Introduction

Cold-water corals (CWCs) are heterotrophic cnidarians with an opportunistic suspension-feeding behavior1. In the food-limited deep sea, carbon processing in CWC reefs is higher compared to other adjacent habitats2,3 and trophic interactions are one of the processes contributing to the transfer and cycling of organic carbon between different functional groups within the reef2. Topography-enhanced hydrodynamics4,5,6,7 and vertical downwelling bringing surface waters to the deep8 are some of the mechanisms responsible for the enhanced transport of organic carbon  to CWC reefs, driving CWC occurrence and distribution9,10,11.

The CWC Desmophyllum pertusum, also known as Lophelia pertusa12, forms biogenic frameworks with three-dimensional complexity13,14 representing a model of ecosystem engineering in the deep sea15. As for other framework-forming CWCs, D. pertusum reefs increase the local biodiversity of associated fauna16,17,18,19,20 by (i) providing a complex habitat with multiple ecological niches in areas of enhanced hydrodynamics favoring increased food supply21,22 and (ii) by locally reducing flow velocity and turbulence in the shadowed central zones of the reef improving food capture23,24 and the settlement of larvae23.

Thriving CWC reefs dominated by D. pertusum25, with the presence of Madrepora oculata26, have recently been found between 331 and 473 m water depth along the Angolan margin (SE Atlantic), coinciding with the center of the local oxygen minimum zone (OMZ). The discovery of the Angolan D. pertusum reefs challenged some of the previously assumed ecological requirements for the species27,28, given that CWCs were found in very low oxygen concentrations (0.5 to 1.3 mL L−1) and relatively high temperatures (6.8 to 14.2 °C)11,25,26. Despite the hypoxic conditions, the Angolan CWC reefs are prosperous and harbor a community of associated mega- and macrofauna, mainly composed of sponges, octocorals, and antipatharians29,30. Most studies on the trophic ecology of CWCs are available for communities occurring in normoxic conditions in the Mediterranean Sea31,32,33 and in the North Atlantic Ocean34,35,36,37,38, with scarce information for other deep-sea regions. To date, no trophic studies have been conducted in the recently discovered reefs of Angola.

Previous works showed that D. pertusum is a passive suspension-feeder, capable of feeding on different food sources, including dissolved organic matter (DOM)39,40, particulate organic matter (POM) in the form of phytodetritus from surface primary production37,41 and zooplankton31,34,42,43,44,45. In the Angolan CWC reefs, high-quality and abundant organic matter (OM) resulting from a productive upwelling system46 is available to benthic communities11. This abundant food supply could be a key mechanism for CWC survival under the Angolan OMZ11,47,48,49. A recent study by Gori et al.50, showed that D. pertusum in Angola can maintain high respiration rates under deoxygenation, suggesting acclimation and local adaptation of the species to the environmental conditions of the reefs. Some of the proposed adaptations of benthic organisms to hypoxia might include changes in body size and shape47,49, changes in energetic metabolic pathways49,50 as well as symbiotic associations with anoxic-bearing bacteria51,52,53.  As far as we know, the effect of OMZs in the trophic ecology of CWCs and associated fauna has not yet been described. However, previous studies of trophic webs in OMZs showed the importance of microbe-mediated trophic processes54,55 in delivering OM to benthic organisms under low-oxygen conditions56, in particular, through the role of nitrifying and denitrifying bacteria55,57. Therefore, investigating the trophic interactions in the Angolan CWC reefs could provide new understanding of the food sources available to these communities, and how CWC reef trophic webs in OMZs might differ from those in more oxygenated regions.

