Cold-water corals (CWCs) are important habitat-forming species including azooxanthellate stony corals (Scleractinia), gorgonians (Alcyonacea), black corals (Antipatharia) and hydrocorals (Stylasteridae) (Roberts et al., 2006). CWCs and their habitats are receiving great interest from the scientific community due to their high ecological value (Roberts et al., 2009). Cold-water reefs formed by stony corals are complex three-dimensional structures and are biodiversity hotspots (Henry and Roberts, 2007; Hovland, 2008; Bongiorni et al., 2010; Mastrototaro et al., 2010; Rueda et al., 2019) acting as nurseries, refugia or feeding grounds for several species of fishes and invertebrates (Baillon et al., 2012; D’Onghia, 2019).

CWCs are threatened by several anthropogenic activities (D’Onghia et al., 2016; Ragnarsson et al., 2016; Montseny et al., 2021). Among them, deep-sea fisheries directly affect CWC reefs (D’Onghia, 2019), up to an estimated 95-98% of the total coral cover (Gianni, 2004). Oil and gas exploitation is extending at deeper depths, increasing the risks of oil-spill accidents or dispersal near CWC reefs (Cordes et al., 2016). Also, deep-sea mining might affect CWC reefs that are often located near sites of exploitation (Roberts and Cairns, 2014; Ragnarsson et al., 2016). In addition, global climate change represents a serious concern for CWCs as the reduction of pH determines a decrease of the depth of aragonite saturation horizon (Ragnarsson et al., 2016; Sweetman et al., 2017) and increasing rates of skeletal dissolution (Gómez et al., 2018). Acidification and higher water temperatures are proved to affect cold-water corals also by increasing their vulnerability to other environmental stressors like chemical pollution (Weinnig et al., 2020). Projections indicate that 70% of the actual locations of CWC ecosystems could become under-saturated by 2099 (Guinotte et al., 2006).

Ecosystem restoration is a key action of several frameworks and conventions both at European and global level. Restoration of damaged CWCs ecosystems will be increasingly required in the future, even because sequestration of carbon dioxide from seawater by CWCs (Turley et al., 2007) constitutes an essential regulating service and, because of this, the deep sea has been proposed as storage environment for the surplus CO₂ produced by humans with the use of fossil fuels (Davies et al., 2007).

Considering that the research field on ecological restoration of deep-sea ecosystems is still in its infancy (Da Ros et al., 2019), only few studies focus on CWCs active restoration actions, as reported in the literature (Montseny et al., 2021). The transplantation of nubbins taken from healthy colonies or reared in aquaria into degraded coral grounds is the preferred restoration action (Da Ros et al., 2019 and references therein). Pilot experiments on transplantation of nubbins of Desmophyllum pertusum were carried out in the Gulf of Mexico, in the North Sea and off Central California, with encouraging results of high survival rates for D. pertusum (Brooke and Young, 2009; Dahl, 2013; Boch et al., 2019). Recently, in the Mediterranean Sea and in the Atlantic (Azores Archipelago), several CWCs, including fan corals, were transplanted to evaluate their survival rate (Linares et al., 2020).

The success of maintenance of CWCs in aquaria can be affected by variations of physical parameters (mainly flow velocities and temperature) that can also influence the feeding behaviour of CWCs like widely demonstrated in literature (Purser et al., 2010; Gori et al., 2015; Orejas et al., 2016).

Artemia salina is often used for keeping CWCs in aquaria and for conducting feeding experiments because it has the same size of target copepods (Orejas et al., 2019 and references therein). Usually, rotifers in aquaria are provided as feeding supplementary to corals characterised by very small polyp size, like black corals (e.g., Bathypathes sp. and Leiopathes glaberrima) or gorgonians (es Dentomuricea meteor) (Orejas et al., 2019). In captivity, nauplii of A. salina (Linnaeus, 1758) and algae are routinely used to also feed D. pertusum (Orejas et al., 2019), while Mysis sp. (Latreille, 1802) and adults of A. salina are used to feed D. cornigera (Gori et al., 2015; Orejas et al., 2019).

