Madrepora oculata forms large frameworks in hypoxic waters off Angola (SE Atlantic)

This study aims to map the occurrence and distribution of Madrepora oculata and to quantify density and colony sizes across recently discovered coral mounds off Angola. Despite the fact that the Angolan populations of M. oculata thrive under extreme hypoxic conditions within the local oxygen minimum zone, they reveal colonies with remarkable heights of up to 1250 mm—which are the tallest colonies ever recorded for this species—and average densities of 0.53 ± 0.37 (SD) colonies m−2. This is particularly noteworthy as these values are comparable to those documented in areas without any oxygen constraints. The results of this study show that the distribution pattern documented for M. oculata appear to be linked to the specific regional environmental conditions off Angola, which have been recorded in the direct vicinity of the thriving coral community. Additionally, an estimated average colony age of 95 ± 76 (SD) years (total estimated age range: 16–369 years) indicates relatively old M. oculata populations colonizing the Angolan coral mounds. Finally, the characteristics of the Angolan populations are benchmarked and discussed in the light of the existing knowledge on M. oculata gained from the North Atlantic and Mediterranean Sea.


Results
Distribution and density patterns of the Angolan Madrepora oculata populations. Madrepora oculata was present on four of the six explored Angolan coral mounds (Buffalo, Castle, Scary, and Valentine mounds; Table 1), where a total of 83 colonies were discovered. The locations of these colonies were marked on high-resolution bathymetric maps (grid cell size: 10 m) providing detailed shaded-relief views of the mounds (Fig. 2b-e). The species was most abundant on Buffalo mound with 39 registered occurrences, followed by 30 on Castle, eight on Scary, and six on Valentine mound ( Table 1). The overall average density of M. oculata-colonies including all coral mounds was 0.53 ± 0.37 (SD) colonies m −2 , covering on average 1.25 ± 1.28 (SD) m −2 of the seafloor within the analyzed video frames (for details see Table 2).
Even though M. oculata occurred over a total bathymetric range of 330-470 m water depth (Table 3), the majority of colonies (91%, recorded on Castle, Scary and Buffalo mounds) thrived in a very narrow depth range between 330 and 390 m depth. Only on Valentine mound, deep occurrences below 420 m were registered www.nature.com/scientificreports/ (Table 3). Consistently, M. oculata colonies could not be located on the shallow Anna ridge and Snake mound (250-330 m water depth).
Madrepora colony size structure. An extraordinary large M. oculata framework was documented on Scary mound with a maximal height of 1250 mm (Figs. 3,4). Two other remarkably large colonies were observed on Castle and Buffalo mounds with maximal heights of 1150 and 930 mm, respectively. Overall, the total height of the colonies varied between 250 and 1250 mm and colonies were on average 610 ± 210 (SD) mm high ( Table 4). The scarce number of records from most locations did not allow for the performance of a proper analysis of the population structure. However, for Buffalo mound thirty-nine colonies were documented from which twenty M. oculata-colonies were suitable for measurements, most colonies were 400 to 600 mm high, but colonies of other size classes were also present. Sizes represented a typical Gauss distribution (Fig. 5). The alive/dead proportions of the measured colonies revealed similar length values for all colonies on all mounds with 170 ± 30 (SD) mm on average for the live part and 420 ± 170 (SD) mm for the dead part of the framework, respectively ( Table 4).  Table 3; Fig. 6). Interestingly, prior studies presented slightly larger ranges regarding DO (0.5 and 1.5 mL L −1 ) and temperatures (6.8-14.2 °C). These studies additionally incorporated repeated measurements by a conventional CTD through the water column and long-term measurements by benthic landers, and explained the higher variability in DO and temperature by the effect of internal waves and the larger geographical coverage 31,32 . With regards to the shallow mound sites with no M. oculata occurrence (Anna ridge, Snake mound), DO revealed similarly low values, while temperature and salinity were slightly higher than for the other mounds (Table 3; Fig. 6). In contrast, for the deepest part of Valentine mound (480-500 m), for which also no M. oculata occurrences were recorded, DO were slightly higher than for the other mounds, while temperature and salinities were significantly lower (Table 3).

