Cold-Water Coral Reefs in the Langenuen Fjord, Southwestern Norway—A Window into Future Environmental Change

: Ocean warming and acidiﬁcation pose serious threats to cold-water corals (CWCs) and the surrounding habitat. Yet, little is known about the role of natural short-term and seasonal environmental variability, which could be pivotal to determine the resilience of CWCs in a changing environment. Here, we provide continuous observational data of the hydrodynamic regime (recorded using two benthic landers) and point measurements of the carbonate and nutrient systems from ﬁve Lophelia pertusa reefs in the Langenuen Fjord, southwestern Norway, from 2016 to 2017. In this fjord setting, we found that over a tidal (<24 h) cycle during winter storms, the variability of measured parameters at CWC depths was comparable to the intra-annual variability, demonstrating that single point measurements are not sufﬁcient for documenting (and monitoring) the biogeochemical conditions at CWC sites. Due to seasonal and diurnal forcing, parts of the reefs experienced temperatures up to 4 ◦ C warmer (i.e., >12 ◦ C) than the mean conditions and high C T concentrations of 20 µ mol kg − 1 over the suggested threshold for healthy CWC reefs (i.e., >2170 µ mol kg − 1 ). Combined with hindcast measurements, our ﬁndings indicate that these shallow fjord reefs may act as an early hotspot for ocean warming and acidiﬁcation. We predict that corals in Langenuen will face seasonally high temperatures (>18 ◦ C) and hypoxic and corrosive conditions within this century. Therefore, these fjord coral communities could forewarn us of the coming consequences of climate change on CWC diversity and function.


Introduction
Cold-water coral (CWC) reefs form complex habitats in the deep sea, supporting rich associated fauna [1][2][3][4]. They play a major role in the circulation of organic carbon in the deep sea [5,6]. The reefs are recognized as vulnerable marine ecosystems (VMEs) by the United Nations [7] and as threatened and declining habitats by the OSPAR commission [8]. Warming and acidifying waters (i.e., decreasing pH and aragonite saturation state, Ω Ar ) are predicted to pose imminent and serious threats to CWC reefs within the next decades [9,10]. Hence, there has been an intensified focus to assess the range of biogeochemical and physical conditions tolerated by CWCs in situ [11][12][13][14][15][16] and in laboratory conditions [17][18][19][20][21]. Finally, a focus is on determining the baseline for optimal conditions for coral growth and reef development [22]. However, there is very little data on the small-scale temporal and spatial variability of biogeochemical and physical conditions in CWC settings, which could be pivotal in understanding their resilience to climate change, as observed for tropical coral reefs [23].
In the North Atlantic, CWC reefs are most commonly build by Lophelia pertusa (syn. Desmophyllum pertusum [24]), with around a third of all known L. pertusa occurrences being from Norwegian waters [25]. L. pertusa reefs are generally found on and around elevated nutrient data are primarily available from global data sets [56,57]. These are useful for making large-scale habitat predictions [13,58], but they lack the fine-scale spatial and temporal coverage that is required to understand the baseline of physicochemical dynamics around CWC reefs that plays a key role in determining the resilience of CWC reefs to future climate change.
This study investigated the natural variability of biogeochemical and physical conditions at five CWC reefs located within a narrow fjord in western Norway. This was achieved by collating water column measurements of carbonate chemistry and inorganic nutrient parameters for multiple stations and timepoints over 1.5 years and combining those with high temporal resolution data from two benthic observatories deployed at one of the reefs. Comparisons of environmental data from reefs growing on vertical walls with those growing on the fjord floor were used to elucidate the relative importance of topography-flow interaction and biogeochemical conditions for CWC occurrences.

Study Site
The Langenuen Fjord in southwestern Norway, south of Bergen (Figure 1), is a 35 km long north-south trending water passage connecting the Korsfjorden with the Hardangerfjord. The system opens up to the Atlantic through Korsfjorden at its northern edge, through Selbjørnsfjorden in the central parts, and through Bømlafjorden at its southern end. Langenuen is~300 m deep. The deepest point measures 568 m and is located in the northern portion between the islands of Huftarøy and Reksteren. The average width of the fjord is 2.9 km in the northern section and 1.7 km in the southern section. The fjord acts as an ocean water inlet to the Hardangerfjord [59]. It hosts several L. pertusa-dominated CWC reefs on the eastern side of the fjord, at depths between 80 and 240 m [60,61]. This study concentrates on five of them in two different settings: corals covering an elevated topographic feature on the fjord bottom here called "bank reef" and corals living on steep walls or cliffs, "wall reefs". The two bank reefs are Nakken (NK, 59.

