The precious coral Corallium rubrum (Linnaeus, 1758) is a charismatic Mediterranean species that suffered extensive exploitation for centuries due to its commercial value in the jewellery industry (Tsounis et al., 2006; Bussoletti et al., 2010; Cattaneo-Vietti and Bavestrello, 2010; Tsounis et al., 2010; Cattaneo-Vietti et al., 2017; Cannas et al., 2019). Harvesting has been traditionally exercised with non-selective and highly destructive trawling gear, while, after their ban in 1994, it has been carried out selectively by professional SCUBA divers; this long-term pressure resulted in a serious over-fishing that put this resource into jeopardy (Cattaneo-Vietti et al., 2017). Currently, this species is included in a number of international conventions supporting its controlled harvesting (SPA/BD Protocol Annex III, Bern Convention Annex III, Habitat Directive Annex V and GFCM Recommendations), and in 2016 it was assessed as Endangered in the IUCN Anthozoans Red List for the Mediterranean Sea (IUCN, 2016), recognizing its important ecological role as habitat-former (Garrabou and Harmelin, 2002; Rossi et al., 2017) and its decline over time. Taking into consideration the IUCN assessment, the inclusion of red coral in Appendix II of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) should be re-evaluated in the near future (Bruckner, 2009; Bussoletti et al., 2010; Bruckner, 2014).

In the past decades, in order to evaluate the distribution and status of the different populations living in the Mediterranean Sea, numerous research studies have been conducted, mainly on littoral banks, often showing an evident state of suffering (FAO, 1988; Cicogna and Cattaneo-Vietti, 1993; Cicogna et al., 1999; Garrabou and Harmelin, 2002; Santangelo et al., 2007; Santangelo et al., 2009; Santangelo and Bramanti, 2010; Tsounis et al., 2010; Bramanti et al., 2014; Montero-Serra et al., 2015). Dramatic size and structural shifts in shallow-water populations have been recognized (Bruckner, 2009), with “forest-like” populations, dominated by large, branched colonies of 30-50 cm height (10-30 mm in diameter) turning into “grass plain-like” populations, dominated by small colonies of 3-5 cm height (5-7 mm in diameter) (Garrabou and Harmelin, 2002; Tsounis et al., 2006; Santangelo et al., 2007; Tsounis et al., 2007; Bramanti et al., 2014). The size shift due to harvesting is not only related to the obvious reason that divers tend to select colonies of larger size, but also to the freeing up of space due to heavy harvesting enhancing larval settling events (Cerrano et al., 1997). The “grass plain-like” condition may persist for a long span of time, because of intra-specific competition (Tsounis et al., 2006), and only in a few Marine Protected Areas has a reversion been recognized after a long-term protection (Bavestrello et al., 2015; Richaume et al., 2021). Moreover, besides harvesting, other factors are known to threaten red coral populations, both direct (e.g., fishing activities and mass mortalities putatively linked to episodic and highly energetic events, like heat waves and volcanic eruptions) and indirect (e.g., climate change-driven events) (e.g., Cerrano et al., 2013; Bavestrello et al., 2014; Lodolo et al., 2017; Cau et al., 2018; Galgani et al., 2018; Enrichetti et al., 2019a).

From a conservation point of view, morphometric data are essential to estimate the health status of the populations, which, in turn, are fundamental to identify the most vulnerable sites and the correct management actions. Although several pieces of evidence are available for shallow-water populations, very little data have been recorded about the morphometric structure of deep populations (Rossi et al., 2008; Priori et al., 2013; Angiolillo et al., 2016; Cau et al., 2016; Carugati et al., 2022). These studies, mainly conducted on a local or regional scale, showed that deep-dwelling populations are generally less harvested than the shallow-water ones, even if high percentages of the colonies are still above the commercial size (≥ 7 mm of basal diameter). However, authors do not agree if there is a size-density relationship in red coral populations with increasing depth, presenting contrasting results.

