The two studied mid-Norwegian Lophelia reef sites represent offshore and inshore environmental conditions of established CWC ecosystems ( Figure 1 ). Offshore, the approximately 14 km long Sula Reef Complex on the Sula Ridge off the coast of Sør-Trøndelag was investigated in the mid-western area of the reef chain at about 300 m water depth ( Figures 1B , 2A ). Inshore, the CWC bank at Nord-Leksa (Nord-Leksa Reef) just outside the entrance of the Trondheimsfjord and about 40 nm from Trondheim was studied at 170–215 m depth ( Figure 1B ). For the less studied Nord-Leksa Reef, a bathymetric map ( Figure 2B ) of the reef was generated based on single beam echo sounder data recorded through the ship-board navigation system and GPS data of the vessel’s route to obtain an idea of the dimensions. The Nord-Leksa Reef rises from about 210 to 145 m water depth and has a dimension of about 1,700 m in west-to-east and about 600 m in north-to-south direction, with a saddle-like depression between two main reef tops ( Figure 2B ). Similar to the Sula Reef, Lophelia pertusa is the most dominant coral species at the inshore reef at Nord-Leksa.

In addition to the two studied reef sites, a side experiment was carried out at the shallow Tautra Reef ( Figure 1B ) at 39–50 m, where water samples for characterisation of the biogeochemistry were taken by divers very close to the polyps. Water sampling at such close proximity to the polyps is usually not possible by CTD sampling in the deeper reef areas and the information might help to further understand the linkages of organic matter recycling processes within the reef.

The geochemical and physical sample collections and measurements as well as benthic lander deployments and recoveries were carried out during two cruises with RV Poseidon (GEOMAR, 2015) in June/July 2013 (POS455) and August 2014 (POS473). Lists of stations and water samples are provided in the cruise reports of POS455 and POS473 by (Form et al., 2014, 2015) and are summarised in Table 1 .

Dives for positioning the lander systems, coral sampling, and characterisation of the reefs via video and still photos were carried out with the manned submersible JAGO (GEOMAR, 2017) at Sula and Nord-Leksa reefs. The highly manoeuvrable research submersible is certified to an operating depth of up to 400 m and accommodates a pilot and an observer. The vehicle is equipped with an underwater navigation and positioning system (USBL), a compass, depth gauges, vertical and horizontal scanning sonar, underwater acoustic telephone communication, digital video (Full-HD 1080p/50p), still cameras and oceanographic sensors. With JAGO’s manipulator arm, sample collection and transportation of small research devices can be carried out such as the positioning of the benthic lander systems.

During the two cruises in 2013 and 2014, JAGO spent 116 hours under water during 40 dives at the two study sites. The JAGO video and still footage was further used to compare general reef characteristics like the reef health status according to Flögel et al. (2014) and the associated fauna between the reef sites. Since the main objectives of the dives were carried out for the purpose of deploying and recovering equipment and collecting samples, the video footage is not suitable for extensive habitat analysis, as distance to the reef changed continuously and no transects were followed. Some marked features can be described and compared between the study sites nonetheless and are described in the results (section 3.1).

Three lander systems equipped with oceanographic and environmental data loggers were deployed in the vicinity of the CWC reefs at the two reef locations Sula and Nord-Leksa for over 13 months ( Figure 2 ). The deployment sites were chosen to allow comparison of the hydrodynamics inshore and offshore (Sula vs. Nord-Leksa, Figures 2A, B ), as well as the central reef and the surrounding reef margin in Nord-Leksa ( Figure 2B ). After a detailed survey with JAGO during POS455 in 2013 to find suitable flat bottom topography in the vicinity of live corals for lander deployment, two benthic landers (Satellite Lander Modules, SLM’s, GEOMAR, Germany, Figure 2D ) were deployed on 2^nd July 2013 on the southwestern part of the Nord-Leksa Reef. The first SLM was deployed at 63°36.487’N and 09°22.956’E at a water depth of 175 m within the living reef structure in the valley between the two reef plateaus (referred to as SLM_reef hereafter), surrounded by a diverse fauna of many live L. pertusa, gorgonians, Acesta excavata and Mycale sp. within a few metres to the west and dead coral rubble and soft corals to the east. The second SLM was deployed at 63°36.454’N and 09°22.915’E at 217 m depth adjacent to the reef (referred to as SLM_off-reef hereafter). The landers were deployed 70 m apart from each other ( Figure 2B ). Both landers were equipped with the following instruments, located 50 cm above the seafloor: an Acoustic Doppler Current Profiler (SBE 300kHz ADCP, Teledyne RD Instruments) to measure water column flow speed and direction in 10-minute intervals, a conductivity, temperature and pressure measuring instrument (SBE 16 PLUS CTD, Sea-Bird Scientific), and sensors to monitor dissolved oxygen (SBE 43), pH (SBE 27 pH and O.R.P. (Redox) sensor), turbidity and chlorophyll fluorescence (WETLabs ECO-FLNTU(RT)D) measuring in 15-minute intervals.