In the Mediterranean Sea, the most common reef-building CWCs are the ‘white corals’ Desmophyllum pertusum (Linnaues 1758), formerly known as Lophelia pertusa (Addamo et al., 2016), and Madrepora oculata (Linnaeus, 1758). These two taxa often co-occur and engineer the CWC-grounds with maximum areal occupancy in the coral provinces identified in the central and western Mediterranean Sea, between ca. 200-800 m (Fanelli et al., 2017; Taviani et al., 2017; Chimienti et al., 2019; Taviani et al., 2019) at a temperature comprised between 11 and 13.9°C (Naumann et al., 2014). In-situ observations carried out through ROVs along Bari canyon and on the Apulian Plateau showed that M. oculata is the dominant species of CWCs in the Central Mediterranean Sea (Freiwald et al., 2009), probably due to the environmental conditions that in this basin are closed to the ecological limits of this species (Davies et al., 2008). The “yellow coral” Dendrophyllia cornigera (Lamarck, 1816) forms colonies from the mesophotic zone down to bathyal depths, where it often mingles with white corals (Castellan et al., 2019). In the Mediterranean Sea, this eurybathic species lives in a temperature range between 7 and 16°C (Castellan et al., 2019).

Here we investigated the feeding preferences of D. pertusum, M. oculata and D. cornigera to establish the best conditions to keep them in captivity and for maintaining them on-board during the oceanographic campaigns. Four food sources were chosen to feed the corals and establish their food preferences: 1) nauplii of A. salina, 2) the green algae Tetraselmis subcordiformis, (Wille) Butcher 1959, 3) two species of rotifers (Brachionus plicatilis, Müller, 1786, and B. rotundiformis, Tschugunoff, 1921) and 4) the crustacean mysid M. relicta (Lovén, 1862).

The main goal of this study is to determine the best conditions for successfully and effectively keeping CWCs alive in aquaria, in face of increasing request of such kind of knowledge for restoration of marine degraded habitats as promoted by the UN Decade on Ecosystem Restoration (UNGA, 2019) at global level and by the EU 2030 Biodiversity Strategy (EC, 2020) at European level. Specific aims of this study are: 1) to determine the isotopic signatures of the exposed specimens to discriminate between ingested vs. assimilated food items and their resource partitioning (i.e., which of the captured preys were preferentially assimilated by the different species) and 2) to increase our understanding on the feeding ecology of CWCs occurring in the Mediterranean Sea.

Three species of cold-water corals were sampled using a ROV (Remotely Operated Vehicle), during two oceanographic campaigns conducted in 2016 in the Mediterranean Sea on-board of RV Minerva Uno. In order to reflect the cold-water coral community structure of the Central Mediterranean Sea, one live colony of D. pertusum (made up of 280 polyps) and 4 colonies of M. oculata (with a total of 1501 polyps) were collected during the SIRIAD16 oceanographic campaign at depth of 244 and 400 m, respectively, in the south-western Adriatic Sea ( Figure 1 ). One colony of D. cornigera (made up of 9 polyps) was collected at 139 m depth, during the RISD_16 campaign, in the northern Ionian Sea ( Figure 1 ). Corals were kept alive inside 20 L PVC darkened aquaria filled with bottom seawater (previously filtered with a 20 µm-mesh) collected simultaneously to corals, without feeding them due to the logistical limitations on board, according to literature (Orejas et al., 2019).

Once in the laboratory, the colonies were kept in different aquaria (one tank for each species, a 30 L tank for D. cornigera, two 50 L tanks for D. pertusum and M. oculata, with the 4 colonies of M. oculata kept together in the same aquaria to better simulate the community structure) filled with bottom seawater (previously filtered with a 20 µm-mesh). Before choosing how to divide the colonies and the volumes of the tanks to use, we checked that the estimated total dry weights of the collected species were similar, according to dry weights values reported in Maier et al., 2012 for D. pertusum and M. oculata and in Gori et al., 2014 for D. cornigera.

Corals were maintained in the dark and at a temperature of 13 ± 0.5°C. Although corals were collected in different sites, the in-situ temperature was similar, ( Table S1 ), thus we set the temperature at ca. 13°C for all the three species as this is the mean temperature of the deep waters in the Western and Central Mediterranean Sea (Danovaro et al., 2010), with the maximum recorded temperature for the occurrence of D. pertusum at 13.8°C in the Ionian Sea (Freiwald et al., 2004). Temperature was maintained constant through a common water bath and a refrigerator (TECO SeaChill Chiller TR5). Seawater was sampled in the central Adriatic Sea and filtered with a 20 µm-mesh prior to gradually mix it with the bottom seawater of the aquaria (final salinity 37 ± 0.2 PSU). Subsequently, ~60% of the seawater in the tanks was exchanged every 10 days. Every day, the bottom of the tank was cleaned with an aquarium siphon to ensure build-up of detritus to be minimal.