A recent large-scale investigation along the Italian coast (Toma et al., 2022), taking advantage of an extensive ROV archive, demonstrated that, at basin scale, red coral still shows a wide occurrence, particularly in a large region of the central Tyrrhenian Sea including Tuscan Archipelago, Pontine Archipelago, Gulf of Naples, and the Sardinian coasts. This study also demonstrated that the species occurrence peaks in the mesophotic zone, between 75 m and 125 m, showing a marked preference for biogenic habitats dominated by Crustose Coralline Algae (CCA) and for vertical substrata. However, the lack of morphometric data limited the considerations on the structure and health status of the populations.

The objectives of this study, therefore, were to i) conduct a large-scale morphometric survey on mesophotic and deep populations in the Ligurian Sea, Tyrrhenian Sea, and Sicily Channel, and highlight latitudinal and bathymetrical patterns in colony average size and density, ii) identify the size-frequency patterns in the most representative forests to depict the putative influence of harvesting pressure, and, ultimately, iii) identify other sources of direct mortality which may affect the health status of those populations.

A large ROV archive was analysed to extrapolate quantitative data with the aim of describing the red coral populations present along the Italian coast. The dataset consisted of 624 ROV dives (each corresponding to a distinct geographical site) carried out between 2006 and 2021 in the Ligurian Sea, Tyrrhenian Sea, and Sicily Channel, between 40 m and 1825 m. Sites were grouped in localities based on their geographic proximity (maximum distance about 5 km), then localities were grouped in areas based on their topographic and oceanographic affinity (e.g., promontory, gulf, shoal), and such areas fell into different macro-areas (4 coastal ones, LIG Ligurian Sea, NCT North-central Tyrrhenian Sea, ST South Tyrrhenian Sea, and SC Sicily Channel, and 1 offshore, SEAM Seamounts) ( Table 1 ).

The number of sites in which Corallium rubrum was present was counted and reported as a percentage of occurrence in the whole dataset. For the quantitative analyses regarding coral areas, only sites (hence localities and areas) with more than 30 colonies were considered.

The depth ranges at which red coral was recorded were noted and ten random frames were obtained for each site targeting the densest coral patches, for a total of 1700 analysed frames. For each picture, density was calculated considering the hardground surface, evaluated using the ROV laser beams, and the exact number of colonies, counted using ImageJ (15730 total counted colonies). The resulting average density of the sites was then defined according to three categories: (1) scattered colonies (< 5 colonies m^-2), (2) densely aggregated colonies (5-20 colonies m^-2) or (3) very densely aggregated colonies (> 20 colonies m^-2). Density values attributed to each category were based on the entire available dataset.

The relationship between density and depth was presented with two different approaches. The first considered the average densities of the sites in four distinct depth ranges (40-80 m, 80-150 m, 150-200 m, and > 200 m). The second analysis considered the percentage distribution of the density categories in the same depth ranges. These latter corresponded to the deepest part of the continental shelf, the shelf break, the upper bathyal range of the continental slope and seamounts, and the bathyal, respectively. The putative effect of latitude on the population average density per macro-area was also investigated by means of a linear regression with r = Pearson’s correlation coefficient.

The height of the colonies was measured in each frame with ImageJ using the laser beams as a reference, to evaluate the morphometry of the populations in the selected coral areas. The relationship between height and depth was investigated by considering the mean height of the populations in the four depth ranges. The putative effect of latitude on the average height was also investigated in the macro-areas by means of a linear regression with r = Pearson’s correlation coefficient. Finally, size-frequency graphs were created for the considered geographic areas, with size classes of 2 cm.

Entanglement to lost fishing gear was evaluated by noting the number of impacted colonies in each frame. Data were presented as the percentage of entangled colonies on the total, as well as the percentage of entangled colonies in each coral area. To investigate the correlation between entanglement and depth, the percentage of entanglement for each area and depth ranges were considered. The putative effect of latitude on the percentage of entanglement of the populations was also investigated for the macro-areas by means of a linear regression with r = Pearson’s correlation coefficient.