At Sula, a SEAGUARD^® Recording Current Meter (TD 262b SEAGUARD^® RCM, AANDERAA Data Instruments, Bergen, Norway) equipped with a ZPulse Doppler Current Sensor (DCS), a pressure/temperature sensor combination, and a conductivity/temperature sensor combination was mounted in a pyramid-shaped POM frame ( Figure 2C ) and deployed with the sensors located 75 cm above the seafloor. Oceanographic data comprised tilt-corrected horizontal flow speed and direction, pressure, conductivity, and temperature (both, conductivity and pressure sensors recorded temperature as well, of which the conductivity sensor-based temperature was used here). The parameters were logged in a 30-minute interval. The SEAGUARD^® lander was deployed on 4^th July 2013 at 64°06.66’N and 08°07.12’E in 305 m water depth 50–75 m from extensive live L. pertusa reef framework in the mid-west of the northern Sula Ridge reef area ( Figure 2A ).

The resolution and accuracy of the measurements of all three landers are shown in Table 2 . In August 2014, almost 14 months after deployment, landers were successfully recovered.

In Nord-Leksa, the benthic landers turned out to be too light to cope with the strong bottom currents at the deployment depths. SLM_off-reef moved uphill twice during the deployment period, with total vertical movement of 5 m. It first moved between 17^th and 19^th November 2013 from 216.5 to 214 m, and again between 13^th to 14^th April 2014 to 211 m depth. SLM_reef moved downwards from deployment depth of 175 m to ~ 180 m within the first hour of deployment, where it remained for 5 days. The lander moved several times over the year between depths of 180 and 213 m. At the end of January 2014, the Lander-CTD stopped working at 207 m depth. Both SLMs were stable at the bottom and recording over three periods: 2^nd – 7^th July 2013, 22^nd August – 16^th November 2013 and 19^th November 2013 – 30^th January 2014. The SEAGUARD^® lander in Sula remained in place and delivered continuous data of all sensors.

From Nord-Leksa, we have reliable time series of ADCP, CTD, oxygen of both landers and turbidity from SLM_reef, while the pH sensors of both landers and the turbidity sensor of SLM_off-reef malfunctioned. The first five days of deployment were used to compare the environmental conditions within Nord-Leksa and only the measurements of SLM_off-reef were used to compare the intra-annual conditions between the inshore and offshore sites. The conductivity sensors had drifts that amounted to -0.3 g kg^-1 (SLM) and to -1.2 g kg^-1 (SEAGUARD) over the deployment periods. These drifts have been considered when the hydrographical variables (S_A, Θ and σ_Θ) were calculated.

To measure water column characteristics at both sites, a CTD system (SBE 911 plus, Sea-Bird Scientific) was used during both cruises. The employed CTD system was built into a rosette housing with 12 10-litre water sampling bottles (Niskin-type). Water samples were taken close to the bottom and in pre-defined depths of the water column (see Table 1 ). In addition to the CTD system, the rosette frame was equipped with sensors measuring dissolved oxygen, fluorescence of chlorophyll-a, and turbidity. Of the latter two only data from POS455 (2013) are available. Calibrations of the sensors were performed prior to the cruise in the laboratory and all parameters yielded coefficients for a linear fit.