Water Column Measurements
Water samples were taken at several depths using a 12-bottle Niskin rosette (Ro) system. For water column hydrography profiling, an instrument measuring in situ conductivity, temperature, and pressure (CTD) attached to the water sampling system was used. A total of 12 CTD and 52 CTD/Ro (CTD/rosette water sampling) casts were carried out across the CWC sites (HH, NK, HN, SN, BK). Measurements were taken during six cruises between February 2016 and August 2017, onboard RV Håkon Mosby, RV Kristine Bonnevie, RV H. Brattström, and MS Periphylla (Table 1). In October 2016 and January 2017, water sampling was performed over a tidal cycle at HH (26-h) and NK (16-h), respectively. On other sites and times, water sampling was performed a maximum of once per site (Table 1). Data collected in October 2016 and August 2017 have been reported earlier in Skjelvan et al. (2016) [62] and Jones et al. (2018) [48], respectively.   Processing of the raw CTD data was performed using the software package SBE Data Processing [63]. The wall reefs (HH, HN, and SN) were investigated above, below, and at the respective depths of living L. pertusa corals. At the bank reefs (NK and BV), samples were taken at the surface, at intermediate depths (at 80 or 100 m and at 150 m), and at~5 m above the reef.

Benthic Landers
Two seafloor monitoring lander systems were deployed at NK bank reef to study the topographical effect of the bank on the flow between May 2016 and April 2017. One of the landers (SLM) was deployed south of the CWC bank at 63.608 • N 9.383 • E at 211 m water depth, while the other one (MLM) was deployed on the CWC bank at 63.609 • N 9.382 • E at 196 m water depth,~125 m apart from each other ( Figure 1C). Both landers were equipped with an Acoustic Doppler Current Profiler (ADCP, Teledyne RD Instruments) to measure water column flow speed and direction at 6-min intervals in between 3 to 30 m above the sea floor (with SLM) and from 5 m above the sea floor upward to the surface (with MLM). The landers were also equipped with a CTD system (SeaBird Electronics) combined with sensors to monitor dissolved oxygen, turbidity, fluorescence, and pH. Sampling intervals were set to 10 or 30 min. Processing of the raw CTD data was performed using the software package SBE Data Processing [63] and the raw ADCP data with the RDADCP Matlab package [64]. The accuracy and sensitivity of these instruments are shown in Table 2. Unfortunately, the lander on the NK CWC bank (MLM) tilted due to high flow velocities in mid-August. This resulted in a separation of the bottom weight and the buoyancy compartment that includes all sensors. Thus, the sensor head was released and floated upwards to~100 m water depth after 3.5 months deployment time in mid-August, where it continued to measure all parameters. The SLM remained at deployment depth until its recovery in April 2017. ADCP, CTD, oxygen, and SLM turbidity sensors provided reliable time-series data. pH and fluorescence sensors malfunctioned. The conductivity sensor had an instrumental drift that amounted to −0.5 g kg −1 over the deployment period. This drift has been taken into account when the hydrographic variables (Θ, S A , σ Θ ) were calculated.

Analysis of Carbonate Chemistry and Inorganic Nutrients
Seawater was collected from the CTD/Ro system for the determination of dissolved inorganic carbon (C T ) and total alkalinity (A T ) into borosilicate glass bottles (250 mL) with ground glass stoppers. The samples (n = 190) were preserved with 0.05 mL of saturated solution of mercuric chloride (HgCl 2 ) and stored refrigerated and dark until post-cruise analysis. Samples (n = 93) for the determination of nutrients and ammonia were collected directly after the samples had been taken for the A T and C T . For nutrient analysis, 20 mL of water was collected in polyethylene scintillation vials, preserved with 0.2 mL chloroform, and kept refrigerated (4 • C) until analysis occurred within a few weeks after sample collection. For ammonium analysis, waters were collected in the same type of vials and kept frozen (−20 • C) without preservatives until analysis.
A T and C T were measured at the IMR's CO 2 Laboratory following standard procedures [48,62,65]. The remaining carbonate system parameters (pH in total scale pH T , partial CO 2 pressure pCO 2 , aragonite saturation Ω Ar and calcite saturation Ω Ca ) were calculated from A T and C T with the program CO2SYS [66] using the thermodynamic constants from Mehrbach et al. (1973) [67] and KSO4 from Dickson (1990) [68], and refitted by Dickson and Millero (1987) [69] on the total scale. The concentrations of silicate and phosphate were set to zero during the calculations, and the errors introduced by this simplification were negligible compared to uncertainties from other sources.

Analysis of Stable Isotopes
For the carbon (δ 13 C) and oxygen (δ 18 O) stable isotope analysis, water samples (n = 37) were collected in 100 mL glass vials and treated with HgCl 2 to prohibit biological activity. The samples were stored in a cool, dark place until measurements were carried out at the isotope laboratory at the Friedrich-Alexander University Erlangen-Nürnberg. Samples were analyzed through a standard procedure with an isotope-ratio-mass spectrometer (Gasbench 2; Thermo Fisher Scientific, Bremen, Germany), corrected to instrumental drift, and normalized to the VPDB (Vienna Pee Dee Belemnite) or to the VSMOW/SLAP (Vienna Standard Mean Ocean Water/Standard Light Antarctic Precipitation) scale, respectively [75]. An acid treatment on a Gasbench was used for carbonate stable isotope analyses. The precision of the control sample was better than 0.1‰ (1 sigma) for δ 13 C and better than 0.05‰ (±1 sigma) for δ 18 O. Oxygen samples were not corrected for the isotope salt effect as this effect has been reported to be neglectable for seawater consisting mainly of NaCl [76,77].