Due to the difficulties in replicating the natural complexity of the environments from which the corals were collected in laboratory settlings, and differently from previous studies that focused on the effects of current velocities on their feeding capacity, here we maintained a constant water recirculation using submersible pumps with a flow rate of ~2000 L h⁻¹ (Orejas et al., 2019). The flow velocity was maintained in a range of 5-10 cm s^-1 to better simulate the natural conditions of collection areas (Turchetto et al., 2007; Langone et al., 2016) with peaks of ~20 cm s^-1 in the tank of D. cornigera (Davies et al., 2009). Pumps water-outlets were coated with a 20 µm-NITEX nylon mesh in order to avoid harming of prey items.

An air stone oxygenated the seawater and was placed at the top of the aquaria preventing any influence on polyps’ activity. The oxygen concentration was maintained at in-situ conditions (5.2 ± 0.2 mL L^-1) (Davies et al., 2008). We started the experiment when the colonies completely extended their polyps without showing stress signals (e.g., mucus production) (Murray et al., 2019). This acclimation period lasted 4 weeks for D. pertusum and M. oculata (Chapron et al., 2018) and 2 weeks for D. cornigera. Since the filtered seawater still contained sufficient organic particles, no extra food had to be provided during the acclimation period (Van Oevelen et al., 2018). Food was provided when at least the 90% of the polyps of the colonies were open.

This study investigated the food preferences of the CWCs, maintained in aquaria at conditions like those encountered during their life in the deep Mediterranean Sea. D. cornigera was the most reactive species, especially after the food supply that stimulates its polyps’ reaction (Orejas et al., 2019). D. pertusum was the most sensitive species to the exposure, even to brief and weak-light expositions, such as those occurring during sampling operations. Being D. pertusum the species most closed to its ecological upper-temperature limit, probably it reacted stronger than the other species to this environmental stressor. Further studies are needed to further explore the role of light. Moreover, this species reacted slowly at the presence of food, confirming previous observations (Mortensen, 2001). Behaviour differences among species could possibly also reflect their reaction to variations in pressure from the site of collection to aquaria conditions, in the order of ca. 20 bars for D. pertusum and ca. 40 for M. oculata (Orejas et al., 2019). The eurybathic D. cornigera was instead collected at ca. 14 bars, thus facing lower pressure variations, in principle less stressing for the animal. M. oculata shows, in general, much higher tolerance to environmental fluctuations (Wienberg et al., 2009), while D. pertusum is a species more commonly found in cold waters with temperatures between 4-12°C (Freiwald et al., 2004). In the central Mediterranean Sea, temperatures measured in live coral habitats during oceanographic campaigns range between 13.4 and 13.9°C (Freiwald et al., 2009). These temperatures are close to the ecological limit of D. pertusum (Brooke et al., 2013; Matos et al., 2021) while the optimal temperatures for this species were estimated to be around 6.2-6.7°C (Davies et al., 2008). These behavioural observations support the hypothesis that the temperature used during the experiments (13°C) was likely more suitable for M. oculata and D. cornigera (Naumann et al., 2014; Gori et al., 2015) than for D. pertusum.

During the whole experiment, the three CWC species captured all types of preys. Available literature data on feeding rates ( Table S2 ) show that prey-capture rates differ among the three CWC species, depending also on the type of prey and on the different life stage of the specimens offered as food (e.g., naupliar stages vs. adults) as well as the temperature and the flow velocity under with the experiments are run.

In our study, prey-capture rates of the three species of CWCs showed a preference for the mysid M. relicta, and subordinately for the branchiopod A. salina. Several studies reported that the diet of CWCs in the field is based on zooplankton, such as copepods (Henrich and Freiwald, 1997; Kiriakoulakis et al., 2005; Naumann et al., 2015). A. salina has the same size of these target copepods (Orejas et al., 2019 and references therein). Among the three different species, M. oculata is the one with smallest polyps (5-10 mm in diameter) (Orejas et al., 2019), so the greater capture rates for rotifers (around 350 µm in length) appears as the result of the selection of a prey with a suitable size compared to the size of the corals’ polyps.