Further processing of the hydrographic data was performed using the software SBE Data Processing^© (V7.26.7, Sea-Bird Scientific (2013)) for SLM landers and CTD cast data. For further conversions and visualisation of the oceanographic and environmental data, MATLAB 9.0 (R2016a, The MathWorks, Inc. (1984)) was used. For subsequent data analysis, the raw CTD and ADCP data were converted. CTD data were converted to absolute salinity (S_A), conservative temperature (Θ), potential density (ρ) and potential density anomaly (σ_Θ), i.e. sigma-theta values according to TEOS-10 standard (McDougall and Barker, 2011). The current measurements were corrected for the local magnetic declination based on IGRF-11 model data (Finlay et al., 2010). The site-characteristic dominant tidal frequencies, ω, and their amplitudes, a, were analysed with the harmonic analysis toolbox T_Tide (Pawlowicz et al., 2002). The tidal signals were analysed by using bottom pressure and horizontal velocity fields. Only tidal signals with a signal-to-noise ratio > 2 are considered to be significant.

For CWC habitat suitability and conservation analyses, large-scale habitat suitability models are utilised, using global databases and regional ocean models (Davies and Guinotte, 2011; Yesson et al., 2017; Morato et al., 2020). Comparison of field observations to these databases and ocean models is crucial, primarily to ground truth model data and to increase model reliability in general. The carbonate chemistry parameters and nutrients are usually obtained from global databases such as GLODAP (Lauvset et al., 2016) and World Ocean Atlas (Garcia et al., 2013) with low resolution of 1° and down-scaled. Since this resolution is not suitable for inshore sites, we compared the global values from 64.5°N, 8.5°E at 200–250 m depth for Sula and 64.5°N 9.5°E at 125–200 m depth for Nord-Leksa as closest approximation to our studied inshore site. For coastal Norway, the coastal currents are modelled with daily 800-m resolution with the NorKyst-model (Albretsen et al., 2011). This model data was compared with lander deployments from the closest model points for summer 2013 to summer 2014 and with the Tautra Reef area.

Both Nord-Leksa and Sula reef sites belong to category I (healthy reefs and mounds) out of three according to the classification of CWC reef statuses described in Flögel et al. (2014), as both reef habitats are characterised by several 100 m² horizontal area and > 2/3 vertical extent of living colonies, therewith comprising large biogenic constructions with discrete morphologies ( Figure 4 ). Nevertheless, the colony morphology differs between sites. At Nord-Leksa, the corals form ‘cauliflower’-like colonies with compact polyp arrangement, while at Sula ‘cauliflower’-like colonies are more dispersedly branched with thinner and more extended polyps ( Figures 4G, H ).

Both sites have rich associated fauna. At Nord-Leksa the living coral reef framework is surrounded by a large coral rubble bed of dead L. pertusa fragments from depths of ~ 200 to 160 m, while the reef framework at Sula is surrounded by sandy bottom with sponge-covered boulders or drop stones. In Nord-Leksa, the rubble zone is ‘coral garden’-like, i.e. it is characterised by a benthic community dominated by non-reef-forming soft corals such as the gorgonians Paragorgia arborea and Primnoa resaediformes, Capnella glomerata, and Anthelia borealis. The rubble zone at Nord-Leksa is abundant in sponges (particularly Mycale lingua) and used as a habitat by a range of mobile fauna such as crustaceans (e.g. Munidopsis sp.), bivalves (including the large file clam Acesta excavata), echinoderms and tunicates. Between about 160 and 152 m, a transition zone follows, which makes up the reef base with first live L. pertusa appearing in the white and orange colour morph similar to the Sula Reef, and occasionally M. oculata colonies settling on dead coral framework and rubble. Upslope, the reef framework is characterised by an increasing number of live L. pertusa colonies accompanied by an increasing number of M. lingua, while the abundance of soft corals ceases.

