Diversity Loss in Coralligenous Structuring Species Impacted by Fishing Gear and Marine Litter

: Coralligenous structuring species (CSS) form a group of marine megabenthic species with an engineering capacity. Since they are highly vulnerable to anthropogenic activities, they have been selected for the Marine Strategy Framework Directive (MSFD) monitoring programs. The pressure and impact of ﬁshing gear and marine litter on these species were evaluated through the image analysis of 54 remotely operated vehicle (ROV) routes along the Campania coasts (Tyrrhenian Sea, Italy). CSS density was calculated as the number of colonies/100 m 2 . Anthropogenic pressure was estimated as the frequency of frames showing longline, nets, other gear, plastic objects, metal objects, and other litter; while the impact was expressed as the frequency showing necrosis/epibiosis, broken/upturned and covered/entangled colonies. Cnidaria dominate in the Napoli, Campanella and Capri areas, while Bryozoa dominate in Cilento N and Cilento S areas. Campanella and Capri appeared to be the least heterogeneous despite their higher CSS densities, which was possibly related to the dominance of a few species. These areas were the most affected by showing the highest numbers of ﬁshing gear (longlines) and marine litter (metal objects) recorded, amongst which longlines are the most abundant. In addition, these ﬁshing areas are either close to a large urban center or located along popular touristic routes. In all the areas, colonies with necrosis/epibiosis (CNE) impact are present with low-moderate values, while the category gears covering/entangling (GCE) impact prevails in the Campanella and Capri areas, and this is strictly connected to the high presence of ﬁshing gear.


Study Area
The study areas (Figure 1) are located along different parts of the Campanian coasts (south-central Tyrrhenian Sea, Italy). A total of six areas, and three randomly replicated sites for each area, were investigated: "the Napoli area" (Nisida, Penta Palummo and Miseno sites), between 20 m and 73 m depth; "the Ischia area" (Secca Forio, Caruso and Casamicciola sites), between 26 m and 60 m depth; "the Campanella area" (Ieranto, Secchitiello and Punta Campanella sites) between 34 m and 102 m depth; "the Capri area" (S. Marcellino, P. Carena and P. Arcera sites), between 30 m and 102 m depth; "the Cilento N area" (Licosa A, Licosa B and Licosa C sites), between 40 m and 95 m depth, and "the Cilento S area" (Acciaroli C, Acciaroli D and Acciaroli F sites), between 49 m and 78 m depth. The Napoli area consists of a sand/mud substrate with scattered coralligenous banks; Ischia island is mostly characterized by a hard bottom due to the presence of cliffs and rocky outcrops of volcanic origins; the sites located in the Campanella area are represented by two rocky cliffs (P. Campanella and Ieranto) and a shoal (Secchitiello); the Capri The Napoli area consists of a sand/mud substrate with scattered coralligenous banks; Ischia island is mostly characterized by a hard bottom due to the presence of cliffs and rocky outcrops of volcanic origins; the sites located in the Campanella area are represented by two rocky cliffs (P. Campanella and Ieranto) and a shoal (Secchitiello); the Capri area also consists of steep rocky cliffs; finally, the Cilento N and Cilento S areas are characterized by a succession of rocky outcrops and coralligenous banks [53][54][55].

Field Activities
The investigation was carried out during several surveys in 2017 and 2018, aimed at assessing the ecological status of coralligenous habitats, following the MSFD protocols [56]). The study was carried out through a remotely operated vehicle (the ROV Ageotec, mod. Perseus), equipped with an HD camera (DVS-3000 high definition), a navigation camera with an underwater positioning Ultra Short Base Line System (USBL) interfaced with the on-board navigation system, two spotlights and two laser pointers providing a 14.5-cm scale for the measurement of the frame area and the size of the organisms. ROV moved at approximatively 0.3-0.4 knots and at a distance of about 0.5-1 m from the sea bottom, with an average visual width of 0.5 m.
Overall, 6 areas with 3 sites each (18 sites in total) were monitored; within each site three 300 m routes were performed (54 routes in total).