Finally, in the sites where dead in-place colonies deprived of living coenenchyme were observed, ten random frames were collected and data of density, height, entanglement, substrate type (rocks, R, and rocks encrusted by CCA, C), and slope of the hardground where C. rubrum was attached were noted to characterize the mortality patches. In the sites where both living and dead populations were present, density values and the average height of the colonies were compared to evaluate the entity and timing of the mass mortality events.

Significant differences in terms of average densities and average heights in the selected coral areas were tested through a Kruskal-Wallis test (data not normally distributed, square root transformed, n=10 - 220, and n=35 - 2913, respectively, for 17 coral areas). Significant differences between average densities, average heights, and average entanglement in the four considered depth ranges were tested through a Kruskal-Wallis test (data not normally distributed, square root transformed, n=3 - 170, for the four depth ranges). All statistical analyses were performed using PAST Software ver. 4.03 (Hammer et al., 2001).

This work considered the widest demographic dataset ever produced on Corallium rubrum, with a specific focus on the mesophotic and upper bathyal depths of the Italian coast. This species confirms to be widespread in the basin, with seventeen coral areas identified based on their geographic proximity and topographic and oceanographic affinity. Among them, some showed high density and considerable size of the colonies (such as those of the Portofino Promontory, Tuscan Archipelago, South and South-western Sardinia, Campanian Archipelago and South Sicily), representing important larval stock for the species, and these may be the target of specific management actions as required by the most recent species and habitat regional assessments (Otero et al., 2017; Montefalcone et al., 2021).

In general, the investigated populations consistently showed a size-frequency distribution skewed towards small-size colonies, with low variability among the areas and among depth ranges. Although it is recognized the possibility of a limited bias in the measures obtained through bidimensional ROV footage with respect to the ones derived by three-dimensional techniques (Lagudi et al., 2022), the overall patterns appear clear. The size of colonial organisms such as C. rubrum is known to be potentially locally constrained by numerous environmental parameters (e.g., food and substrate availability, temperature, currents, depth, slope) (e.g., Cau et al., 2016; Galli et al., 2016) as well as by anthropic pressures, mainly harvesting and fishing (e.g., Tsounis et al., 2006; Rossi et al., 2008; Bruckner, 2009; Cattaneo-Vietti et al., 2016). Over the spatial scale here considered, these factors may be synergically responsible for the recorded variations, especially when considering the occurrence, in certain coral areas (western Ligurian Sea, Pontine Archipelago, Tyrrhenian Calabria, Sicily Channel), of multiple cohorts of colonies. However, overall, it seems that, among all factors, depth does not influence the size of the colonies. This is in accordance with Carugati et al. (2022), reporting no significant variation of colony height with depth in the North-western Sardinian populations (50-290 m), but contrasts with the findings of other authors sustaining the major size (as well as diameter and branching) of deep-dwelling colonies, a trend mainly attributed to a lower harvesting pressure (e.g., Tsounis et al., 2007; Rossi et al., 2008; Priori et al., 2013; Angiolillo et al., 2016; Cau et al., 2016). Moreover, the same authors provided evidence of a size/density relationship within C. rubrum populations, with larger, but more scattered colonies in deeper waters, that did not result from the present dataset.

Together with indeed ecological considerations, the homogeneous reduced size of the red coral populations in the entire study area could be related also to a stressful situation due to the long-term harvesting pressure carried out in all the considered basins (Cattaneo-Vietti et al., 2016). Already Bruckner (2009, 2014), based on data derived from the landed harvested material, supported the use of size reduction in time as the main indicator of the harvesting impact on the demography of the populations. Similar conclusions were also derived from in-situ local studies, carried out both on shallow and deep populations. Tsounis et al. (2006); Tsounis et al. (2007) and Rossi et al. (2008) clearly demonstrated that the harvesting pressure is the main constraint driving population structure at Cap de Creus (Spain, NW Mediterranean Sea), reporting a colony size in shallow overexploited populations noticeably smaller than that of deeper, less harvested populations (average colony size: 3.0 ± 1.7 cm, and 7.8 ± 3.3 cm, respectively, in shallow and deep populations). Significant differences in terms of basal diameter and maximum height emerged from the comparison between harvested and non-harvested populations around Marseilles (France) (Garrabou and Harmelin, 2002). Finally, larger colony sizes have been observed also in populations located inside protected areas, where harvesting is forbidden, as in the Medes Islands Marine Reserve (Spain) and on Ischia Island (Regno di Nettuno MPA, Campania, Italy), suggesting that the harvesting pressure has a deleterious effect on the population structure of the precious coral (Linares et al., 2010; Angiolillo et al., 2016). Our data, coming from a wide sampling effort, provide additional support to this hypothesis.