At the offshore study site at Sula, the reef is not surrounded by large rubble fields, but the sandy, rocky seabed (at ~ 302 m) with hard-bottom sponge aggregates transits directly into the base of reef framework of mostly dead L. pertusa (~ 302–296 m) with a very diverse fauna. Associated species in the Sula Reef encompass the same phyla as in Nord-Leksa but appear to have a larger variety on the species level among poriferans, cnidarians and crustaceans. Numerous large actinians and shrimps (Pandalus sp.) are prominent, as well as chaetognaths as abundant zooplankton species. The most remarkable difference between the reef sites (inshore vs. offshore), however, is the abundance and distribution of sponges in the live coral reef zone. The reef top at Nord-Leksa (~ 152–140 m) is interspersed with Mycale sp. ( Figure 4B ) and occasionally also gorgonians, cluttered with mobile fauna such as Munidopsis sp. and Gorgonocephalus sp., whereas the living reef cover at Sula (~ 296–283 m) is free of sponges ( Figure 4A ) and only the dead basis of the reef in between broken-up colonies inhabits a large variety of particularly large sponges (e.g. very commonly Geodia sp.) and large gorgonian colonies of Paragorgia and Primnoa. In comparison to Nord-Leksa, only few A. excavata were found at Sula and often only shells. The few individual living A. excavata had far brighter, i.e. cleaner shells, whereas at Nord-Leksa shells were overgrown and more strongly eroded by bioeroders.

Similar fish species were spotted at both sites, including saithe (Pollachius virens), tusk (Brosme brosme), monkfish (Lophius sp.), and red fish (Sebastes sp.). The latter is the most common species living within the reef framework but was more abundant at Sula. Most common cartilaginous fish included chimaeras at Sula and the velvet belly lantern shark at Nord-Leksa, but also at Sula shark and ray eggs were observed.

During the first week after deployment (from 2^nd to 7^th July 2013), both landers remained stable at their deployment positions ( Figure 7 ). Although SLM_reef was located in shallower depths, the water was cooler (Θ = 7.78 ± 0.01°C), more saline (S_A = 35.23 ± 0.04 g kg^-1) and, hence, denser (σ_Θ = 27.36 ± 0.03 kg m^-3) than at SLM_off-reef (Θ = 7.79 ± 0.01°C, S_A = 35.07 ± 0.02 g kg^-1 and σ_Θ = 27.23 ± 0.02 kg m^-3) ( Figures 7A–F ). The variance of the hydrographic variables was higher at SLM_reef than at SLM_off-reef. Oxygen concentration levels were similar at both lander positions during the first week of deployment, but fluctuated more at SLM_reef as well ( Figures 7G, H ). The bottom flow was stronger at SLM_reef (U_mean = 13.53 cm s^-1, U_max = 41.59 cm s^-1) than at SLM_off-reef (U_mean = 12.9 cm s^-1, U_max = 37.66 cm s^-1) ( Figures 7I, J ). The flow direction oscillated between SSE and ESE at SLM_reef whereas at SLM_off-reef, the flow oscillated between E and WSW ( Figures 7K, L ). For the other comparable periods (22^th August – 16^th November 2013 and 19^th November 2013 – 30^th January 2014), the SLM_reef lander measured slightly higher salinities (ΔS_A = 0.2 g kg^-1) and bottom flow speeds (ΔU_mean = 1 cm s^-1) than SLM_off-reef, but fluctuations and range of the oceanographic variables were close to each other (not shown).