Data Analysis of Hydrography and Flow
Data visualization and CTD and ADCP data conversions were performed with Matlab R2018a. For subsequent data analysis, the raw CTD and ADCP data were converted. CTD variables were converted according to TEOS-10 standard with GSW Oceanographic Toolbox to absolute salinity (S A ), conservative temperature (Θ), potential density (ρ Θ ) and potential density anomaly (σ Θ ), i.e., sigma-theta values [78,79]. The water column stratification was estimated from CTD casts with Brunt-Väisälä, or buoyancy, frequency N 2 = g 2 ρ Θ −1 ∆ρ Θ ∆z −1 [79], where g is the gravitational acceleration and z is the depth. The flow measurements were corrected for the local magnetic declination based on International Geomagnetic Reference Field, IGRF-11 model data [80]. To estimate the mean flow direction, the horizontal velocity components (eastward, u e , and northward, u n ) were rotated using variance ellipses from the jlab data analysis toolbox [81]. Accordingly, the mean direction velocity components (u m ) are in the direction of the most energetic fluctuations, while components perpendicular are interpreted as cross-flow (u c ). The vertical velocity component, u w , was not altered. The tidal frequencies, ω, and their amplitudes, a, were analyzed with the harmonic analysis toolbox T_Tide [82]. The tidal signals were analyzed by using bottom pressure and horizontal velocity fields at 3 (SLM) and 5 (MLM) meters above the sea floor. Only signals with a signal-to-noise ratio >2 were considered to be significant.
The health statuses of the five included reefs were estimated using the known spatial extent of the reefs combined with data on density and vertical height of living coral colonies [15,22]. The flow state based on topography-flow interaction at the Nakken bank reef was determined following methods described in [22].

Cold-Water Coral Occurrences
The shallowest reefs in the study area are HH and SN, where framework-building scleractinian corals L. pertusa and Madrepora oculata are observed at depths of 100-200 m and 80-220 m, respectively. At Straumsneset, the densest coral cover is recorded between 130 and 180 m depth. The maximum continuous L. pertusa cover (length < 6.5 m, width < 3.2 m, colony height < 1.2 m) is larger than maximum continuous M. oculata cover (length < 2.5 m, width < 0.2 m, colony height < 0.5 m, with 0.2-0.3 m dead coral skeleton). The deepest L. pertusa colonies are found at HN, where they live at depths between 220 and 240 m. At NK and BV bank reefs, scattered L. pertusa patches live at 190-220 m and 200-210 m depths, respectively. At Nakken, patches have heights of 1-2 m and diameters < 4 m. The bank is~200 m wide. All sites have rich communities of associated megafauna, including the bivalve Acesta excavata, sponges Geodia sp. and Mycale lingua, and octocorals Paragorgea arborea, Primnoa resedaeformis, Paramuricea placomus, and Anthothela grandiflora. Based on coral coverage, spatial extent, and the proportion of living to dead corals, Langenuen CWC reefs are all in health category II [22], with deeper bank reefs NK and BV and wall reef HN having more coral rubble and dead coral framework than shallow wall reefs HH and SN.

Water Column Hydrography
Waters in Langenuen consisted of seasonally forced surface waters down to~100 m. The Norwegian coastal water (NCW) and the North Atlantic water (NAW, S A > 35 g kg −1 ) were present beneath the surface waters ( Figure 2). The temperature at water depths above the CWC reefs (<80 m) ranged between 5.  Table 3).  . Measured values include temperature, salinity, sigma-theta, buoyancy, oxygen, total alkalinity, dissolved organic carbon, nitrate, phosphate, silicate, stable isotopes, monthly mean speed, monthly maximum speed, southward direction, and monthly mean flow direction. Also shown are calculated values for in situ pH T (total scale), pCO 2, and saturation states of aragonite and calcite. These parameters were calculated using CO2SYS [66].     Table 3). In general, the northern sites were saltier than the southern sites indicating the NAW route from Korsfjord and fresh water output from Hardangerfjord. The surface layer (<80 m) was well stratified (N > 7 ×