Data Management
The ROV videos were analyzed using the VisualSoft ® software, rending the HD videos overlapped with the navigation data. The frames used for the analysis, were obtained by video defragmentation each 10 sec with the DVDVideoSoft ® software. Taking into account the length of each route (around 300 m), a number of 74 images per route was chosen for uniformity between the different routes. Only high-quality images were analyzed.
The density of coralligenous structuring species identified by ROV-imaging technique (CSS, sensu MSFD protocols) was calculated as the number of colonies of each species on 100 m 2 (n. CSS colonies/100 m 2 ) and compared with CSS frequency (percentage times that CSS appeared in each route of the total images number with coralligenous habitat), in order to assess the spatial distribution of species [57]. Moreover, quantitative dominances (Di), as the ratio of a specific CSS density with total density, were also calculated for each area.
Anthropogenic pressure on coralligenous habitats was assessed by frequencies of coralligenous bioconstructions presenting fishing gear or marine litter, calculated as a percentage of frames. Likewise, the anthropogenic impacts were assessed as frequency, calculated as a percentage of frames with damaged coralligenous structuring species.
Anthropogenic activities impacting structuring species colonies can be direct or indirect and they were classified into three categories [34]: the direct impacts are BUC (broken/upturned colonies) and GCE (gears covering/entangling), while the indirect impact is CNE (colonies with necrosis/epibiosis).

Statistical Analyses
The experimental design, with fishing gears and marine litter not causing damage to the coralligenous communities as the null hypothesis, used statistical analyses involving two factors: area (Ar: fixed, six levels) and site (Si: random, 18 levels) with n = 3 (routes per site). Prior to the analyses, abundances were square-root transformed to reduce the weight of very scanty CSS.
Univariate and multivariate analyses were performed on total densities and single species densities, respectively. In particular, following the Terlizzi et al. (2007) methodology [58], univariate analyses of variance PERMANOVA [59] based on Euclidean distance, was conducted on CSS densities, while the Bray-Curtis similarity was used for the remaining analyses [60]. Subsequently, a pairwise test (corresponding exactly to Gosset's original t statistic in PRIMER v.6 + PERMANOVA (software) [61] was performed in order to detect differences among areas.
Multivariate patterns were visualized through the multivariate canonical analyses of principal coordinates (CAP) [62]  index > 0.4 were also shown in CAP plot. Distance-based permutation multivariate analyses of variance (PERMANOVA) were carried out on densities in order to test for differences in assemblage CSS structure among areas and sites.
Anthropogenic pressures patterns were visualized by the multivariate three-dimensional non-metric multidimensional scaling (3D-nMDS) of Areas centroids. CSS densities were related to them through the multivariate distance-based linear modelling (DistLM [64]), using Step-wise as section procedure and adjusted R2 as a selection criterion.
Finally, fishing gears (independent variable) being significantly correlated with CSS densities were related by linear regression (ordinary LS) to direct impacts (GCE + BUC), indirect impacts (CNE) and heterogeneities (dependent variables).
All the statistical analyses were carried out through PRIMER v.6 + PERMANOVA software [61], except the linear regressions, which were performed with Past4 ® software [65].
Such findings were confirmed by the CAP analysis (Figure 4), showing: a cluster composed by Campanella and Capri areas, polarized in the positive part of the CAP 1 axis, strongly related with cnidarian species, and a cluster composed by Cilento N and Cilento S areas, polarized in the negative part of the CAP 1 axis, strongly related with bryozoan species. Napoli and Ischia areas are scattered along the whole CAP 1 axis. Diversity 2021, 13, x 7 of 16 Such findings were confirmed by the CAP analysis ( Figure 4), showing: a cluster composed by Campanella and Capri areas, polarized in the positive part of the CAP 1 axis, strongly related with cnidarian species, and a cluster composed by Cilento N and Cilento S areas, polarized in the negative part of the CAP 1 axis, strongly related with bryozoan species. Napoli and Ischia areas are scattered along the whole CAP 1 axis.