Because of their remoteness and therefore the difficulty to obtain samples or conduct in situ experiments, disentangling energy flows within CWC ecosystems is challenging; nonetheless, different methodologies may allow to overcome the inherent difficulties. Gut content analysis is a widely used technique to decipher the diet of both terrestrial and marine organisms and has also been used for deep-sea animals58,59. However, data on the gastrovascular content of CWCs are very rare60 because of the complications in identifying preys due to the small size of the food particles61 and the difficulties in assessing the gut content by dissection of the corals’ polyps (in particular the ones from scleractinian corals). In situ detection of feeding interactions is possible through visual observations62,63, but such interactions are not easily captured on camera, and a large amount of data, over a large temporal scale, is needed. For these reasons, carbon (δ13C) and nitrogen (δ15N) stable isotope analyses have become the most widely used tools to investigate the trophic structures of deep-sea communities64,65,66,67,68,69,70. Stable isotope signals do not decay over time, thus integrating long-term diets and giving information on the dietary choices of an organism over a wide temporal range (from days to months, depending on the tissue turnover)71. In a standard food web, δ15N increases in a stepwise manner from one trophic level to the next, usually resulting in an enrichment of 3.4–5.0 ‰ in δ15N72 from the prey to consumer, thus, allowing to estimate the trophic position of the different organisms73. In addition, there is a small enrichment in δ13C of approximately 1 ‰ from one trophic level to the next74. Furthermore, the δ13C signature provides information of the primary source of the OM supporting the trophic web75. For example, enriched 13C ratios point to photosynthetic production (e.g., from − 22 to − 14 ‰31), while more depleted δ13C values suggest a trophic web fueled by chemoautotrophy (e.g., from − 50 to − 15 ‰ in hydrothermal and methane seep communities76).

In this study, we provide a first insight in the trophic structure of the Angolan CWC reefs, using stable isotope analyses. This work is a pivotal step to understand the trophic interactions of CWC reefs under hypoxic conditions and will contribute with new baseline knowledge on the ecological functioning of a poorly investigated deep-sea area of the SE Atlantic Ocean. This could shed light on how CWCs and associated deep-sea communities might cope with the forecasted impacts of climate change, specifically with the expected expansion of OMZs in the future77.

Results

Carbon and nitrogen stable isotope (δ13C and δ15N) analyses were conducted on benthic megafauna samples and on three types of Particulate Organic Matter (POM; filtered Suspended Particulate Organic Matter (SPOM), SPOM from a sediment trap and POM from sediment) collected from the Angolan CWC reefs. The lowest measured δ13C value was for SPOM sampled at 342 m water depth (− 22.46 ± 0.34 ‰) and the highest was for Echinus sp. (− 12.81 ± 1.65 ‰), while δ15N values ranged from 2.00 ± 1.48 ‰ for SPOM sampled at 532 m water depth to 21.90 ‰ in a Hexactinellid sponge. The carbon and nitrogen stable isotope bi-plot (Fig. 1) represents a total isotopic range of 9.65 ‰ for δ13C and of 19.9 ‰ for δ15N, with a correlation of R = 0.74 (p-value < 2.2 × 10–16) between δ13C and δ15N. The four clusters obtained with the k-means analysis match with the trophic position (TP) calculations (Table 1), indicating that each cluster represents a different trophic guild in the Angolan CWC reefs.

Figure 1
figure 1

Bi-plot of the mean carbon (δ13C) and nitrogen (δ15N) stable isotopes of the main benthic megafauna groups on the Angola CWC reefs and the analyzed particulate organic matter, sampled at different depths: suspended particulate organic matter (SPOM), settling SPOM collected with a sediment trap (SPOMtrap) and sediment. Dashed grey lines represent standard deviation. Polygons represent k-means clusters, with the optimal number of clusters decided based on the elbow, silhouette and gap statistic methods. Labels in the bi-plot represent the abbreviations of the different taxonomic groups, found in Table 2. The values presented here for SPOM342, SPOM532, SPOMtrap342 and SPOMtrap526 were previously published by Hanz et al.9.

Table 1 Mean ratios of δ13C and δ15N (± standard deviation (s.d.)) and number of samples (n) analyzed for each type of particulate organic matter and taxonomic group, with information of the label used in Fig. 1.