The near-reef flow was relatively strong both at Nord-Leksa and Sula but was in the range of flow speeds reported for L. pertusa reef distribution in Norwegian waters (Buhl-Mortensen and Freiwald, 2023) and other CWC reefs in the NE Atlantic (White and Dorschel, 2010; Lim et al., 2020), with flow speeds on topographic highs > 20 cm s^-1 (Mortensen et al., 2001; Thiem et al., 2006). The mean bottom flow speed was stronger in Nord-Leksa (~ 20 cm s^-1) compared to Sula (~ 8 cm s^-1). High flow speeds (> 80 cm s^-1) occurred more regularly at Nord-Leksa than Sula throughout the year. Given the bathymetry, single high flow speed events will likely have a higher impact at the Leksa Reef, where waters are pushed to a narrow strait/fjord. Maximum flow speeds recorded in late autumn were in a similar range at both sites (100–150 cm s^-1), which is about five-fold higher than previously reported maximum flow velocities at the Sula Ridge (< 30 cm s^-1, Eide, 1978; Roberts et al., 2005). These maximum flow speed events coincided with strong wind speeds ( Supplementary Figure S1 ) and therefore suggests storm events during this time. In November/December, the forcing resulting from this caused the movement of both landers at Nord-Leksa. Similarly high maximum flow speeds of 76 cm s^-1 and 114 cm s^-1 were recorded at two sites in the upper Porcupine Bank Canyon on the Irish continental shelf, which likewise led to landers being toppled upslope, suggesting even higher current velocities than recorded (Lim et al., 2020). Lim and colleagues suggest that in the Porcupine Bank area a higher live coral abundance is found at sites with lower coral-facing mean flow speeds and that sites that are consistently exposed to higher flow speeds have higher coral rubble percentages due to enhanced physical and biological erosion of the coral framework. In contrast, flow velocity at the inshore site of this study was consistently higher directly at the flourishing reef compared to the off-reef lander location surrounded by rubble where this could be compared, suggesting that the differences in flow direction (De Clippele et al., 2018) and environmental conditions play a greater role in coral distribution at the Nord-Leksa Reef than variations in current speed. In mid-Norway, the reefs are well-developed, large and dense framework structures (Hennige et al., 2014), which could make them less susceptible towards high current velocities than thickets and individual mounds. Nonetheless, current speeds were generally higher at coral sites when compared to non-coral bearing sites in the study by Lim et al. (2020) as well, underlining that corals require high flow speeds to thrive.

Relatively high flow speeds are thought to increase the food encounter rates of the corals (Hunter, 1989; Sebens et al., 1998) and prevent the polyps from clogging with sediments (Brooke et al., 2009; Larsson and Purser, 2011). In contrast, laboratory experiments have demonstrated the efficient prey capture rates of CWCs like L. pertusa to peak at relatively low flow speeds < 7 cm s^-1 (for zooplankton at flow rates of ~ 2.5 cm s^-1 and for phytoplankton ~ 5 cm s^-1), as with stronger flow the prey could escape from the polyps (Purser et al., 2010; Orejas et al., 2016). The CWC framework slows down the ambient flow due to friction (Roberts et al., 2009) and dense coral framework creates stability for the colony (Chamberlain and Graus, 1975; Roberts et al., 2009; Hennige et al., 2014). The CWC framework thus acts as a natural sediment trap allowing the corals to capture food and utilise the enhanced particle delivery. The corals are adapted to a very dynamic environment as various hydrodynamic processes on different temporal scales cause food supply to come in periodically rather than constantly (Maier et al., 2023), linked to small-scale processes such as internal tidal activity (Davies et al., 2009; de Froe et al., 2022) as well as larger scale circulation patterns and specific water masses (Schulz et al., 2020; Mohn et al., 2023). On continental margins like the shelf edge along Norway and plateaus such as Rockall Bank, currents continuously interact with the seafloor, creating good feeding conditions for the corals (Frederiksen et al., 1992; White et al., 2005; Thiem et al., 2006). In the fjords, tidal effects play a more important role for food supply than offshore and higher fluorescence as well as higher turbidity at depths of CWC occurrence at Nord-Leksa may indicate higher local food supply at the time of sampling due to more turbulent conditions compared to Sula.

The colony morphology determines how much the water flow slows down within the framework. Coral colonies with long and thin branches slow down the flow less than more compact colonies with short and thick branches (Kaandorp and Sloot, 2001; Chindapol et al., 2013). The long-branched colony morphology, which is more common at Sula, is associated with low near-reef flow speeds (Kaandorp, 1999; Chindapol et al., 2013). The recorded near-bottom flow speed at Sula was < 7 cm s^-1 for 60% of the time during lander deployment, while at Nord-Leksa it was only 19%, thus most of the time higher flow speeds were recorded during the lander deployment. The difference in the mean near-reef flow speeds likely explains the more extended colony shapes at Sula. High flow speed events (> 100 cm s^-1) in winter may break exposed CWC skeleton in Sula more easily than at Nord-Leksa due to the thinner branches. Considering < 7 cm s^-1 to be the optimal flow rate for prey capture not only in experimental conditions but also in the natural environment suggests that prey capture efficiency is supported better at the Sula Ridge than at Nord-Leksa based on the year-long continuous flow measurements presented here.