Discussion
Despite the rather extreme environmental setting along the Angolan margin with overall very low DO (minimum DO: 0.5 mL L −1 ) and rather high temperatures (maximum temperatures: 14.2 °C 31,32 ), which, according to previous studies, is expected to cause severe stress on deep-sea benthic organisms such as CWCs 28,36 , vivid and large reef structures were found to be widely distributed there. The coral community in this location was dominated by  Table 2) was clearly higher than the range of values previously documented for the NE Atlantic (0.01 ± 0.03 (SD) to 0.04 ± 0.04 (SD) coverage m −24 ). This pattern might be related to the exceptional large colonies with heights of up to 1125 mm that were documented for the Angola coral mounds ( Table 4). The reason for this different demographic pattern is unclear at this point. However, it is well known that recruitment (in plants) is a result of different processes like settlement, growth and mortality of seedlings 37 , which might also be applicable to coral larvae. Accordingly, we hypothesize that the M. oculata populations off Angola, comprising of few but large colonies, might experience less recruitment as reported for terrestrial forests. There, the presence of many small trees indicated ongoing recruitment, while forests with fewer but larger trees point to less intense recruitment 37 . Nevertheless, the population structure observed on Buffalo  www.nature.com/scientificreports/ mound (where most M. oculata colonies were documented) indicates that the majority of the colonies had heights of 400-800 mm, while only a small amount of large (up to 930 mm in height) as well as of small colonies were present (Fig. 5). The low number of large colonies could be explained by the maximum size that colonies can reach due to physical constrains (e.g., maximal weight and height supported by the coral structure), whereas the low number of small colonies (recruits) might be a result of a so-called "demographic accident". This might be a result of highly variable larval supply recruitment, which results in a large number of individuals of the next size class (Bramanti pers. comm.). Such an event would explain why the M. oculata population on Buffalo mound appeared to be a growing population that has experienced a so called "recruitment accident" (Fig. 5).
Even if the overall low number of M. oculata recorded on the Angolan coral mounds avoids a robust statistical analysis, the depth-related environmental changes-in particular changes of DO and temperature measured with the CTD during the ROV video observations in the direct vicinity of the coral communities ( Fig. 6; Table 3) -provide first insights into potential controls of their distribution and colony size spectrum. This might be an explanation for the observed intra-mound variability (Tables 2, 4). Regardless of the presence or absence of M. oculata, DO were very low ranging from 0.5 to 1.3 mL L −1 (Table 3) at all depths analysed (250-500 m). The majority of M. oculata-colonies (91%) as well as the largest colony sizes (930-1250 mm; Table 4) were documented between 330 and 390 m water depth (Castle, Scary, and Buffalo mounds), which correspond to DO of only 0.5 and 0.9 mL L −1 (Fig. 6, Tables 3, 4). Despite the moderate fluctuations, these values are by far the lowest DO previously observed for living CWCs (North Atlantic: ~ 1-2 mL L −118,29,38 ). Nevertheless, they are slightly Table 3. Depth range of living occurrence of Madrepora oculata (DR-LOM) recorded for each cold-water coral mound. Additionally, the range and mean values of environmental parameters (DO: dissolved oxygen concentrations) recorded by a CTD installed on the ROV during video observations are provided (DR-CTD: depth range of CTD measurements).

Coral mound DR-LOM (m) DR-CTD (m)
DO (mL L −1 ) Temperature (°C) Salinity  www.nature.com/scientificreports/ above values reported for an extinct coral population on the Namibia shelf, where DO of below 0.5 mL L −1 prevents a present-day coral colonization 31 and even may have caused their local extinction about 4500 years ago 39 . However, the presence of vivid CWC reefs off Angola indicates that the prevailing low DO constitute no severe stressor per se and that the Angolan coral communities have adapted to such low oxygen levels 32 . It is assumed that the high amount and quality of food off Angola (illustrated by a high net primary production (NPP) of > 3400 g C m −2 day −1 , see Table 6) counteract potential deleterious effects of hypoxia 31 . Additionally, aquaria experiments demonstrated that continuous food supply can neutralize or at least minimize the negative   (Table 3). In contrast, temperatures are considerably higher (mean: 11-12 °C) at shallower depths (250-330 m, Anna ridge and Snake mound) and slightly lower (mean: < 8 °C) at deeper depths (> 470 m, deep part of Valentine mound), where no living specimens were documented (Tables 1, 2). Therefore, Table 4. Measurements of total length and alive and dead proportions of Madrepora oculata colonies (R: range, Av.: average). For each of the 30 colonies, 3-5 measurements were taken to cover the intra-colonial variability; here the minimum and maximum values (displayed as R length) of all measurements of each colony are provided.