The three types of POM analyzed are included in cluster 1, presenting a mean stable isotopic composition of − 21.57 ± 0.59 ‰ and 4.83 ± 1.58 ‰ for δ13C and δ15N, respectively. Within this cluster, sediment sampled at 345 m water depth presented the most enriched values (− 20.84 ‰ for δ13C and 6.19 ‰ for δ15N), being 4.19 ‰ more enriched, in terms of δ15N, than SPOM sampled at 532 m water depth. The two types of SPOM analyzed (filtered SPOM and SPOM from the trap), presented a mean δ13C value of − 21.87 ± 0.44 ‰ and a mean δ15N value of 4.23 ± 1.64 ‰, while sediment had more enriched mean values of − 20.96 ± 0.17 ‰ for δ13C and of 6.05 ± 0.21 ‰ for δ15N. The SPOM samples collected at deeper sites present more enriched mean δ13C signatures (− 21.65 ± 0.35 ‰) than the ones collected at shallower sites (− 22.10 ± 0.52).

The stable isotopic signature of consumers ranged from 10.52 ± 1.07 ‰ to 21.9 ‰ in δ15N, corresponding to M. oculata and a hexactinellid sponge, respectively, whereas the consumer with most the δ13C-depleted signature was D. pertusum (− 19.77 ± 1.0 ‰) and with the most δ13C-enriched signature was Echinus sp. (− 12.81 ± 1.65 ‰).

The consumers included in cluster 2 varied from − 19.77 ± 1.0 ‰ to − 17.25 ‰, in terms of δ13C and from 10.52 ± 1.07 ‰ to 14.00 ‰, in terms of δ15N, presenting a mean δ13C value of − 18.70 ± 0.83 ‰ and a mean δ15N value of 12.90 ± 1.00 ‰. The mean calculated TP of this cluster is 3.54 ± 0.29, with the maximum TP (3.87) occupied by a polynoid annelid (Polynoidae sp.). The consumers included in this cluster are all analyzed CWC species as well as the above-mentioned annelid (Polynoidae sp.), a hydrozoan (unidentified), a sea anemone (Actinaria) and a fish (Myctophidae sp.). The distance between the mean stable isotope ratios of cluster 1 (POM) to cluster 2 is 2.9 ‰ for δ13C and 8.07 ‰ for δ15N, which corresponds to 2.37 trophic levels, if a trophic enrichment for δ15N of 3.4 ‰ is considered.

All the echinoderm taxa analyzed, plus the polychaete Eunice norvegica are part of cluster 3. This group presents the most enriched carbon stable isotope signatures with mean δ13C of − 14.00 ± 1.08 ‰ and mean δ15N of 15.00 ± 1.59 ‰. The mean TP of this group is 4.16 ± 0.47.

The organism most enriched in δ15N was a glass sponge (Hexactinellida sp.; 21.90 ‰). All analyzed sponge taxa (cluster 4) presented the highest mean δ15N ratios (mean 20.20 ± 1.87 ‰) consequently corresponding, according to the TP calculations, to the highest mean TP in the trophic web (mean TP of 5.69 ± 0.53).

Discussion

The basic structure of the food web of CWC reefs offshore Angola was deciphered, for the first time. Our results indicate a reef trophic web composed of four different trophic guilds (Fig. 2). The carbon isotope signatures of POM (− 22.46 to − 20.84 ‰) indicate that the CWC reefs are sustained by photosynthesis-derived primary production from the photic zone, consistent with what was first shown by Le Guilloux et al.78; nonetheless, the gap of one trophic level between POM (cluster 1) and suspension feeders (cluster 2) suggests that an additional food source was missed in our sampling. Although the significant correlation between δ13C and δ15N indicates a network supported by a single primary source65,79,80, the enriched δ13C signatures of predators and detritus-feeders (− 15.39 to − 12.81 ‰) and the high δ15N values of sponges (17.33 to 21.90 ‰) point to different food sources for the consumers at the Angolan CWC reefs.

Figure 2
figure 2

Schematic representation of the proposed trophic web for the Angolan cold-water coral (CWC) reefs. Trophic groups are based on carbon and nitrogen stable isotope (δ13C and δ15N) analyses. Mean trophic position (TP) (± s.d.) is indicated for each group as well as the proposed food sources. Created with BioRender.com.