However, optimal flow conditions for coral feeding change within a reef system as the flow speed and direction change on a diurnal and seasonal time scale, thus different parts of the reef have different flow conditions for feeding. This affects the growth direction of the corals, which might help them to optimise food capture (Wheeler et al., 2007; Buhl-Mortensen et al., 2010; De Clippele et al., 2017, 2018).

Moreover, local flow dynamics also change with the complexity of the framework, with larger structures modifying their surrounding flow environment and potentially reducing the favourable flow conditions for other colonies in the vicinity, as suggested in a recent modelling study by Hennige et al. (2021) using the ‘Goldilocks Principle’. That corals effectively optimise their own local flow environment through habitat engineering is supported by the finding that surface complexity within and between colonies is similarly variable independent of the differences of the compactness at inshore versus offshore reef sites (i.e., the more compact growth form of corals exposed to strong currents in fjords for example) (Sanna et al., 2023).

The flow regime is controlled by short-term phenomena like tides and storms in Nord-Leksa and by seasonal atmospheric forcing at Sula. The short-term variability in the flow field is controlled by tides and storms. In winter 2013-2014, two storms with winds > 24.5 m s^-1 were recorded over Trøndelag on 16–17^th November 2013 (MET-info, 2013) and 12^th December 2013 (MET-info and Fagerlid, 2014), which coincided with steep increases in flow speeds recorded by the landers ( Supplementary Figure S1 ; Figure 8 ). The first storm coincided with the vertical lander movements in Nord-Leksa showing that the whole water column in the fjord system is affected when the waters are pushed towards the fjord entrance. At Nord-Leksa, the flow regime is tidally-driven with strong semi-diurnal (M₂) flow (~ 20 cm s^-1) towards the main current direction. This results in a larger amplitude than the average semi-diurnal tidal current on the central Norwegian shelf (amplitudes of 3–12 cm s⁻¹) (Gjevik and Straume, 1989; Haugan et al., 1991). The tidal flow controls the bi-directional flow at Nord-Leksa supporting the observed ‘cauliflower’ colony shape. At Sula, the tidal flow is weak (~ 2 cm s^-1), rotated westward due to an anticyclonic (clockwise) topographic effect of the Haltenbank and differs from the main current direction (northward).

At Sula, the seasonal variations in the flow regime are driven by the quasi-stationary topographical effects of Haltenbank and Frøyabank and the seasonal changes in the NAC-NCC -interaction. The flow is mostly directed by a northward flowing coastal branch of the NCC (Ljøen and Nakken, 1969; Eide, 1979; Poulain et al., 1996; Sætre, 1999). This strong current creates a distinct temperature front between the coastal branch of the NCC and the water masses further offshore (Audunson et al., 1981). The stratification is weakest in late autumn and winter, which yields to observed fresher and warmer surface water at greater depths ( Figures 8A, D , 9A ). In winter, the anticyclonic eddies are strongest due to changes in the prevailing wind direction (Eide, 1979; Sætre, 1999). This yields intrusions of high-salinity Atlantic Water (AW) to the shelf (Haugan et al., 1991) and the observed pronounced strong southward flow with increasing salinity in mid-winter ( Figures 8D, N , 9A, C ).

In the Trondheimsfjord, the seasonal variations in the flow field are controlled by fresh-water input and the dominant wind directions at the coast. Although we do not have flow measurements from Tautra, the flow has previously been reported to be similarly dominated by impacts of fresh-water input and strong tides as in Nord-Leksa (Mortensen and Fosså, 2001). The Trondheimsfjord has a mean fresh water runoff of 725 m³ s^-1 with a spring flood maximum in April/May of up to 6,430 m³ s^-1 (Sakshaug and Killingtveit, 2000). This partially coincides with the period of northerly winds and the shallow NCC that are believed to explain the inflow of dense AW to the fjord from February to June (Sakshaug and Killingtveit, 2000).