Bottom Water Hydrography
The annual bottom temperature at Nakken bank reef ranged from 7.4 to 8.3 • C, with a mean temperature of 7.8 • C. This included two distinct phases: cool temperatures (<7.8 • C) observed from May to mid-November and warm temperatures (>7.8 • C) measured from late November to April (Figure 3), indicating heat dissipation from the surface during autumn and water column mixing in spring. March was the warmest month. September and October were the coolest months (Figures 3 and 4). Bottom water salinity varied between 35.00 and 35.34 g kg −1 with the lowest salinities (<35.2 g kg −1 ) observed in autumn and winter, and the highest salinities observed in late spring, indicating the water source variability between North Atlantic water (S A > 35 g kg −1 )-dominated and Norwegian coastal water-dominated waters. The salinities measured by MLM were~0.05 g kg −1 higher than those measured by SLM (Figure 3 (Table 3). Both the lowest (C T = 2135 µmol kg −1 ) and the highest (C T = 2192 µmol kg −1 ) C T values at CWC living depths were measured at the NK bank reef at 200 m depth during the 16 h sampling period in January 2017. The highest A T value (2343 µmol kg −1 ) was also measured during the same period and place. The lowest A T of 2300 µmol kg −1 was measured at HH wall reef at 200 m depth in August 2017. In general, the northernmost sites (BV and SN) had higher A T in autumn and higher C T in winter and autumn at 200 m depth than sites farther north, but short-term variability recorded over tidal cycle at a single location (HH or NK) was larger than this between sites difference. pH T ranged from 7.94 to 8.08 throughout the water column over the study period ( Figure 5, Table 3). Low values (<7.98) were measured at depths >140 m at BV bank reef and at depths >100 m at NK bank reef in winter (January 2017 and February 2016), at 80 m depth at wall reefs HH and SN in May 2016, at~200 m depth at HH and NK in April 2017 and at depths >100 m at all sites except BV in autumn with a minimum at 150 m depth. High values of >8.03 were measured at the surface (<40 m) at all sites and times (Table 3) Table 3). Carbonate chemistry sampling was performed over several hours at two stations at two different times: At wall reef HH (corals at depths 80-220) in October 2016 and at bank reef NK (corals at depths 190-220) in January 2017. During a 26 h sampling period at 200 m depth at the HH wall reef in October 2016, Θ and S A were relatively stable with ranges of ∆Θ = 0.14 • C and ∆S A = 0.11 g kg −1 (Figure 6). Over the sampling period, the carbonate system parameters changed by: ∆A T = 10 µmol kg −1 , ∆C T = 10 µmol kg −1 , ∆pH T = 0.032, ∆pCO 2 = 45 µatm, ∆Ω Ar = 0.12, and ∆Ω Ca = 0.19. This is comparable to the seasonal variability between winter and summer from single measurements at HH at this depth ∆A T = 11 µmol kg −1 , ∆C T = 34 µmol kg −1 , ∆pH T = 0.112, ∆pCO 2 = 123.6 µatm and ∆Ω Ar = 0.4 ( Figure 6). The flow direction at nearby Nakken reef was southward with speeds between 4 and 43 cm s −1 .  (Figure 6). The carbonate chemistry followed the changes in flow direction with high C T and low A T values observed with northward flow (=warm and fresh phase). The highest A T value (2343 µmol kg −1 ) was also measured during northward flow conditions. Over the sampling period, the carbonate system parameters showed large ranges of: ∆A T = 26 µmol kg −1 , ∆C T = 57 µmol kg −1 , ∆pH T = 0.11, ∆pCO 2 = 118 µatm, ∆Ω Ar = 0.41, and ∆Ω Ca = 0.64. This is larger than the seasonal variability between spring and summer from single measurements at NK at this depth ∆A T = 10 µmol kg −1 , ∆C T = 41 µmol kg −1 , ∆pH T = 0.04, ∆pCO 2 = 47.6 µatm, and ∆Ω Ar = 0.14 ( Figure 6).  (Table 3).

Stable Isotopes
The distribution of δ 13 C within the water column varied between the different sampling times (Figure 7). In January 2017, δ 13 C decreased with depth. In April 2017, the water column was well mixed with respect to δ 13 C, and in May 2016, δ 13 C increased with depth. The δ 13 C value was generally higher at bank reefs (0.41‰ ± 0.06‰, depths 190-220 m) than at shallower wall reefs (0.37‰ ± 0.07‰, depths 80-220 m) ( Table 3). δ 18 O increased with depth and had similar ranges for all sampling times. Therefore, δ 18 O was higher at bank reefs (0.39‰ ± 0.03‰) than at shallower wall reefs (0.34‰ ± 0.06‰) ( Table 3). In May 2016, when samples were taken from most of the sites, δ 18 O had the largest range at~150 m depth of ∆δ 18 O = 0.12 ‰ (Figure 7). With a relatively low sample size (n = 37), these results are preliminary and only provide us the first look at the distribution of stable isotopes in Langenuen. Thus, these results have to be confirmed by future research.

Flow Regime
The flow below 180 m was strong (mean 18 cm s −1 , max 60 cm s −1 ) with a semidiurnal tidal component (Figures 4 and 9, Table 3). The monthly mean flow speed (U mean ) varied between 12 and 25 cm s −1 . Low mean flow speeds (U mean < 15 cm s −1 ) were measured in June and July, and high mean flow speeds (U mean > 20 cm s −1 ) were measured in March and in late autumn in between October and December. During the remaining months, U mean varied between 16 and 18 cm s −1 . At depths above 180 m (recorded with MLM), flow speeds varied more. From May to August, U mean was < 15 cm s −1 in between 115 and 180 m depth. U mean reached maxima at 90 m and at <50 mm depth with 27 and >45 cm s −1 , respectively ( Figure 9A).
The maximum monthly flow speeds (U max ) below 180 m depth varied between 30 and 70 cm s −1 . U max was <45 cm s −1 in May and June and~60 cm s −1 between October and December. During the other months, U max varied between 50 and 55 cm s −1 . Between 125 and 180 m water depth, U max was <50 cm s −1 . At depths of <117 m, U max was >100 cm s −1 except in August, with peak speeds (>200 cm s −1 ) measured around 60 and 100 m depths ( Figure 9B). The current direction was determined by the orientation of the fjord and was dominantly southward (between 90 • and 270 • ) beneath 180 m, varying in between 80 and 180 m, and southward (May and June) or northward (July and August) near the surface ( Figure 9C, Table 3). In July and December, the flow direction was dominantly northward ( Figure 9C). The direction of the most energetic flow, u m , was dominantly southward across the water column (150 • -163 • , at depths >180-208 m and 160 • -170 • at depths <180 m). The northward flow was measured at~150 m depth in May and depths <100 m in August ( Figure 9D).
The flow regime at NK bank reef was mostly subcritical, but it supported hydraulic jumps with flow velocities >U mean . These are observed as rapid changes in hydrographic parameters. For example, in the winter months, temperature, salinity, and turbidity fluctuated within 12 h ∆Θ = 0.4 • C, ∆S A = 0.15 g kg −1 , and ∆turbidity >0.2 NTU. Following the categorization of Juva et al. (2020) [22], the flow state is thus partially subcritical (PB sc ).