The Angolan basin is an area with exceptionally high primary productivity, derived from the Benguela upwelling system46,81, and fueled by nutrients delivered by the Cuanza and Congo Rivers82. It has been argued that a nutritive food source delivered by the upwelling system could play an important role in allowing the corals to tolerate the potential stress that hypoxia and high temperatures off Angola could cause25,26,83,84, since an abundant and easily digestible food source could help balance the energetic metabolic demands of species under oxygen stress47,48. Indeed, Hanz et al.11, showed that high-quality and abundant OM, in the form of fresh phytodetritus, is available to the CWC reefs, corroborated by the low isotopic signature of δ15N of SPOM, as well as the low C/N ratios and the little degraded phytopigments of the analyzed OM11. In addition, fluvial input of terrestrial OM slightly enriches the δ13C signature of the SPOM and sediment (− 22.46 to − 21.40 ‰)11. Our results show that D. pertusum and M. oculata are 2 ‰ enriched compared to the mean δ13C signature of the analyzed POM (see cluster 1, Fig. 1), indicating that possibly the corals do not directly feed on the terrigenous-enriched POM. However, the high input of nutrients associated with the terrestrial material transported with the riverine discharges could, instead, be important to maintain the primary productivity that supports the baseline of the trophic web85. The OM fluxes to the Angolan basin fluctuate seasonally86, because of seasonal changes in the upwelling system and riverine discharges87. It is possible that this seasonality is also reflected in the diet of the CWCs, with the corals exploiting fresh phytodetritus, when available, but also opportunistically feeding on other types of food, when fresher phytodetritus is absent1,45,88.

At the Angolan CWC reefs, D. pertusum and M. oculata presented slightly higher δ15N values (11.56 ± 1.37 ‰ and 10.52 ± 1.07 ‰, respectively) than the ones reported for other CWC reefs (7.6 to 10.1 ‰ for D. pertusum and 6.9 to 10 ‰ for M. oculata) in the NE Atlantic Ocean34,35,37,89 and the Mediterranean Sea31, even though the δ15N signature of SPOM in our study (2.00 to 5.32 ‰) is comparable to the ones reported in literature (2.2 to 8.3 ‰31,34,35,37,41,44,79). In our study area in the SE Atlantic Ocean, δ15N ratios show an enrichment of 7.33 ‰ and 6.29 ‰ between SPOM and D. pertusum and M. oculata, respectively. In other studies, the observed δ15N enrichment between D. pertusum and SPOM ranged from 2.7 to 7.1 ‰31,34,35,37,44. The observed isotopic enrichment at the Angolan reefs is only similar to the one observed in the Galicia Bank (7.1 ‰34) where the authors could not define a conclusive food source for the CWCs, suggesting a mixture of different sources34. In fact, isotopic enrichment might differ between deep-sea environments and conditions35,44,70 depending on the available food sources, the metabolic rate of organisms and the isotopic composition of the surrounding water column. Furthermore, the observed δ15N enrichment in our study could also point to the presence of an additional trophic level between the phytodetritus and the corals, missed in our sampling. This gap could indicate that, in addition to fresh phytodetritus, zooplankton might be an important food source for the CWCs offshore Angola, as previously observed in other CWC reefs dominated by scleractinians31,34,89,90. Indeed, zooplankton is a nutritional rich food source for cold-water scleractinians45,91, octocorals92,93, and antipatharians31,33,94. The Angolan basin harbors abundant and diverse zooplankton communities95 whose dial vertical migrations coincide with the water depth where CWCs are found (331 to 473 m)96. The zooplankton δ15N ratios (ranging from 3.5 to 10 ‰) from other regions of the Atlantic sampled in the same depth range (300 to 500 m) as our study34,37,97,98,99 fit with the expected enrichment of the intermediate trophic level. Thus, zooplankton prey should be considered available to the examined CWCs. Accordingly, we hypothesize that the corals’ diet is most likely based on fresh phytodetritus, when available100, but zooplankton might contribute the most to the CWCs diet. However, the role of zooplankton as primary food source in the Angola CWC reefs still needs to be investigated in future studies.