CWCs live in relatively cold (Θ = 6.83–8.97°C) and oxygen-rich (O₂ = 5.7–6.6 mL L^-1) waters in mid-Norway compared with other known CWC reef sites (Roberts et al., 2006; Dodds et al., 2007; Freiwald et al., 2009; Davies et al., 2010; Davies and Guinotte, 2011; White et al., 2012; Ramos et al., 2017; Hanz et al., 2019). The very low salinity values down to 31.12 g kg^-1 recorded at Sula are most likely a result of sensor errors (values are peaks in unfiltered data). Disregarding those peaks shows a comparable salinity range (33.50–36.06 g kg^-1) to other CWC habitats (Freiwald et al., 2004; Davies et al., 2008). However, our results revealed a broader range of salinities than previous studies suggested for the Sula Reef (35.0–35.2 psu) but keeping in mind that most comparable studies represent data solely from summer conditions, higher variability throughout the year can be expected. The measured annual range of hydrographic variables was larger than expected (Sula: > 1.5°C and > 2 g kg^-1 and Nord-Leksa: > 1.5°C and > 1 g kg^-1), but moderate when compared to the largest measured temperature fluctuations at a CWC site of 9°C within a day in the Cape Lookout area, NW Atlantic (Mienis et al., 2014). Due to coastal fresh-water influence and relatively warm winter bottom water temperatures, the bottom water was less dense over the whole year at Nord-Leksa and less dense from late autumn to winter at Sula than the suggested density range for NE Atlantic CWC sites (σ_Θ = 27.35–27.65 kg m^-3) (Dullo et al., 2008).

Oxygen concentrations in Nord-Leksa ranged in a well-oxygenated state compared to other CWC reef or mound sites (Hebbeln et al., 2020) and showed relatively little variability throughout the entire year. Despite an off-set (~ 0.3–0.9 mL L^-1) between lander and CTD oxygen sensors, both systems revealed values in the same order of magnitude within the range of previously recorded oxygen concentrations in Norwegian and other NE Atlantic CWC habitats (Dullo et al., 2008; Findlay et al., 2014; Hebbeln et al., 2020). Similar to observations in White et al. (2012) maximum values of dissolved oxygen were found in March/April and decreased to lowest values in November/December and is likely due to the supply and decay of organic matter. Lower oxygen concentrations in Nord-Leksa than in Sula may be due to higher O₂ consumption rates inshore as a result of biological activity and eutrophication, which might be influenced by aquaculture activities as well (Aksnes et al., 2019). Despite being located at the entrance of the fjord and not enclosed in the fjord system as well as experiencing similarly high flow velocities like in Sula, it may also be that the open ocean shelf reef sites have higher renewal rates of dissolved oxygen due to the more direct influence of oceanic AW.

The carbonate chemistry variables measured here were comparable to values previously reported for Norwegian CWCs (Büscher et al., 2019) and indicate saturation of calcium carbonate, promoting calcification at Nord-Leksa and Sula. In the Norwegian Sea, the ASH is ~ 2,000 m (Jones et al., 2018). The CWCs at the studied sites thrive in shallower and therewith saturated waters (Ω_Ar = 1.57–2.77) in a similar range like other CWC occurrences in the NE Atlantic in the Rockall Trough region and the Outer Hebrides (McGrath et al., 2012; Findlay et al., 2014). For CWCs in the Gulf of Mexico in the NW Atlantic, occurring at deeper depths than studied here, between ~400 – 600 m, lower aragonite saturation values (Ω_Ar ≈ 1.15–1.44) were reported, as a result of higher mean DIC values (Georgian et al., 2016). Flögel et al. (2014) demonstrated that living CWCs in the NE Atlantic are linked to low DIC concentrations (< 2,170 µmol kg^-1). This is confirmed for Sula and Nord-Leksa with DIC values ranging from 2,070 to 2,157 µmol kg^-1, while in waters outside the NE Atlantic, live CWCs occur at higher DIC concentrations up to 2,470 µmol kg^-1 (in the Marmara Sea), attributable to different water masses present in those areas (McCulloch et al., 2012; Findlay et al., 2014). McGrath et al. (2012) have shown that DIC concentrations have increased in the Rockall Trough by nearly 20 µmol kg^-1 in subsurface waters over two decades (1991 – 2010), concomitant with an equivalent decrease in pH and Ω_Ar and a shoaling of the ASH. Hence, environmental conditions are changing in the North Atlantic and may shift boundary conditions outside the tolerance limits of benthic organisms such as CWCs.