Differently from height, density is strongly driven by depth, with a linear decrease from shallow to deep sites, confirming the preliminary qualitative results obtained by Toma et al. (2022) and in accordance with known literature (e.g., Rossi et al., 2008; Carugati et al., 2022). In addition, on the continental shelf and shelf break, higher variability in terms of density was observed, with dense aggregations and populations made of sparse colonies equally present, while at bathyal depth, populations of scatted colonies were dominant. A similar pattern was already reported in Cap de Creus (North-western Mediterranean Sea) at 50-230 m (Rossi et al., 2008). The present work enhances the idea that the self-thinning process (i.e., the intra-specific competition regulating population density and structure, based on space limitations, namely hardgrounds availability) (Cau et al., 2016), is a valid explanation for the density-depth pattern observed on a large scale for C. rubrum, but this trend is not necessarily associated to an increase of the colonies’ size with depth. It has to be considered that this latter trend of size with depth emerged from studies targeting single populations located in restricted geographic and depth ranges, and often deeply harvested since historical times in shallow waters. The present study, instead, considers a much larger area (about 680,000 m²), and a wider depth range (from the deep circalittoral to the deep bathyal), including coastal and offshore sites subjected to a larger variety of environmental constraints and human pressures. This allowed obtaining a larger picture of the population structure of red coral, damping local influences. Interesting large-scale patterns, both in terms of density and height, emerged from the latitudinal point of view, probably related to unknown gradients of environmental (e.g., currents, temperature, silting, nutrients) and biological (e.g., competition with other co-occurring components of the assemblages) characteristics. In fact, populations showing higher densities and smaller colony heights characterize the mesophotic hardgrounds of the northern areas, with respect to the southern ones.

In terms of coastal-offshore gradient, seamounts, despite hosting only a few sparse red coral colonies generally of large size, offered some insights into the demographic trends of this species. The distance of these structures from the coast seems to act as a barrier for the larval dispersion of the species, which is generally considered on a short-medium range (Costantini et al., 2007; Martínez-Quintana et al., 2015; Costantini and Abbiati, 2016; Guizien et al., 2020). High densities of C. rubrum observed on other Mediterranean seamounts may be related to their closer proximity to the coast (de la Torriente et al., 2018), as well as to more favourable trophic conditions than those found on the offshore reliefs investigated here (30-70 NM from the nearest coastal populations). However, once settled, offshore red coral populations may potentially be subjected to lower levels of anthropic impact since these structures have never been interested by coral harvesting (Cattaneo-Vietti et al., 2016) and are known to have a modest fishing pressure (Bo et al., 2014), hence supporting larger sizes.