Black coral and octocorals occupy the same trophic guild of D. pertusum and M. oculata (cluster 2, Fig. 1) however, their δ13C and δ15N signatures are more enriched than those of the latter two. These differences might be explained by taxon-specific (morphological and physiological) requirements36,101, since colony morphology102 and polyp size103 influence the rates of food capture104 and the selection of prey type and size61 between CWC species. This possibly leads to a less-opportunistic feeding strategy and food selection61 focused in capturing more of one type of food source, potentially zooplankton, resulting in differences in their trophic strategies93 and, consequently, isotopic signatures.

The crustacean decapod Munida sp., a fish of the family Myctophidae, and a polynoid polychaete also belong to the same trophic group as CWCs. This result highlights their suspension-feeding behavior and, most likely, all of them exploit the same food sources as the CWCs. Moreover, given that small fish are known to preferably feed on zooplankton97, this finding further supports the hypothesis of the importance of zooplankton as a food source to cluster 2. On the other hand, the polychaete E. norvegica and the analyzed echinoderms (Asteroids, Ophiuroids and Echinus sp.) are the most 13C-enriched organisms (δ13C from − 15.39 to − 12.81 ‰) with a corresponding mean TP of 4.16 ± 0.47 (cluster 3). The polychaete E. norvegica is known to live in close association with D. pertusum and M. oculata13,105,106,107 and to contribute to reef formation106, where the polychaete could benefit from extra food supply from the coral host by feeding directly on detritus in the corallites105. In the Angolan CWC reefs, E. norvegica was present at the basal part of the reef-building coral colonies29. The δ15N of E. norvegica is in the same range as the one for the other polychaete analyzed in this study. However, the comparative δ13C enrichment of E. norvegica indicates an exploitation of different food sources that could be related to its association with D. pertusum since it has been shown that, in the presence of the coral, E. norvegica assimilates more carbon through selective feeding on bigger particles108. It is possible that bigger particles consumed by E. norvegica indicate fresher particles, with a higher content of chlorophyll-a, resulting in 13C-enrichment109.

The high standard deviation in δ15N of asteroids and in the δ13C of Echinus sp. could be attributed to the opportunistic feeding behavior of the organisms in these groups110,111,112,113. Given that all the organisms included in this trophic group are mobile species, they are potentially able to exploit different food sources within the reef, including detritus31,114, zooplankton37,115 or, by directly predating corals and sponges116. The latter seems more likely, since ROV video observations have shown the sea urchin Echinus sp. on top of D. pertusum colonies (Fig. 3), and the dissection of collected specimens confirmed the presence of fragments of D. pertusum in their gut’s contents29, supporting the evidence of its predatory feeding behavior. Moreover, echinoderms in the CWC reef presented the most enriched δ13C ratios (an enrichment of 5 ‰ compared to cluster 2). One possible explanation to the 13C-enrichement of this group could be a “deep-sea sponge loop”40,70,117,118, where the sponges’ feeding on bacteria and DOM, will result in the release of 13C-enriched particulate detritus that can be assimilated by associated fauna, both through detrital feeding and by directly feeding on sponges through a predatory pathway40,118. However, this interpretation should be taken with caution since in situ experiments are needed to support a “deep-sea sponge loop”.

Figure 3
figure 3

Sea urchins from the genus Echinus sp. on top of Desmophyllum pertusum at Valentine Mounds. Photo Credits: MARUM ROV SQUID.

Sponges are active filter-feeders and, therefore, it was expected that sponges act as primary consumers on the Angolan CWC reefs. However, sponges exhibited the highest δ15N values (from 17.33 to 21.90 ‰) of all samples analyzed in this study. These results are in the range of, and sometimes higher than, values usually attributed to top predators in marine trophic webs119,120. According to TP calculations following Post,72, the 15N-enriched ratios of sponges correspond to a mean TP of 5.69 ± 0.53 (Table 1), being inconsistent with the expected filter feeding behavior in the trophic web. Hanz et al.70 showed that TP calculations based on δ15N might not be appropriate to infer the TP of deep-sea sponges because when other biomarkers are used (such as compound specific isotope analyses), sponges act as normal filter feeders. Instead, the high δ15N values of sponges possibly indicate that sponges rely on additional food sources that have not been included in our stable isotope food model.