Nutrient concentrations are scarcely reported from CWC reefs. Findlay et al. (2014) assembled reported data from known CWC reefs sites in the NE Atlantic, revealing a large range for dissolved inorganic nitrate (4.1–23.4 µmol L^-1) and a smaller range for phosphate (0.6–1.6 µmol L^-1). Analysed nitrate (7.8–13.0 µmol L^-1) and phosphate (0.6–0.7 µmol L^-1) values were in the lower range of those reported for the Rockall Trough region, but similar to values found in a comparable CWC reef setting (the Mingulay Reef Complex) at the Outer Hebrides.

Phosphate values within the reef structures at Tautra Reef were considerably lower (< 0.45 µmol kg^-1) than the water above the reef and observed at other NE Atlantic CWC sites. Perhaps, these lower phosphate values close to the polyps point to higher turnover rates in the nutrient cycling within the reef structures. de Froe et al. (2019) recently demonstrated that CWC reef communities and specifically the living CWCs influence benthic nitrogen cycling by circumventing nitrification and NH₄ ⁺ production. High NH₄ ⁺ concentrations close to the coral reefs may thus be indicative of a NH₄ ⁺ release of the corals as a result of nitrogen cycling. However, due to the sensitive nature of nutrient analysis, especially ammonium, and this being only a single point measurement, similar sampling and measurements should be repeated for comparison before jumping to conclusions.

Both sites have rich associated fauna using the CWC framework for shelter, reproduction and as feeding grounds (Henry and Roberts, 2007; Baillon et al., 2012; Buhl-Mortensen et al., 2017), but higher diversity of coral-associated invertebrates was previously found in Nord-Leksa and other mid-Norwegian inshore reefs compared to the Sula Reef (Mortensen and Fosså, 2006). The reef framework provides substrate for other benthic filter feeders such as sponges (Orejas and Jiménez, 2017), which can form a substantial component of CWC reef biomass (Hogg et al., 2010). Reef sponges function as nutrient recyclers, substrate stabilizers, bioeroders, and as a food source and habitat for other organisms (Wulff, 2001). At Tisler Reef in the Oslofjord in southern Norway where similar environmental conditions prevail as at Nord-Leksa, both Geodia sp. and M. lingua were found growing in dead L. pertusa framework, but only M. lingua was found growing within live L. pertusa (De Clippele et al., 2018). Thus, M. lingua does not compete with the corals for hard substrate and can co-occur by growing within Lophelia colonies as was observed in Nord-Leksa, while Geodia sp. presumably competes for the same substrate as the corals and may not be able to outcompete L. pertusa’s growth rates in regions where substrate availability is limited (Purser et al., 2013; De Clippele et al., 2018). The coverage of M. lingua within the framework of L. pertusa at Nord-Leksa, whereas found on dead coral substrate between colonies at Sula, supports the observation that substrate availability is limited at Nord-Leksa and species have to co-exist, despite largely comparable environmental conditions like offshore. The absence of Geodia sp. could also be explained by small-scale differences in environmental conditions, e.g. with lower water temperatures by approximately half a degree at Sula on average and considerably lower bottom flow speeds. Temperatures at Tisler Reef, where Geodia sp. occur, are however generally higher than temperatures reached at Nord-Leksa at its maximum (compare De Clippele et al., 2018).