Tides
Tidal analysis of the pressure data explained 87.2% and 95.6% of the pressure fluctuations with 28 and 22 significant constituents at SLM and MLM, respectively. At both lander sites, the semidiurnal (M 2 ) signal generated the largest amplitude of 0.34 dbar. The other significant constituents were diurnal, semidiurnal, and higher frequency constituents (Table 4). Table 4. Tidal analysis for bottom pressure and flow record based on the harmonic analysis toolbox T_Tide [82]. Shown are the harmonics with amplitudes >0.02 dbar and >1.5 cm s −1 for pressure and flow, respectively, in three categories: diurnal, semidiurnal, and short period constituents. Explained variance through the tidal model in percent is given next to the parameter. Abbreviations: p: pressure, u e u n : bottom horizontal flow.  (Table 4).

Discussion
In this study, we investigated the dynamics of the flow and environmental (hydrography, carbonate chemistry, and inorganic nutrients) conditions around three L. pertusadominated wall reefs and two bank reefs in a narrow fjord on diurnal, seasonal, and annual time scales. Our results suggest that both hydrodynamics and hydro-biogeochemistry regulate the distribution and health of the reefs in Langenuen and that the fjord is likely to be in a state of change that has been ongoing since the 1980s.

Flow Dynamics
All five CWC reefs are healthy, but their limited spatial extent suggests that areas with suitable living conditions for CWCs are narrow both vertically and horizontally. CWC occurrences, both on the walls and banks coincide with the most dynamical parts of the fjord.
The flow regime in Langenuen is controlled by the Norwegian coastal current and modified by bathymetry, wind conditions, and tidal forcing [83][84][85][86]. In the upper 100 m, the flow is strong with peak speeds of >100 cm s −1 . At L. pertusa living depths (100-220 m), water flow is predominantly southward with mean speeds of <20 cm s −1 and peak speeds of 60 cm s −1 (Figure 9). This is comparable to other CWC sites in the NE Atlantic [27,31,87]. The flow down to~120 m is driven by southerly winds and Norwegian coastal water entering the fjord above the sill depth from the south [88,89]. Flow in deeper water layers is driven by seasonal density differences outside the fjord system. For example, the flow is reversed to northward-dominated at 120-200 m depth in December and in January ( Figure 9C), when prevailing northerly winds ( Figure 4D) [90,91] create coastal upwelling that pushes warm and deep North Atlantic water over the sill and into the deep fjord basin [92]. Intrusions of high salinity and relatively warm North Atlantic water renew the basin water in the fjords [89,93] and maintain aerobic conditions at the bottom [94]. At Nakken, the tidal flow was small compared to the residual flow (~5 vs.~20 cm s −1 ). This is common in these coastal waters [95,96]. Similar amplitudes have been recorded in Hardangerfjord, the main fjord to which Langenuen is a tributary [93].
In the Langenuen Fjord system, the distribution of CWC reefs, both on walls and banks, seems to be governed by small-scale hydrodynamics. At the wall reefs of HH and SN, L. pertusa corals live just beneath the seasonal thermocline and the fastest flow layer (>100 m). The upper vertical limit of the corals could be set by this fast flow layer as in strong currents, the polyps could bend backward, reducing the feeding surface [97], and prey could escape from the polyps [98], restricting energy uptake in the coral. The 100 m depth also coincides with the strongly stratified layer in late summer and autumn (N > 20 × 10 −3 s −1 , Figure 2), which could reduce the zooplankton migration to CWC living depths [99,100]. Below 100 m, flow is slower, and the framework of the CWCs further reduces flow speeds due to friction [28] toward the efficient prey capture speeds of the L. pertusa [34,98]. Moreover, the CWC framework acts as a natural sediment trap allowing the corals to capture and use the enhanced particle delivery. At these depths, thriving colonies of L. pertusa and M. oculata occur on the vertical walls. The lower vertical distribution limit of the CWC growth on vertical walls is created by the lack of hard bottom substrate when moving beneath 200 m depth at Straumsneset and Huglhammaren and below 240 m depth at Hornaneset.
At the bank reefs, the topography-flow interaction is suggested to directly influence the health and growth of CWCs [22]. At the Nakken bank reef, the flow supports periodic hydraulic jumps at high flow speeds. This will create a link between the surface and the deep reefs. During these turbulent events, resuspended organic particles that have settled to the sea bed would locally elevate food supply to the reefs compared to the surrounding deep sea-bed. These processes could be particularly important during periods of food limitation [101]. However, during the periods of periodic hydraulic jumps also mineral particles will be resuspended and settled. It is assumed that patchy reefs with similar flow regimes would have high vertical growth rates to prevent burial [22]. Maier et al. (2020) [101] estimated the linear growth at NK to be 13 mm year −1 for new polyps. This is 30% to 100% higher than the linear growth measured for other Norwegian reefs located farther north [102,103]. These areas occur likely in a more dynamic flow state preventing the settling of particles and keeping the reef clear from heavy sedimentation. This would reduce the need for large vertical growth while increasing the need for horizontal growth and bridging between polyps, enforcing the coral skeleton to withstand high physical forcing without breaking [22]. At the Nakken bank reef, turbulent conditions created by hydraulic jumps are only present in winter, i.e., from October to December (U mean >20 cm s −1 , and U max >50 cm s −1 , Figure 9). It is plausible that the emergence and growth of new polyps, taking place at this particular reef from December to March [101], is initiated by the increased sedimentation caused by the periodic hydraulic jumps. Growth rates 60% higher than those measured at NK have been documented for L. pertusa in the Gulf of Mexico, where corals form similar patches as observed in NK [104].
