Comau Fjord is located in the northernmost part of the Chilean fjord region, east of Chiloé island (42°10′ to 42°50′ N and 72°40′ to 72°60′ W) (Figure 1A). It has a total length of 45 km and reaches a maximum width of 8.5 km (Häussermann et al., 2012). As a typical fjord in northern Patagonia (Silva, 2008), Comau Fjord is characterized by a strong vertical stratification, as it receives freshwater from precipitation (approx. 6,000 mm yr^–1) and rivers, mainly in winter and early spring (September–October), which creates a low salinity surface layer (0.5 m in summer—10 m in winter) with salinities as low as 2 (PSU) (Pantoja et al., 2011; Addamo et al., 2021). Below 18 m water depth, however, salinity is relatively constant at 32 (PSU) (Addamo et al., 2021), and the intermediate water layer, consisting of Modified Sub-Antarctic Water (MSAAW), compensates the freshwater outflow (Palma and Silva, 2004; Valle-Levinson et al., 2007). Due to the high tidal amplitudes in the region (Schwabe et al., 2006) organisms in the upper water layer (down to approx. 15–20 m) of the fjord are exposed to pronounced fluctuations in salinity and temperature (Häussermann and Försterra, 2009; Jantzen et al., 2013b). Below the pycnocline, monthly mean water temperatures fluctuate seasonally between 8 and 12°C (Häussermann et al., 2012; Jantzen et al., 2013b).

The experiment was conducted at two sites, (1) “XHuinay,” a site at the western coast of the central fjord area (42°23′25″ S, 72°27′32″ W) and at (2) “’Lilihuapi” (42°09′43″ S, 72°35′27″ W), located at the mouth of the fjord (Figure 1A). Sampling sites were chosen based on the prevailing horizontal gradient in seawater carbonate chemistry (e.g., pH, total alkalinity (TA), and Ω_arag) between the end and the mouth of the fjord (Jantzen et al., 2013b). On the basis of previous assessments of the environmental conditions at the selected experimental sites (see Jantzen et al., 2013b), it was expected that the water at XHuinay would display a lower pH, TA, and Ω_arag than at Lilihuapi.

In February 2014, vertical profiles of pH and salinity were taken at six stations from the end of the fjord to its mouth (i.e., Station 1, Station 2, XHuinay, Station 4, Station 5, and Lilihuapi; Figure 1A), using a CTD multi-probe (SBE 19plus-SEACAT Profiler - Conductivity, Temperature, and Pressure with RS 232 Interface, Sea-Bird Electronics Inc.) equipped with a pH sensor (PHC301, Hach Lange GmbH, Germany). All casts were conducted by hand, and with maximum depths between 83 and 216 m; only the relevant averages for water depths between 17 and 23 m are reported here, as the experimental corals were located in a water depth of 20 m (for the full dataset see Laudien et al., 2014).

Additionally, seawater samples were taken between January and February 2015, at a water depth of 20 m at the two experimental sites (n = 4 per site), using a 2.5 L Niskin-type water sampler (Hydrobios GmbH, Altenholz, Germany). The pH was immediately determined manually with a portable pH probe (PHC301, Hach Lange Company, Düsseldorf, Germany). To further assess the carbonate chemistry at the experimental sites, seawater samples (n = 4 per site) were collected on the same day at a depth of 20 m, using SCUBA. For the sampling, a 500 mL borosilicate bottle (Schott Duran, Germany) was filled with seawater at a depth of 20 m, leaving no air space. Before these seawater samples were analyzed for total alkalinity (TA; i.e., the total concentration of bases and total dissolved solids), they were filtered through glass microfiber filters (Whatman GF/F 25 mm, GE Healthcare Europe GmbH, Freiburg, Germany) to remove particulate organic matter. For each sample, triplicate measurements of TA were performed using potentiometric titration (TitroLine alpha plus+TA05plus, SIAnalytics GmbH, Mainz, Germany and TitriSoft 2.72, Schott Instruments, Mainz, Germany) (Gran, 1952) using certified seawater as reference material (CRM) (Andrew Dickson, Scripps Institution of Oceanography). Further parameters of the carbonate system, i.e., dissolved inorganic carbon (DIC), total alkalinity (TA), partial pressure of carbon dioxide (_pCO₂), aqueous carbon dioxide [CO₂(aq)], as well as aragonite and calcite saturation (Ω_arag and Ω_calc, respectively), were then calculated with the R package Seacarb (Lavigne and Gattuso, 2013) using first and second carbonate system dissociation constants of Lueker et al. (2000), as well as the dissociations of HF and HSO₄^–(Dickson and Goyet, 1994), respectively.

In addition, ambient seawater temperature was recorded at the two experimental sites and at a depth of 20 m throughout the experimental period (1 year, from February 2014 till January 2015) using autonomous data loggers (TidbT V2 Temp Logger, Onset Computer Corporation, United States) with a 15-min recording interval.