Coral mound
No. of colonies   www.nature.com/scientificreports/ the observed rather narrow temperature window is seemingly most suitable for the Angolan M. oculata populations, and interestingly, it perfectly matches the preferred temperature range (8.5-10 °C) reported for living M. oculata occurrences in the North Atlantic 43 . Consequently, as the overall prevailing hypoxic conditions already pose a certain degree of stress to the Angolan CWCs, it is possible that M. oculata is not capable to compensate a further stressor, such as temperatures above or below its preferred temperature range. A comparison of the environmental setting off Angola with environmental conditions prevailing in some Atlantic and Mediterranean reef sites, where M. oculata is present, shows a distinct trend regarding temperature, salinity, DO, and NPP (data compiled from various sources ( 29,(44)(45)(46) ; see Table 6). Madrepora oculata thrives in a wide thermal envelope ranging from cold temperatures of ~ 6 °C at the most northern CWC sites off Norway and Ireland to ~ 12 °C off Angola, and even up to 13 °C in the Mediterranean Sea (Table 6). This remarkably wide thermal tolerance of M. oculata has previously been documented in aquaria experiments 25 (Table 6). Even more impressive, M. oculata displays a wide (inter-regional) tolerance for DO as the species occurs under maximum  www.nature.com/scientificreports/ values of 6.7 mL L −1 off Norway, but also under minimum values as low as 0.5 mL L −1 off Angola. The NPP data shows an opposite trend with the highest NPP corresponding to the southern upwelling areas off Mauritania and Angola (Table 6). On the one hand, this negative correlation between DO and NPP can be explained by the increased oxygen consumption due the remineralization of high fluxes of organic matter. On the other hand, the high availability of food might in turn compensate for the oxygen deficiency causing metabolic stress for the corals 31,32,40 . The interaction between DO (stressor) and food (compensator) has been assumed for the Angolan coral populations 31 , though aquaria experiments with L. pertusa reveal regionally dependent thresholds for DO 27,28 , which point to (a potentially genetic) regional adaptation 30 . This case exemplifies that we are still at the beginning of understanding how multiple factors control the proliferation of M. oculata (and other CWCs) and how local adaptation might enhance the complexity of the system. In this context, the estimated average age of 95 ± 76 (SD) years hints at fairly old M. oculata populations off Angola, which might underline that the species is capable to deal with the regional seasonal and even decadal environmental variability. However, it is questionable, if local adaptation capabilities of CWCs could keep pace with environmental changes related to global warming. Further, it should be taken into account that the ages presented here are only preliminary estimates and that growth rate measurements of Angolan specimens will be essential to confirm or readjust these age estimations. Interestingly, M. oculata exhibits partially similar growth rates in regions with very different environmental conditions such as the Norwegian shelf and the Mediterranean canyons 34,35 (Table 6). On the contrary, these experimental studies also indicated a high variability in growth rates among specimens collected from the same region 34 . Therefore, the (potentially) wide range of ages estimated for the Angolan M. oculata colonies (16-369 years; Table 5) might simply reflect the high intraspecific growth rate variability. In addition, regionally different environmental conditions might be one of the reasons for the overall highly variable colony dimensions (varying from some branches of a couple of centimetres to colonies of > 1 m height; e.g., 24,47 ) and plasticity (displaying different morphotypes and tissue coloration) documented for M. oculata at various sites in the Atlantic and Mediterranean (Fig. 7). In order to understand the response of marine species to environmental changes, it is essential to understand how phenotypic as well as genotypic characteristics have been shaped by the dispersal-selection balance through space and time 48 . The high phenotypic plasticity of M. oculata (Fig. 7) and the scarce knowledge on its phenology and genetic characteristics 49 , makes it difficult to identify the regionally important environmental factors and to decipher their complex interplay favouring (or suppressing) the occurrence, growth and longevity of M. oculata. The observed regional differences highlight the importance of considering the environment as it might well be the case that environmental differences along geographical gradients play an equally important role as the species eco-physiological limits documented in the experimental works. Indeed, environmental factors can act as important selective agents shaping genotype and phenotype composition of populations 50 . Consequently, genotype and phenotype adaptations must be taken into consideration to improve the understanding of the geographical as well as the bathymetric distribution and performance of marine organisms on both global and regional scales.