Due to surface warming in spring and summer, the seasonal isopycnal reaches~100 m. The uppermost CWCs on the wall reef setting (at 80-100 m, HH) experience temperatures >12 • C in late summer before the water is mixed in late autumn. This is~4 • C warmer than the mean temperature at these shallow reefs and likely enhances the metabolism of L. pertusa and increases their energetic demand [36,41]. Together with strong flow speeds (limiting prey capture rates in L. pertusa), seasonal high temperatures could restrict the upper limit of the CWCs on the fjord wall. Beneath the seasonal warming layer of 100 m, annual ranges of hydrographic variables were small (∆Θ = 1.92 • C, ∆S A = 0.96 g kg −1 ) particularly when compared to the largest measured temperature fluctuation at any CWC site, namely the~9 • C (5.8-15.2 • C) temperature fluctuation within a day registered in the Cape Lookout area, NW Atlantic [105]. Bottom temperatures decrease in late winter due to dissipation from the surface and remain low throughout summer due to the pronounced water column stratification established by the spring freshwater flood [106]. Due to relatively warm winter temperatures in the deeper layer and the freshwater influence, reefs occur in less dense waters than suggested for healthy CWC sites in NE Atlantic (i.e., σ Θ = 27.35-27.65 kg m −3 , [12]).
At Langenuen, the dissolved inorganic carbon was at times high, and a maximum value of 2192 µmol kg −1 was observed. This is higher compared to what is previously linked to thriving NE Atlantic CWC occurrences (C T < 2170 µmol kg −1 , [15]). On the other hand, total alkalinity (2300-2343 µmol kg −1 ) was similar to other NE Atlantic CWC sites (2287-2377 µmol kg −1 , Table 5). For comparison, the typical carbonate chemistry properties in Atlantic core water in the Norwegian Sea is about 2160 µmol kg −1 and 2310 µmol kg −1 for C T and A T, respectively [107]. Regional studies of carbonate chemistry surrounding CWC ecosystems in the Gulf of Mexico [14], Gulf of Cadiz [15], Mauretania [15], Mediterranean [10,15], and Marmara Sea [10] have shown that C T > 2170 µmol kg −1 is common for CWC sites outside the NE Atlantic (Table 5) while significantly higher (A T > 2500 µmol kg −1 ) alkalinity levels are measured only at the Mediterranean and the Marmara Sea CWC sites. Together with regional hydrographic (Θ, S A ) conditions, the relatively high A T result that the aragonite saturation is also relatively high (Ω Ar > 2.5) in the Mediterranean CWC sites compared to other basins, where values <2.0 are common (Table 5). Consequently, this study emphasizes the importance of estimating both C T and A T since it is the relationship between them that determines the CaCO 3 saturation. These regional ranges are obtained from single time point measurements that do not include the variability on the diurnal level. References for nutrient data, nd: no data reported, * Only mean salinity reported in [108]. The short-term variability in carbonate chemistry in Langenuen is large, and the highest values are linked to the tidal cycle. Similarly, high C T concentrations have been observed in dynamical NE Atlantic CWC mounds in the southern Rockall Bank [53]. There, values up to 2186 µmol kg −1 were measured over a tidal cycle, indicating that L. pertusa is resilient to C T levels exceeding 2170 µmol kg −1 , at least over short time scales. At Rockall Bank, the tidal range of C T was 58 µmol kg −1 . This is similar to short-term (within 16-h) variability measured at NK reef (∆C T ≈ 57 µmol kg −1 ) in January 2017 and to the annual ranges of C T at CWC living depths at other Langenuen reefs (∆C T ≈ 51 µmol kg −1 at HH, and ∆C T ≈ 39 µmol kg −1 at SN). The short-term changes in carbonate chemistry at Langenuen reefs cannot be explained by salinity and temperature changes alone. For example, the pCO 2 change is about 4.2% per 1 • C, implying that the temperature effect on pCO 2 could explain~5 µatm of the 124 µatm difference recorded at Nakken reef during 16 h in January 2017. This large variability is most likely caused by rapid change in the dominant water mass from Norwegian coastal water to North Atlantic water. NAW has high pCO 2 , thus large C T , but also relatively high A T . Therefore, pH T and Ω Ar are generally higher in NAW than in waters containing freshwater (such as NCW), and large changes in the whole carbonate chemistry are possible during short time periods. It is also plausible that the presence of North Atlantic water with higher Ω Ar compared to Norwegian coastal water is buffering ocean acidification in Langenuen, and strengthening of the coastal stratification outside the fjord system caused by warming could reduce its presence at the Langenuen coral reefs in the future.