Multivariate PERMANOVA analysis on the species densities dataset showed significant differences both among sites (p = 0.0002) and areas (p = 0.0066); in particular, pairwise tests show that Cilento N and Cilento S were significantly different from Campanella and Capri (p < 0.05). Indeed, the PERMDISP analysis (p = 0.0076), and measure of β-diversity in terms of small-scale heterogeneities, showed the lowest values for Campanella and Capri and the highest values for the other areas ( Figure 5).  Such findings were confirmed by the CAP analysis (Figure 4), showing: a cluster composed by Campanella and Capri areas, polarized in the positive part of the CAP 1 axis, strongly related with cnidarian species, and a cluster composed by Cilento N and Cilento S areas, polarized in the negative part of the CAP 1 axis, strongly related with bryozoan species. Napoli and Ischia areas are scattered along the whole CAP 1 axis.
Multivariate PERMANOVA analysis on the species densities dataset showed significant differences both among sites (p = 0.0002) and areas (p = 0.0066); in particular, pairwise tests show that Cilento N and Cilento S were significantly different from Campanella and Capri (p < 0.05). Indeed, the PERMDISP analysis (p = 0.0076), and measure of β-diversity in terms of small-scale heterogeneities, showed the lowest values for Campanella and Capri and the highest values for the other areas ( Figure 5). Multivariate PERMANOVA analysis on the species densities dataset showed significant differences both among sites (p = 0.0002) and areas (p = 0.0066); in particular, pairwise tests show that Cilento N and Cilento S were significantly different from Campanella and Capri (p < 0.05). Indeed, the PERMDISP analysis (p = 0.0076), and measure of β-diversity in terms of small-scale heterogeneities, showed the lowest values for Campanella and Capri and the highest values for the other areas ( Figure 5).   Anthropogenic pressure is mostly represented by longlines (67.6%), followed by nets (20.4%); while metal objects were the most frequent among marine litter (0.1%) ( Figure 6). Capri is the most stressed area with the presence of gear and litter in the 32.7% of frames; while, Cilento N represents the less stressed site, with value of 4.6%. In particular, the medium values of fishing gear and marine litter for each area resulted: Ischia 14.2 ± 7.9%, Napoli 24.3 ± 26.3%, Campanella 30.3 ± 23%, Capri 32.7 ± 23.4%, Cilento N 4.5 ± 4.8%, and Cilento S 4.7 ± 4.8%. The nMDS ordination plot (Figure 7) performed on the different areas centroids show three evident clusters: Campanella-Capri, Cilento N-Cilento S, and in the middle Napoli- Anthropogenic pressure is mostly represented by longlines (67.6%), followed by nets (20.4%); while metal objects were the most frequent among marine litter (0.1%) ( Figure 6). Capri is the most stressed area with the presence of gear and litter in the 32.7% of frames; while, Cilento N represents the less stressed site, with value of 4.6%. In particular, the medium values of fishing gear and marine litter for each area resulted: Ischia 14.2 ± 7.9%, Napoli 24.3 ± 26.3%, Campanella 30.3 ± 23%, Capri 32.7 ± 23.4%, Cilento N 4.5 ± 4.8%, and Cilento S 4.7 ± 4.8%.   Anthropogenic pressure is mostly represented by longlines (67.6%), followed by nets (20.4%); while metal objects were the most frequent among marine litter (0.1%) ( Figure 6). Capri is the most stressed area with the presence of gear and litter in the 32.7% of frames; while, Cilento N represents the less stressed site, with value of 4.6%. In particular, the medium values of fishing gear and marine litter for each area resulted: Ischia 14.2 ± 7.9%, Napoli 24.3 ± 26.3%, Campanella 30.3 ± 23%, Capri 32.7 ± 23.4%, Cilento N 4.5 ± 4.8%, and Cilento S 4.7 ± 4.8%. The nMDS ordination plot (Figure 7) performed on the different areas centroids show three evident clusters: Campanella-Capri, Cilento N-Cilento S, and in the middle Napoli- The nMDS ordination plot (Figure 7) performed on the different areas centroids show three evident clusters: Campanella-Capri, Cilento N-Cilento S, and in the middle Napoli-Ischia. Moreover, the distLM analysis highlights that longlines are the only gear significantly affecting coralligenous assemblages structures (p = 0.0004, r 2 = 0.71), explaining 94.5% of the total variation. Ischia. Moreover, the distLM analysis highlights that longlines are the only gear significantly affecting coralligenous assemblages structures (p = 0.0004, r 2 = 0.71), explaining 94.5% of the total variation. Figure 7. 3D-nMDS plot of anthropogenic pressure, from Euclidian distance matrixes. Points represent centroid of areas levels clusters in 3D multidimensional space, and the distance between any two points represents the difference between those two areas. The high quality of the ordination is indicated by a low-stress value (3D stress = 0.03).
