneity.
The principal ecosystem engineers and the main contributor in dissimilarities were L. racemus and L. corallioides, the latter included in the Annex V of the EU Habitat Directive [29]. In particular, L. racemus was the dominant species in all the study areas except from Armeleia shoal, where L. corallioides was dominant. The rhodolith assemblage was particularly heterogeneous at Tremiti Islands MPA where, despite the large cover of L. racemus and N. brassica-florida, it was not observed the marked dominance of one species. RBs shallower than 25 m depth, as Tremiti's one, are also rather uncommon in the Mediterranean basin, being mainly present in the northern Adriatic Sea and in the Gulf of Gabés (Eastern Mediterranean, Tunisia) [25]. Indeed, most Mediterranean RBs known thus far occur between 30 and 75 m, although they have been broadly found from 9 to 150 m of depth [25].

Rhodolith Morphotypes and Main Hydrodynamic Regime
The clear separation of Armeleia shoal from the other two areas (Figure 6) suggested that a combination of environmental drivers (e.g., currents and sedimentation) possibly shaped a different rhodolith assemblage in this area in terms of species composition, rhodolith growth forms and cover. Besides temperature, wind waves, nutrients and irradiance levels as a function of photosynthesis, calcification and respiration process [86][87][88], bottom currents represent important variables governing rhodolith occurrence and development. In fact, different hydrodynamic regimes can influence RB composition and rhodolith morphotype [33,89]. In particular, spherical and densely-branched rhodoliths seem to be typical of exposed sites (i.e., subject to strong bottom currents), while open-branched rhodoliths are found in more stable environments (i.e., subject to moderate bottom currents) [22,23,44]. The dominant presence of pralines and boxworks at Tremiti Islands MPA could be due to the relevant exposure of the archipelago to strong water currents coming from the north Adriatic [55][56][57]. Both densely-branched rhodoliths (L. racemus) and open-branched ones (L. corallioides) were observed off Gallipoli. This variety may be driven by the different local conditions due to the presence of large coralligenous outcrops scattered on the coastal detritic bottom with RBs (mosaic of coralligenous and RBs) that may affect water masses movements on the seabed. Finally, RBs at Armeleia shoal were mainly characterized by unattached branches with few, long and thin ramifications, likely due to a more hydrodynamically stable environment [53,66]. These results seemed to support the hypothesis that Mediterranean RBs are mainly present around islands and capes (as Tremiti Islands), around banks and shoals (as Armeleia shoal), as well as on top of submarine plateau and on marine terraces (as off Gallipoli) [25].

Threats and Conservation
Habitats featured by high variability of assemblages and complex patterns of distribution require different protected sites for a representative conservation policy. Heterogeneity in rhodolith species composition, morphologies and cover resulted locally very high in all the RBs, representing eligible sites for conservation measures, such as the institution and/or the implementation of Natura2000 sites according to the EU Habitat Directive [29]. To that regard, areas as Armeleia shoal are good candidates for strategic protection initiatives due to the co-existence of priority habitats [27,29], such as coralligenous bioconstructions and RBs.
The observed high patchiness of the rhodolith assemblage at Tremiti Islands could be enhanced by a number of anthropic stressors including high touristic frequentation, anchoring activities and some fishing practices that occur in the area [47,57], albeit no evident signs of destructive fishing practices (e.g., trawl marks) have been observed on the RB. However, trawling is allowed only below 50 m depth in the Mediterranean Sea and our survey was carried out mostly shallower than this depth, so it is assumed that no trawling activities are carried out on the studied RBs. Beside the direct effect of fishing activities, the indirect human pressures might negatively influence the habitats as reported along the Tyrrhenian coasts where the great amount of marine litter was found to affect seabed [90] and litter abundance was negatively correlated to rhodolith cover [91]. However, the conservation status of RBs in a 25-year-old no-take area of the north-western Mediterranean (Columbretes Islads Marine Reserve) [92] underlined a beneficial protection from fishing activities on RBs cover and species richness, although the proportion of living rhodoliths was similar in protected and in unprotected areas.
Further evidences of human pressures could be highlighted by the presence of the green algae Caulerpa cylindracea. This alien and invasive species was frequently observed in many sites, and resulted locally very common at Tremiti Islands MPA. Local heavy overgrowth of C. cylindracea on RBs may be facilitated by anthropogenic impacts [93] and it could affect the rhodolith assemblage [94,95]. In particular, besides the shading due to the settling on the rhodoliths, C. cylindracea may significantly influence the quantity and biochemical composition of sedimentary organic matter across the RBs [96,97].
The ongoing ocean acidification and rising of water temperature might hamper the formation of RBs in the near future or result in the degradation of present assemblages [98][99][100][101][102][103]. As recently highlighted for other calcifying habitat formers such as cold-water corals [3,104], knowledge about patterns of distribution and main habitat features are fundamental to assess ongoing and future changes [105]. Understanding the status of RBs without consistent information about species compositions, dominant morphologies, proportion of living thalli, habitat features and disturbances (both natural and anthropogenic) shaping presence and distribution of RBs is very unlikely. This information, together with the assessment of RBs sensitivity to human disturbance, is essential for the proper protection and management of this productive marine habitat.

Conclusions
Our results showed the potential of applying coordinated monitoring at regional and national level that, in the near future, could confirm the presence of complex spatial and diversity patterns possibly driven by a combination of natural processes and anthropic pressures (e.g., organic inputs, coastal human activities, fishing pressures). Visual techniques proved to be effective, but sampling is still needed for taxonomic identification and for both calibration and validation of remotely-collected data (both geophysical and visual). The identification of strategic areas which are as representative as possible of the different Mediterranean RBs is essential, together with the application of ad hoc conservation measures to protect this unique marine habitat.
Supplementary Materials: The following are available online at http://www.mdpi.com/2077-1312/8/10/813/s1, Table S1. Coordinates and depth range of the ROV transects carried out in the four study areas, with three sites per area and three transects per site.