Region DO (min-max) (mL L −1 ) T (min-max) (°C) S (min-max) NPP (min-max) (g C m −2 day −1 ) Data source
Finally, the expected ocean warming and deoxygenation are predicted to be two of the most important changes affecting CWCs 28,51 . Furthermore, predictions anticipating a decline of fluxes of particulate organic carbon from the surface waters 28,36 will result in a lower energy supply for the CWCs and, thus, potentially reducing their capabilities to cope with increasing stress. The most recent habitat suitability model for the North Atlantic anticipates a reduction of 30% of suitable habitats for M. oculata with a northern shift in median latitudinal distributions (ranging from 1.9° to 4.6° in latitude), and a shift of median suitable depths towards deeper waters 3,28 . However, the geographical differences reflected by the variability of environmental parameters compiled for various Atlantic reef sites with the occurrence of M. oculata further highlight the importance to integrate potential regional adaptions of the species considered, which will be a future challenge for upcoming habitat suitability models.

ROV video surveys.
During the RV Meteor cruise M122 (ANNA) in 2016, coral mounds with thriving reefs at their summits were discovered along the Angolan continental margin 52 . These mounds extended over a distance of ~ 30 nautical miles (nm) and were limited to water depths of ~ 250-500 m (Fig. 2). High-resolution mapping with the two multibeam echosounder systems KONGSBERG EM710 and EM122 (for technical details see 52 ) revealed that their morphologies ranged from complex mound structures to long ridges with heights of up to 100 m (Fig. 2). The remotely operated vehicle (ROV) MARUM-SQUID (manufactured by SAAB Seaeye, UK) was utilized during the cruise M122 to explore the seafloor´s benthic communities colonizing the Angolan coral mounds. During this cruise, a total of seven dives (transects) were performed crossing six coral mounds (see Table 1, Fig. 2). The ROV was equipped with five video and still cameras (for technical details see 52 ). Out of these, the main camera for video documentation was the Insite Pacific MiniZEUS MKII-a full HD camera with a resolution of 2.38 megapixels. A DSPL Wide-i SeaCam with a considerably lower resolution of 750 × 576 pixels was used as a downward-looking camera providing a full overview of the area underneath the vehicle. An Imenco Tigershark still camera acquired images at a resolution of 12 megapixels. For size measurements of organisms on the seafloor, the ROV was further equipped with two Imenco Dusky Shark line lasers, which have the capacity to project two parallel laser beams (lines) at a distance of 30 cm.
Occurrence, density, coverage and size measurements of Madrepora oculata colonies based on ROV footage. The ROV video footage recorded by a MiniZEUS HD camera was visualized with the free software VLC to document and to further analyse the occurrence and distribution of M. oculata colonies on the coral mounds. In addition, screen shots were taken from the MiniZEUS video records, when the living and dead www.nature.com/scientificreports/ parts of the colonies were clearly visible and the laser beams were switched on for size calibration (Fig. 3). These screen shots were used to measure the dead and living parts of the individual colonies following the methodology of Vad et al. 53 and using the free software Image J. For each measured colony (total number: 30), three to five repetitive measurements were taken to cover the intra-colonial variability. The video frames recorded by the DSPL Wide-I SeaCam downward-looking camera were used to calculate species density and coverage of the seafloor as these provided (i) a complete coverage of the seafloor underneath the vehicle, and (ii) facilitated the calculation of the total area covered by using the two parallel laser beams as a scale. Downward-looking video frames appropriate for analyses were selected based on the detailed information regarding the colony locations documented in the MiniZEUS video footage. Despite the low resolution, it was possible to identify the colonies without difficulty as they stood out from the surrounding seafloor and other benthic organisms. Madrepora oculata-colonies were clearly visible in a total of 20 video frames, which were ultimately used to calculate the total area depicted in the video frame. The percentage of the area covered by M. oculata-colonies (expressed as coverage m −2 ) and the colony density (expressed as colonies m 2 ) were determined by using the above-mentioned free software Image J. Due to the small number of colonies that could be measured (as a result of the limited number of suitable video frames), a statistical analysis regarding potential differences in occurrence among mounds was not feasible. www.nature.com/scientificreports/