Comparing measurements of prey-capture rates from different experiments is difficult because capture rates may overestimate the real ingestion rate if the prey is not efficiently transferred to the gut (Purser et al., 2010). Trapped food may be partly lost due to the sloppy feeding (Møller, 2004) or could not be assimilated, so prey-capture rates cannot directly be translated into ingestion rates or even in assimilation rates (Orejas et al., 2019). Because of this, we integrated capture-rates results with SIA outcomes to determine the assimilated food.

The values of δ ¹³C measured at T₀ in this study for the three CWCs species (from –21 to –19.5 ‰) fit well with that of North Atlantic CWCs (from –22.2 to –19.3 ‰) (Duineveld et al., 2004; Sherwood et al., 2008). These values are slightly more negative than that for CWCs from the Strait of Sicily, South of Malta (from –18.9 to –18.2 ‰, CNR cruise CORAL of the RV ‘Urania’, 450–600 m) (pers. comm. M. Taviani), but this can be justified by the different characteristics of the Adriatic basin and of the Sicily Strait (higher salinity and temperature of this latter, Simoncelli et al., 2014), which typically influence the isotopic composition of the POM (Fanelli et al., 2013; Conese et al., 2019). Our results are consistent also with data from the CWCs coral province of Santa Maria di Leuca for D. pertusum and M. oculata (δ ¹³C from –19 to –21 ‰) (Carlier et al., 2009). In our experiment, a decrease in δ ¹³C in D. pertusum and M. oculata was observed at T_f, although δ ¹³C values remain like those reported for the Ionian Sea (Carlier et al., 2009). These results likely support the hypothesis of a high similarity, in terms of food sources, between the South Adriatic (Bari Canyon) and S. Maria di Leuca CWC provinces. On the contrary, an increase in δ ¹³C for D. cornigera was observed probably due to the preferential ingestion of M. relicta. The polyps of this coral measure 20-40 mm in diameter (Orejas et al., 2019) and mysids are likely more suitable to the dimensions of this species (Gori et al., 2015; Orejas et al., 2019).

The values of δ ¹⁵N at T₀ in all the species were more negative (about halved) than those obtained for D. pertusum and M. oculata analysed in the Santa Maria di Leuca CWC province (values of δ ¹⁵N from 6.9 to 10.1 ‰) (Carlier et al., 2009) or in the Atlantic (Duineveld et al., 2004; Kiriakoulakis et al., 2005). The ¹⁵N-depletion could be caused by both thermal shocks that occurred during the recovery from the seabed, and/or by light exposure and decompression (Orejas et al., 2019). In fact, corals started eating only after a period of acclimation. The results from the simmr mixing model show that at T_0, all the three investigated species had an isotopic composition close to that of the POM of the region (i.e., the Bari Canyon; Carlier et al., 2009). POM is easier to consume and less energy-expensive to capture compared to whole animal preys (living or frozen), so our results confirm that CWCs are able to use also fine organic particles as food source (Mortensen, 2001; Mueller et al., 2014; Orejas et al., 2016). These results suggest that corals filtered POM from surrounding seawater. It is probable that CWCs can sustain their basal metabolism feeding only on POM when other food sources are not available. As an example, the values of δ ¹⁵N of D. pertusum and M. oculata in the Rockall Bank are very close to that of obligate filter-feeding taxa like tunicates and bivalves that are known to feed only on organic particles (Duineveld et al., 2007). In our experiment, after 30 days, the isotopic values of corals reared in aquaria were closer to those of POM of the northern Adriatic Sea that is probably incorporated in their soft bodies faster than the other food sources. Among them, according to simmr results, the mysids M. relicta contributed to the corals’ diet for ~17%. This prey seemed to be the preferred one, also in terms of feeding rates. This crustacean is a member of the family Mysidae, the same family of Boreomysis arctica (Krøyer, 1861) and B. megalops (G.O. Sars, 1872) that are among the most abundant supra-benthic species on the upper and middle slope of the Ionian (Madurell and Cartes, 2003) and the Catalan Sea (Cartes et al., 2011), and are likely among the main natural preys of CWCs living in these areas. Considering together results of prey-capture rates and results of SIA allows to suggest that the maintenance in aquaria of CWCs could be optimised by a diet based on the supplying of frozen Mysis sp. instead of the most common used and live-prey A. salina. Supplying the corals with Mysis sp. can also solve the problem of the depletion of some components, like fatty acids, in the tissue of corals fed only with A. salina (Larsson et al., 2013). Discrepancies between prey-capture rates and the degree of assimilation of the food sources in the soft bodies of the corals should be correlated with the utilization of the ingested food for respiration, for maintenance and growth, tissue growth and storage, reproduction and the release of mucus as dissolved organic matter (Orejas et al., 2019) that can also be re-ingested ad assimilated as a strategy to withstand several months without food supply (Mueller et al., 2014).