Sponges are capable of opportunistically exploiting a wide range of both organic and inorganic food sources, including DOM45,121, phyto- and bacterioplankton118,122,123,124,125, or even small zooplankton through the specialization of a group of sponges (Family Cladorhizidae) to utilize carnivory feeding modes126,127. In fact, the enriched δ15N values of hexactinellid sponges in the Angolan reefs are not entirely surprising since similarly high ratios have also been reported for deep-sea sponges in several other studies in the North Atlantic37,79,98, Pacific125,128, Arctic70,114,129 and Antarctic66,130 Oceans. The 15N-enriched values of sponges could be attributed to intense nitrogen cycling pathways in the sponge’s tissue70,131 driven by (1) the sponge holobiont37,79, defined here as the sponge and its associated microbial communities, and/or (2) the sponge’s feeding on heterotrophic re-suspended bacteria that are 15N-enriched because of the uptake of waste material of higher trophic levels70,125,130.

The Angolan CWC reefs occur in the center of the local OMZ and OMZs are zones with thriving microbial communities55 and complex nitrogen cycling dynamics54,132,133. The tissue of sponges is rich in microbial communities53,129,134,135 and the nitrogen isotope fractionation in sponges changes according to their bacterial richness53,70,125, i.e., if High Microbial Abundance (HMA) sponges or Low Microbial Abundance (LMA) sponges70,136. The sponge holobiont takes energy from DOM137 and, as a result of OM degradation, will release ammonium (NH4+)70,123,138, an isotopically depleted metabolic end-product139, leading to increased δ15N in the sponge tissue. Moreover, symbiotic relationships with bacteria, capable of both denitrification and anammox131,140,141, play an important role in sponge survival under low oxygen environmental conditions52,53,142,143. Likewise, it has been shown by Middelburg et al.52 that denitrification can occur in D. pertusum, due to the dominance of denitrifying bacteria in the coral’s holobiont and this process is enhanced by low-oxygen conditions. It is beyond the scope of our study to identify the exact physiological mechanisms used by benthic organisms to cope with low-oxygen environments, but the observed δ15N ratios of sponges in our study hints that interactions with microbes could be a plausible mechanism to withstand the hypoxic conditions. Given the slightly enriched δ15N values of D. pertusum, the CWC holobiont could also potentially play a role in the coral’s coping with the hypoxia conditions of the Angolan CWC reefs50,52, although this is more likely to be more significant for sponges given their enriched stable nitrogen isotope values.

This study gives the first insights into the trophic structure of the Angolan CWC reefs, where different feeding strategies are utilized by the organisms inhabiting in this OMZ. Our observations provide the first evidence pointing to an important role of zooplankton as food source for CWCs in Angola, and to a strong reliance of sponges, through their holobiont, on bacteria as food source in hypoxic conditions. The use of other biomarkers in future studies, such as compound specific stable isotope and fatty acids analyses, or in situ experiments with isotopically labelled food will contribute to a better understanding of CWCs and sponge’s nutrition. We further suggest that, in the future, special attention should be given to the potential role of the holobiont in allowing CWCs and sponges to withstand the conditions of the Angolan and other OMZs.

Methods

Study area

The Angola basin is located along the SW African continent in the SE Atlantic Ocean and belongs to the Benguela Current Large Marine Ecosystem (BCLME)144. The region is influenced by the Benguela upwelling system, one of the world’s most productive145, with an estimated primary production of 156 million t C yr–1146. Induced by coastal parallel winds, the upwelling of nutrient-rich cold waters results in enhanced primary productivity81, while the remineralization of high concentrations of OM in the water column results in severe mid-depth oxygen depletion and in a pronounced OMZ147 that coincides with the presence of the Angola CWC mounds25. In addition, the area receives a high input of riverine discharges from the Congo and Cuanza Rivers, which leads to further increase in primary production due to a high terrigenous nutrient input11,82. At approximately 17°S, the cold nutrient-rich waters transported northward by the Benguela Current interact with the warm nutrient-poor Angola Current148, forming the Angola-Benguela Frontal Zone (ABFZ)149. Oxygen-depleted and nutrient-rich waters are characteristic from 70 to 600 m water depth, indicating the presence of the South Atlantic Central Water (SACW)150, which is transported southward by the Angola Current.