In Figure 10, the frequencies of indirect and direct impacts detected in each area are shown. Campanella and Capri resulted in the most impacted areas (29.2% and 17% of frames), followed by Cilento S (7.9%), while Ischia, Napoli and Cilento N resulted in the less impacted areas (4.9%, 4.1% and 3.8%). In particular, CNE impact is the most frequent in Ischia, Napoli, Cilento N and Cilento S areas (3.5 ± 3.3%, 2.8 ± 4.7%, 3.5 ± 8.8% and 6.6 ± 5.2%, respectively). Instead, GCE is the most frequent impact in Campanella and Capri areas (17.8 ± 15.7% and 9.6 ± 6.7%, respectively). Linear regressions between longlines and direct impact, indirect impact, and heterogeneity are showed in Figure 11 and Table 3. In particular, a significant positive relation (p = 0.0086, r 2 = 0.77) is detected between longlines and direct impact; while, no significant relation is detected between longlines and indirect impact. Finally, a significant negative relation (p = 0.035, r 2 = 0.77) between longlines and heterogeneity. Linear regressions between longlines and direct impact, indirect impact, and heterogeneity are showed in Figure 11 and Table 3. In particular, a significant positive relation (p = 0.0086, r 2 = 0.77) is detected between longlines and direct impact; while, no significant relation is detected between longlines and indirect impact. Finally, a significant negative relation (p = 0.035, r 2 = 0.77) between longlines and heterogeneity.  Table 3. Mean (± SD) of linear regression independent (longlines) and dependent variables (direct impact, indirect impact, and heterogeneity).

Discussion
Campanella and Capri represent the areas with the highest number of organisms. In all areas, however, as evident in Figure 2, density and frequency of CSS show a regular trend, typical of a random spatial species distribution, possibly due to the absence of limiting factors controlling species eco-physiology (e.g., temperature, water, substrate type) [66].
Although the species richness in terms of CSS number among areas is consistent, changes in the species composition were observed; in particular, the phylum Cnidaria dominates in Napoli, Campanella and Capri, while the phylum Bryozoa dominates in Cilento N and Cilento S. This result may also depend on the different abiotic factors among areas, such as the substrate type, the hydrodynamic regime, the depth range, or the water trophism; for example, large gorgonians show a higher preference for trophic and hydrodinamic waters [67]. Coralligenous assemblage's structure and composition are strongly related to edaphic factors, which may act on their heterogeneity [51,68]. Some studies revealed that extreme environmental conditions may promote the growth of coralligenous assemblages and large suspension feeders [69]; indeed, the most abundant gorgonian species found in the study area grow in low-light conditions, induced probably also by the high turbidity and the prevalence of strong currents [57,70,71], that in the study areas may be near-bottom or upwelling, generating water turnover and promoting nutrient dispersion [72,73].