Location
Observed differences between the five reefs were not consistent between the different time periods or seasons sampled. This variability is at least partly a consequence of the short-term variability that is not captured with single time point measurements or replicates taken close to each other in time. The observed variability over the 16 h sampling period in January 2017 was larger both between the sites and at NK compared to other months. The high fluctuation in winter could be caused by a combination of the storm event and dynamical winter conditions (i.e., weaker stratification, stronger flow, water source variation between the north (Korsfjorden) and south (~Utsira) more than during other seasons), see Figures 2,4,6 and 9. This kind of extreme event is likely to occur each year, and based on meteorological reports [90,91], it is likely that even more energetic events occurred in winter 2016-2017. Since the tidal component causes large variability in the carbonate chemistry, the full seasonal and diurnal variability of the carbonate and nutrient chemistry should be measured in further ocean acidification studies [48].

Long-Term Changes
The measured time period (from February 2016 to August 2017) is not long enough to fully capture the interannual variability nor the long-term trends of environmental conditions occurring in the Langenuen Fjord. However, because of the water exchange between Langenuen and the open ocean above sill depth, the mean hydrographic conditions at Langenuen CWC depths can be correlated to the fixed coastal station Utsira south of Langenuen ( Figures 1A and 10) [110]. Saetre et al. (2003) [38] reported that both temperature and salinity decreased between 1950 and 1989 along the Norwegian coast. At Utsira, the winter temperature (JFM) decreased from 7.6 to 7.0 • C and salinity from 35 to 34.8 [38] at 150 m depth during this period. These trends are affected by basin-wide phenomena such as "great salinity anomalies" (GSAs) [111][112][113] and the North Atlantic oscillation (NAO) [114]. GSAs are low-salinity/low-temperature events that are formed by cold winters, freshwater outflow, strong northerly winds, and a large sea ice extent in the northwestern Atlantic [112]. A positive NAO index is associated with mild winters, an increase in westerly winds, higher winter precipitation over Scandinavia, and a deeper Norwegian coastal current. GSA salinity minima are reported at Utsira coastal station in 1977-1978, 1987-1989, and 1994-1996 [38]. High temperatures and salinities in between these periods are caused by an increase in the Atlantic inflow combined with the atmospheric conditions associated with periods of positive NAO and a general rise in temperature over the past decades [115]. The decadal forcing acts together with climate change to determine the past, the present, and the future deep-water hydrography within the region and at Langenuen. The whole water column had warmed after the GSA in the 1980s when only salinity returned to previous levels ( Figure 10), indicating ocean warming. At Utsira, the increase has been 0.45 and 0.25 • C per decade at 50 and 200 m depths, respectively, since 1975 ( Figure 10). If we assume a similar warming trend in bottom temperature at Langenuen, the waters at CWC living depths were~1 • C cooler in the 1970s compared to the study period 2016-2017. Furthermore, if these warming rates continue at Langenuen, temperatures could seasonally increase to >18 • C at 50 m depth and >10 • C at 200 m depth ( Figure 10). The summer thermocline with temperatures >12 • C would reach depths >100 m by the year 2100. The warming within the past 40 years has likely already increased the energy demand of CWCs compared to that in the 1970s, with possible effects on coral energy reserves and reproductive output if these are not met by increased food uptake rates [41]. The warming of coastal waters caused by climate change has likely led to a decreased frequency of intrusions of dense, oxygen-rich North Atlantic water with relatively high Ω Ar levels and has caused a general oxygen decline in the basin waters of some western Norwegian fjords with shallow sills (i.e., <100 m) [45,116]. Furthermore, warmer water contains less oxygen. In the Masfjorden, north of Bergen, this multidecadal decline corresponds to a loss of 2.0 mL L −1 over 42 years and an associated 1 • C rise in temperature [45]. A similar decline in oxygen has been reported in Byfjorden, off the city of Bergen, with an accompanying shift in benthic communities toward domination of opportunistic benthic species [116]. Given the observed warming in Utsira, oxygen concentrations have likely decreased in Langenuen as well and are likely to decrease even further in the future. Hebbeln et al. (2020) [35] suggested that L. pertusa populations are highly sensitive to low oxygen conditions of 40%-50% lower than the ambient oxygen values. From observed conditions, a decline of 40% from present values would set the lower oxygen limit in Langenuen to~2.5 mL L −1 compared to the global limit of <1.5 mL L −1 [36,117,118]. If a decline of 0.5 mL L −1 per decade continued linearly in Langenuen, corals would be exposed to hypoxic conditions by the end of the century and to conditions <2.5 mL L −1 by the year 2070.
Besides warming waters and oxygen decline, salinity in the surface layer has been decreasing along the Norwegian coast [38]. At Utsira, the decline at 50 m depth has been up to −0.05 g kg −1 per decade between May and September since 1975 ( Figure 10), but salinity has increased slightly during other months and at other depths. This longterm decrease in salinity at the surface is partly caused by increased precipitation and retreat/melting of glaciers, substantially increasing freshwater run-off. Førland  Driven by warming, the water column has become less dense within the past 40 years in Utsira at a rate of 0.037 kg m −3 per decade at 50 m and~0.02 kg m −3 per decade at 200 m depth. If density has changed similarly in Langenuen, waters at CWC living depths would have been within the suggested sigma-theta range for thriving CWC sites in the NE Atlantic (>27.35 kg m −3 ) [12] during the 1970s but are lower than that now. Such changes may be critical because the depth zonation of CWCs on the vertical walls may be related to physical boundary conditions at specific depths that act to concentrate food particles [122], and corals cannot physically move to adjust to that.
Freshwater from rivers, rain, and melting glaciers have been observed to decrease A T and Ω Ar substantially and play a large role in increasing ocean acidification in surface waters [123,124]. These waters could reach the coral depths during autumn and winter mixing even though the summer stratification is likely getting stronger. Therefore, there is a weak positive indication that the influence of fresh coastal water into the deeper water masses of the fjord may locally accelerate ocean acidification at CWC living depths.
Our hindcast suggests that ocean warming and acidification may already today affect the functioning of Langenuen CWCs. As oceanic uptake of atmospheric CO 2 continues [115], warming and ocean acidification continue in the western Norwegian fjords. We, therefore, stipulate that these shallow fjord reefs could serve as windows into the future and forewarn about the likely effects of ocean acidification and warming on CWC reef biodiversity and productivity. Studies from tropical coral sites suggest that the rate of calcification is lower under recent conditions (~400 ppm) compared to pre-industrial (280 ppm) times [125,126], and enhanced reef growth has been observed in situ in lagoons where pCO 2 levels have been manipulated to pre-industrial levels. McGrath et al. (2012) [127] showed that C T has increased in subsurface waters in the NE Atlantic between 1991 and 2010. This was concomitant with a decrease in Ω Ar and a shoaling of the ASH. Within this century, 70% of the known CWC habitats are predicted to be in corrosive waters [49]. The ocean acidification monitoring program of Norwegian waters [48] has shown an annual decrease in Ω Ar and pH of 0.007 and 0.02 in Korsfjorden at depths of~670 m between 2007 and 2017, respectively. The decrease in Ω Ar and pH has been reported to be greater in coastal areas than in deeper waters (~2000 m) offshore [48]. If the carbonate system changed at similar rates at~200 m depth in Langenuen, CWCs would be exposed to corrosive conditions (Ω Ar < 1) by 2090. Given their shallower location in a narrow fjord, it is likely that the rate of change is even larger, and Ω Ar < 1 conditions would occur even sooner [51]. The living coral can withstand some corrosive conditions with pH up-regulation [10,128], and in the Mediterranean, Ω Ar < 0.92 was found to be the calcifying limit for L. pertusa [129]. However, the exposed dead coral skeleton, which frequently forms the largest portion of the CWC reef, starts to dissolve in corrosive conditions [130].