The remoteness of a site, influencing its anthropic accessibility, both in terms of depth and distance, has been considered a factor driving the demography of the red coral populations (Rossi et al., 2008). In this study, demersal artisanal and recreational fishing activities emerged as serious and widespread threats to the habitats including red coral at almost all coastal sites and depths. Colonies entangled in fishing gear were visible in almost 90% of the coral areas, with those in the Ligurian Sea and Sicily Channel being among the most impacted, and those in the Tuscan Archipelago and eastern Sardinian coast among the least. This pattern has been highlighted in numerous studies conducted on the deep megabenthic assemblages of these regions (Bo et al., 2014; Angiolillo et al., 2015; Cattaneo-Vietti et al., 2017; Cau et al., 2017). Highly impacted benthic communities have been documented along the Ligurian coast, especially in the western sector (Enrichetti et al., 2019a; Enrichetti et al., 2019b). On the contrary, the Portofino Promontory resulted among the less impacted Ligurian coral areas. This could be due to the proximity of the studied mesophotic populations to the coastal Marine Protected Area. Here, in fact, a reversion in the size-structure shift was observed by Bavestrello et al. (2015), reporting flourishing dense populations of large colonies at shallow depths after almost two decades of protection. The positive effects of restrictions on red coral populations have been reported also in North-West Sardinia, where large colonies (modal size classes 10-15 cm and 15-20 cm) were registered by Onboard Scientific Observes in catches from areas subjected to stricter harvesting measures compared to other countries within the GFCM competence area (Follesa et al., 2013; Carugati et al., 2020). This specific area within the Sardinian Sea was already reported as a hotspot of occurrence of C. rubrum (Toma et al., 2022). These works support the importance of MPAs and other protection measures in the conservation and management of red coral. This is particularly true if we consider the high frequency of the indirect damages of fishing activities occurring in the deepest part of the continental shelf and along the shelf break, where the densest and more diversly aggregated populations were observed.

The present study also highlighted how, besides fishing and harvesting pressures, mass mortality events, registered down to 200 m, may represent significant stresses for mesophotic and upper bathyal red coral populations. Our data enlarge the current knowledge about red coral mass mortality events identifying a Tyrrhenian and Sicilian belt of dead red coral zones. The causes of these catastrophic events remain, however, uncertain. Similar phenomena in coastal ecosystems (0-40 m), including also red coral, have been linked to modifications in environmental conditions triggered by global changes (e.g., thermal anomalies) (Harvell et al., 1999; Cerrano et al., 2000; Pérez et al., 2000; Garrabou et al., 2001; Harley et al., 2006; Garrabou et al., 2009; Cerrano et al., 2013). However, in the deep regions where red coral mortalities have been reported, far from the influence of the seasonal thermocline, other putative causes should be addressed. For example, the red coral mass mortality event reported along the coast of Provence in 1987 (Rivoire, 1991; Arnoux et al., 1992), between 80 and 160 m, was attributed to polluted waters driven by the dominant currents in that area. The deep circulation of warm waters in proximity to volcanic areas (Bavestrello et al., 2014) combined with the seafloor movements in seismic areas (Lodolo et al., 2017; Bavestrello et al., 2021) may explain the extensive mortality events observed in the Pontine Archipelago, Campanian Archipelago and the offshore banks of the Sicily Channel. For the Sardinian sites, however, there is no clear explanation of the putative factors triggering the coral death, neither for the population along the North-eastern sector (Carugati et al., 2022) nor for those in the southern one. Most of the mortalities were recorded in submarine canyons, which may support the fact that the particular environmental conditions (e.g., turbidity currents) occurring in these systems very close to the shorelines, may have caused the events (Moccia et al., 2021). However, the co-occurrence, in the same sites, of different structuring filter-feeding anthozoans (e.g., Antipathella subpinnata (Ellis and Solander, 1786), Eunicella cavolini (Koch, 1887)) in apparently good health status would exclude a wide clogging event triggered by a massive deposition of silt (Moccia et al., 2022). Certainly, mass mortality events, both in shallow and deep waters, should be considered as additional factors significantly shaping the demography of red coral populations in the short term.

This work provides valuable information on the distribution and population structure of a precious overexploited species from a large-scale perspective. Deep coral areas, with particularly high density and size, were identified, as well as the areas most affected by fishing impact and mass mortality events. All this information is fundamental to have a comprehensive picture of the demography of an endangered species and to depict the best conservation actions.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Conceptualization, MT, GB, MB. methodology, validation, formal analysis, MT, GB, MB. investigation, resources, data curation, writing-original draft preparation, MT, MB, GB, SC, AC, MA. writing-review and editing, MT, MB, GB, SC, AC, MA. visualization, supervision, project administration, GB, MB, SG, FA, SC. All authors contributed to the article and approved the submitted version.