The Angola CWC reefs cover long ridges to individual coral mounds that can reach up to 100 m above the surrounding sea floor. These CWC ridges and mounds have formed on millennial and longer time scales151, through successive periods of CWC reef development. They are composed of CWC fragments and other shells loosely embedded in fine hemipelagic sediments25,29,30.

Sample collection

Samples were collected in January 2016 during the M122 (“ANNA”) expedition on board the R/V Meteor29. In total, 18 reef sites topping seven different CWC mound complexes (Fig. 4, Table 2) were sampled for stable isotope analyses. Samples of organisms belonging to the taxa Porifera, Cnidaria, Arthropoda, Annelida, Echinodermata and Chordata were collected by means of a box corer (box dimensions: 50 × 50 cm, 55 cm high), a Van-Veen grab sampler or the Remotely Operated Vehicle (ROV) SQUID (MARUM, Bremen, Germany)29 (Supplementary Table S1). Each organism was identified to the lowest possible taxonomic level, rinsed in seawater and put in plastic bags or vials.

Figure 4
figure 4

Study area map. (a) Location of the cold-water coral (CWC) mounds off Angola (red box) and of the Angola Current and the Angola-Benguela Frontal Zone (ABFZ). Bathymetry grid from GEBCO compilation160. (b) Locations, where different types of samples for stable isotope analysis were collected on the various CWC reefs (black boxes). Contour lines represent 50 m depth intervals. Bathymetry data was acquired during R/V Meteor cruise M122161. Maps created using the Open-Source software QGIS Version 3.20.3-Odense (http://www.qgis.org).

Table 2 Metainformation of the number of samples collected (n), type of sample (Fauna, Sediment or suspended particulate organic matter (SPOM)) and sampling gear at the different sampling stations (Station ID).

To investigate potential food sources for the megabenthic fauna, three types of POM were collected: SPOM from the water column, settling material from the sediment trap (SPOM trap), and sediment. SPOM was collected at two depths (342 and 532 m) at 40 cm above the seafloor using a McLane phytoplankton pump, attached to the ALBEX lander (NIOZ), with a maximum of 7.5 L of water pumped every 2 h for a period of 48 h over 24 GF/F filters (47 mm Whatman™ GF/F filters pre-combusted at 450 °C) (for more details see Hanz et al.11). Settling SPOM (SPOM trap) was also collected with a sediment trap (Technicap PPS4/3) attached to the ALBEX lander, with the collector at 2 m above the bottom over a sampling duration of 2.5 days, at both 342 and 526 m depth. Additionally, two sediment samples were collected at two depths (259 and 345 m): one with the box corer and another one with the Van-Veen grab sampler. All organisms and POM samples were stored on board at – 20 °C and, afterwards, were freeze-dried in the laboratory and stored at − 20 °C until further analysis.

Sample preparation for stable isotope analyses

Fragments of freeze-dried faunal samples were ground to powder with a mortar and, depending on the quantity of sample available, two subsamples of 10 to 50 mg were weighted. One subsample was left untreated for δ15N determination, whereas the other was acidified with the addition of 10% HCl drop-by-drop until effervescence ceased and used for δ13C determination. The acidification of samples is a common practice in stable isotope analyses and aims to remove inorganic calcium carbonate from the organisms, known to interfere with δ13C ratios112. All samples analyzed for δ13C were acidified, except fish samples due to their low carbonate content152. After acidification, the subsamples were oven-dried at 50 °C for 72 h. Afterwards, three pseudo replicates of each subsample were weighted with a precision balance (± 0.001 mg) into tin capsules (11 × 4 mm, Elementar Microanalysis) to conduct the isotopic composition analyses. Depending on the taxonomic group, subsamples of 0.4 to 4 mg were weighted.