CSS densities in the Campanella and Capri areas are significantly different from those of the Cilento N and Cilento S areas, and appear to be less heterogeneous in terms of β-diversity, despite the higher densities values. This condition may be the result of high dominance values of few species, possibly due to anthropogenic impact [74]. Indeed, Campanella and Capri are mostly dominated by the large gorgonians E. cavolinii and P. clavata that are highly impacted, since their morphological structure may increase the likelihood to be entangled and damaged by fishing gear. On the contrary, Cilento areas are characterized by the presence of small bryozoans, such as M. truncata and P. fascialis that may decrease the likelihood to be entangled. Several studies have demonstrated that low diversity is linked to a high ecosystem impact. For example, some authors, i.e., [75], showed that impacted assemblages on artificial bottoms present fewer taxa and lower values of diversity, suggesting that those support a less diverse and more homogeneous assemblage than natural substrates. Moreover, communities characterized by high β-diversity should be more resistant to disturbance than others, acting as refuges for neighboring patches [76,77]. High diversity levels make communities able to persist and absorb fluctuations, increasing the resilience potential and reducing the fragility of marine communities and ecosystems [47].
Campanella and Capri are the most stressed areas with the highest numbers of fishing gear and marine litter, followed by Napoli and Ischia, while Cilento N and Cilento S are the least stressed ones with very low values. Generally, longlines dominate in all the sampled areas, followed by nets, with very high numbers in Campanella and Capri. Nets are more abundant in the Napoli area, where the coralligenous bioconstructions are present in the form of banks surrounded by muddy substrate. Longlines are widespread because of their dual-source from both professional and recreational fishing, while nets should be considered mainly professional gear [78]. Moreover, longlines can be used on different bottom types, while the nets are usually set on soft bottoms, even though they may get stuck on hard bottom, with relevant damages for the benthic communities [32,79,80].
Marine litter is the most abundant along the coastal areas of Capri, Napoli and Campanella; in particular, the Napoli area is close to a large urban center, while Capri and Campanella areas are located along popular touristic routes. Despite the highest impact of fishing gear, linked to the developed professional and recreational fishing activities, the use of the coasts for recreational tourism purposes may be responsible for the accumulation of anthropogenic debris, as well as the rivers input [44]. In addition, recreational beach activities can possibly contribute to marine litter accumulation along the coastal areas [81].
CSS colonies with necrosis or epibiosis are present in all areas, with the CNE impact results the highest in Ischia, Napoli, Cilento N and Cilento S. However, they appear lower than 5% in Ischia, Napoli and Cilento S, and between 5 and 10% in Campanella, Capri and Cilento N. These values may be considered respectively low and moderate since the impact is usually considered elevated for values higher than 10% [52,82]. Fishing gear may either eradicate the large colonies or get entangled in the ramifications, scraping their coenenchyma with consequent necrosis tissues and favoring the development of epibionts, which may lead to a burdening of the colonies and to a greater mechanical stress, increasing their resistance to water movement [28,80]. Necrosis, and more extensively, colonies mortality, may also be caused by different stressing factors such as water temperature increase, possibly linked to climate change [83]. The heatwave of 2003 in Europe caused an anomalous warming of seawater, which played a key role in the observed mass mortality event of fan corals [84,85], as well as climatic anomalies, which can generally cause biodiversity loss and ecological shifts [86,87]. Necrosis may also be caused by a high sedimentation rate, as a consequence of resuspension caused by trawl-fishing [88], altering metabolic functions, enhancing respiratory and interfering with the prey-capturing apparatus, and also compromising the settlement [89]. For this reason, the identification of the effective cause of this indirect impact can be difficult.
Direct impact, mostly represented by GCE, prevailed in Campanella and Capri areas, where it is strictly connected to the high presence of fishing gear. Erect or branched organisms among Cnidaria, Porifera and Bryozoa are the most endangered by fishing gear since they are often broken and upturned [28,[90][91][92]. Direct impact was mostly caused by the presence of abundant longlines, while no relation was found between indirect impact and longlines. Moreover, an indirect relation between heterogeneity and longlines was detected, demonstrating that fishing activity may cause severe species diversity loss [93][94][95].
A better knowledge of the anthropogenic activities impact on benthic communities of mesophotic reefs is increasingly needed in order to improve the management tools to protect one of the most fragile and valuable Mediterranean habitats. This should be reached through sustainable fishing techniques, adequate protection measures and long-term monitoring programs. Environmental protection, also through Marine Protected Areas, should respond not only to an ecological purpose but also to socioeconomic dimensions of sustainability [96].