M. oculata seems to be a more generalist species, with the greatest trophic diversity, as evidenced by the wide δ ¹³C range and the greatest CD values, and the highest feeding plasticity compared to the other two CWC species. A generalist behaviour in deep-sea ecosystems represents an advantage, as the species can rely on available food source (POM, zooplankton, phytodetritus) in an environment where food availability may be heterogeneous in time and space (Gori et al., 2018). Opportunistic feeding behaviour was previously reported for D. pertusum (Mortensen, 2001; van Oevelen et al., 2009; Mueller et al., 2014; Orejas et al., 2016), M. oculata and D. cornigera (Gori et al., 2018). In the Mediterranean basin, M. oculata seems to be the most abundant CWC species: it is up to 50 times more abundant than D. pertusum in Cap de Creus and Lacaze-Duthiers canyons in the Gulf of Lion (Gori et al., 2013), exclusively present in the eastern Ligurian sea (Fanelli et al., 2017), and dominates the CWC communities in the Santa Maria di Leuca coral province (Vertino et al., 2010), in Bari Canyon (Freiwald et al., 2009) and in the Alboran Sea (Corbera et al., 2019). The higher abundance of this species in the basin may be related also to its greater range of food items and its wider ability to exploit different food sources.

There is growing evidence that conservation measures alone, such as the creation of offshore MPAs or Fishery Restricted Areas (FRAs), albeit necessary to avoid the negative effects of the bottom trawling (Huvenne et al., 2016), are not sufficient to protect these vulnerable habitats from the numerous synergistic impacts that threaten them (Ragnarsson et al., 2016). Moreover, it is known that CWC colonies are slow growing species, which require decades to reach a diameter of 1.5-2 m and possibly thousands of years to build a reef (10-30 m thick; Fosså et al., 2002). Due to their low growth rates, the expected natural recovery rates can be very slow.

Active restoration actions for CWC reefs will require rearing and/or maintenance in aquaria of nubbins taken from healthy donor colonies and transplantation of these fragments into degraded mounds (Van Dover et al., 2014). Promising results of pilot transplantation experiments have been reported in literature for D. pertusum (Da Ros et al., 2019 and references therein). The outcomes of our experiments confirm that it is possible to maintain CWCs in aquaria and allow us to identify the best feeding conditions to keep them in aquaria. Our study provides novel information on the trophic items preferred by D. pertusum, M. oculata and D. cornigera in the Mediterranean Sea. Maintaining corals in aquaria and improving their wellness in captivity may positively impinge on their growth or reproduction success, which may turn into the generation of nubbins successfully transplantable in degraded reefs. Additionally, transplanting healthy corals will increase the chances of obtaining a positive outcome of the effort made, thus contributing to the achievement of the goal of a successful restoration of degraded ecosystem as several frameworks and directives foresee for the next future.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

RD, MT, AD’A conceived the study. RD, MT, EF and AD’A designed the sampling strategy. LA and MT performed the sampling. ZDR and EF analysed the samples. ZDR and EF performed the statistical analyses. ZDR wrote the original draft of the paper, with contributions from EF, RD, MT, AD’A and LA revised the original draft. All authors have read and agreed to the published version of the manuscript.

This research was carried out within the frame of the MIUR “Flagship project RITMARE” and the European Union Horizon 2020 projects, Marine Ecosystem Restoration in Changing European Seas (MERCES), grant agreement No. 689518.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.