Sample preparation for the determination of δ15N and δ13C for the three types of POM considered (filtered SPOM, SPOM trap and sediment) is described in detail in Hanz et al.9. These samples were acidified by fuming HCl (20%) vapor over night for δ13C determination. Three replicates of sediment (1.6 to 3.6 mg for δ13C and 9 mg for δ15N) and sediment trap material (1 mg δ13C and 4 to 5 mg for δ15N) were weighted into tin capsules. For SPOM, a ¼ section of each GF/F filter was transferred into tin capsules for δ13C and δ15N analyses.

Stable isotope analyses

Analyses of benthic megafauna samples were performed using the Elementar IsoPrime 100 isotope ratio–mass spectrometry (IR–MS) instrument (IsoPrime Ltd.) coupled to a CNS elemental analyzer (Elementar Vario Pyro Cube EA CNS; Elementar Analysensysteme GmbH). Stable carbon isotopic values (δ13C) were quality checked and calibrated by using the reference materials Glucose (BCR-657) and Polyethylene (IAEA-CH-7), while for the stable nitrogen isotopic (δ15N) values potassium nitrate (USGS32), caffeine (IAEA600) and ammonium sulfate (USGS25) were used.

The δ15N and δ13C values of filtered SPOM, SPOM trap and sediment were analyzed by a Delta V Advantage IR–MS coupled online to an elemental analyzer (Flash 2000 EA-IRMS) by a ConFlo IV (Thermo Fisher Scientific Inc.). Benzoic acid and acetanilide were used as standards for δ13C, whereas acetanilide, urea and casein were used as standards for δ15N (for more details see Hanz et al.9).

Vienna Pee Dee belemnite (V.P.D.B.) for carbon, and atmospheric N2 (Air) for nitrogen, were used as reference materials, and stable isotope values are here reported with respect to those. Precision is based on the standard deviation between replicate analyses and was < 0.50 ‰ for δ15N and < 0.20 ‰ for δ13C for faunal samples, and < 0.15 ‰ for δ15N and δ13C for POM.

Stable isotope ratios (δ13C and δ15N), in relation to the standards, were calculated as following:

$${\updelta }^{13}\mathrm{C or }{\updelta }^{15}\mathrm{N }(\mathrm{\permil }) = [\frac{\mathrm{R sample}}{\mathrm{R standard}}-1] \times 1000$$
(1)

where R corresponds to 13C/12C or 15N/14N of the analyzed sample (Rsample) and standard used (Rstandard).

Data analysis

The Trophic Position (TP) of consumers was calculated following Post72, as in similar studies on deep-sea habitats98,115,153,154. The TP of each consumer was calculated using the following equation:

$${\text{TP}}_{{\text{n}}} = \lambda + \left( {\delta^{{{15}}} {\text{N}}_{{\text{n}}} - \delta^{{{15}}} {\text{N}}_{{{\text{baseline}}}} } \right)/{\text{TEF}}$$
(2)

where TPn corresponds to the trophic position of the organism, δ15Nn corresponds to its stable nitrogen isotope ratios, δ15Nbaseline represents the nitrogen isotopic composition of the baseline, TEF corresponds to the Trophic Enrichment Factor for δ15N and λ represents the trophic level occupied by the baseline. The δ15N of the baseline considered was 4.23 ± 1.65 ‰, corresponding to the mean ratio measured for SPOM (λ = 1). A TEF of 3.4 ‰ for δ15N was considered, since it has been regarded as appropriate in other CWC habitats31,37.

To explore the different trophic guilds of the Angolan CWC reefs, a k-means clustering analysis was applied to the δ13C and δ15N data, computed using the “factoextra” package155 within R Studio156 under R version 4.1.3157. The optimal number of clusters (k = 4) was investigated using the “NbClust” R package158 by applying three different methods: elbow, silhouette, and gap statistic methods (Supplementary Fig. S2). The Pearson correlation between δ13C and δ15N ratios was calculated using the “cor.test” function available in the “stats” package in R. Since fauna and POM samples were collected along different CWC mounds, it was not possible in our study to assess how the stable isotopic ratios would differ between mounds.