Conclusions
The distribution of cold-water corals in the narrow Langenuen Fjord in southwestern Norway (health category II) is limited by physical environmental boundaries and the hydrodynamical setting driven by seasonal and short-term forcings. Summer thermocline with temperatures of over 12 • C reached the uppermost populations on the vertical wall reefs (~80 m). This is >4 • C warmer than the mean temperature (7.8 • C) at the coral living depths and may be over their tolerance limit, thus limiting the vertical extent of coral growth. Variability of chemical parameters during a recorded winter storm (∆A T = 26 µmol kg −1 , ∆C T = 57 µmol kg −1 ) was comparable to the measured annual variability (∆A T = 42 µmol kg −1 , ∆C T = 57 µmol kg −1 ). This short-term variability was driven by rapid changes in flow conditions and water masses. For all reef settings, but in particular the bank reefs, our findings highlight the importance of sampling at different phases within the tidal cycle when collecting samples for environmental monitoring programs.
Norwegian coastal waters outside Langenuen have warmed since the 1970s on average at a rate of 0.25 • C per decade at depths >100 m and up to 0.44 • C per decade at 50 m depth. Waters have warmed more during summer and autumn compared to the remaining seasons. This warming has reduced water mass density and its oxygen concentration and strengthened summer and autumn stratification, changing the physical boundaries toward "non-optimal" for coral growth in Langenuen. The depth zonation of CWCs on the vertical walls may be related to physical boundary conditions at specific depths that act to concentrate food particles, and corals cannot physically move to adjust to that. Clearly, this aspect needs more attention and should be tackled in future research.
If warming continues at a similar rate in the future, the summer thermocline with temperatures >12 • C would reach depths >100 m, and corals would be exposed to hypoxic and corrosive events by 2100. In conclusion, more frequent short-term stress events and the gradual change of mean conditions may restrict the depth zonation of corals and change the benthic species composition in Langenuen, as well as other CWC sites, dramatically.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.