In February 2014, 20 specimens of T. endesa (Figure 1B) with an average height of 4.1 ± 0.9 cm (mean ± SD) were collected by hand in a depth of 20 m at each of the two experimental sites (total = 40) by SCUBA divers. To minimize the risk of damage and stress during the sampling process, mainly corals growing on a biogenic substrate (primarily on shells of alive limpets Crepidula sp.) were collected together with their substrate. For the transport to the laboratory, specimens were placed in plastic containers, sealed water-tight at the collection site, ensuring no contact with either the brackish surface water layer or air. Experimental specimens were maintained for 48 h in a 30 liter flow-through aquarium, supplied with unfiltered seawater, continuously pumped up from 25 m water depth. Corals were then cut from their settling substrate, using a grinding disc attached to a rotary tool (Dremel 4,000, Dremel, The Netherlands), and any additional organic material adhered to the base of coral specimens was removed. Individual corals were then fixed (Super Flex Glue Gel, UHU GmbH und Co KG, Germany) to polyethylene (PE) screws and their initial skeletal mass was determined using the buoyant-weighing method (see below). Finally, corals were brought back to the field and attached (facing upwards) on a customized coral holder installed at a depth of 20 m at the respective sampling sites (Figure 1C).

Of the original 20 specimens collected per site (i.e., XHuinay and Lilihuapi), ten were returned to the site of origin, serving as a control referred to as “X” and “L” corals, respectively. The remaining corals were cross-transplanted to the respective other site (Figure 1C) and are now referred to as “X to L” (for corals transplanted from XHuinay to Lilihuapi) and “L to X” (for those transplanted from Lilihuapi to XHuinay). In this way, 20 T. endesa specimens were installed at each site, with ten control and ten cross-transplanted specimens randomly placed on the previously mounted holding racks. This cross-transplantation of corals between the two experimental sites was conducted to eliminate the possibility of incorrectly transferring the growth of two subpopulations, each genetically adapted to its own location, to the population occurring in the entire fjord.

The growth rate of T. endesa (as mass increase in mg per year and normalized for % increase per day) was determined using the buoyant-weighing technique (Jokiel, 1978; Davies, 1989), a method widely used for tropical, temperate, and CWCs (e.g., Marubini et al., 2001; Jantzen et al., 2013a; Comeau et al., 2014; Büscher et al., 2019). The buoyant mass of the corals in water (m_water in grams) was assessed using an analytical balance (CP 225D-OCE, Sartoris, Germany, precision 0.001 g) at the beginning of the experiment in February 2014 and then again at the end in February 2015. The mass of the specimens in air (m_air) was then calculated after Jokiel (1978):

where ρ_water is the seawater density (g cm^–3), and ρ_aragonite is the skeleton density of aragonite (g cm^–3).

To calculate seawater density during the weighing process (Bialek, 1966), temperature and salinity were assessed using a temperature sensor (WTW ama-digit, Wissenschaftlich-Technische Werkstätten GmbH, Germany) and a refractometer (Sinokit Enterprise Limited, Hong Kong). Seawater temperature was kept at 12 ± 0.5°C. The difference between the coral mass in 2014 and 2015 was taken as the skeletal growth per year (mg yr^–1). In order to compare the growth rates of T. endesa with those of other CWC and tropical coral species, growth rates (G), were normalized to mass increase per day, as a percentage of the initial mass of the coral (G % d^–1) using Equation 2:

where m₁ is the initial dry mass (in mg), m₂ is the dry mass at the end of the experiment, and t₂ – t₁ is the duration of the experiment between the final (t₂) and initial (t₁) time in days.

To compare the skeletal mass increase of T. endesa between the four different groups (i.e., “X,” “L,” “X to L,” and “L to X”) and to test for significant differences (p ≤ 0.05) between the in situ skeletal growth at the two experimental sites (i.e., XHuinay and Lilihuapi), pairwise Wilcoxon tests and Kruskal-Wallis tests were used. All statistical analyses were performed using R (Foundation for Statistical Computing, Austria, Version 3.5.3).

In 2014, the seawater pH (averaged for a depth between 17 and 23 m; for the full dataset, see Laudien et al., 2014), assessed at the six sampling sites (i.e., Station 1, Station 2, XHuinay, Station 4, Station 5, and Lilihuapi; Figures 1A, 2A) highlighted the pronounced horizontal gradient from the end of the fjord toward its mouth. The seawater pH ranged from 7.57 ± 0.01 (Station 1; mean ± SD) via 7.63 ± 0.02 at XHuinay, to an average of 7.90 ± 0.01 at Lilihuapi (Figure 2A and Table 1). The additional manual pH measurements from 2015, conducted only at the experimental sites (i.e., XHuinay and Lilihuapi), at a depth of 20 m further underline these particularly low seawater pH values in the region, with 7.70 ± 0.02 at XHuinay and 7.90 ± 0.01 at Lilihuapi (Table 1; for full dataset see Diercks et al., 2015a,b). Low TA also characterized the seawater at the two sites, with 2,234 ± 36 μmol l^–1 at XHuinay, and 2,216 ± 19 μmol l^–1 at Lilihuapi, resulting in an undersaturation (Ω_arag 0.98 ± 0.05) at XHuinay and a low aragonite saturation (Ω_arag1.47 ± 0.03) at Lilihuapi (Table 1).

Between February 2014 and January 2015, the mean water temperature at the central fjord at XHuinay was 11.5 ± 0.5°C, with monthly variations ranging from 10.55 ± 0.15°C in August 2014 (austral winter) to 12.64 ± 0.89°C in March 2014 (austral summer; Figure 2B and Table 1). The highest water temperature of 15.30°C was recorded in December 2014, and the lowest of 9.93°C was logged in August 2014. Maximum monthly temperature fluctuations of Δ0.89°C and Δ0.77°C were recorded in March 2014 and December 2014, respectively. At Lilihuapi, the average annual water temperature was 12.9 ± 0.9°C, with the water coldest in August 2014 (10.41 ± 0.17°C) and warmest in February 2014 (12.85 ± 0.93°C; Figure 2B and Table 1). The highest water temperature at this site (15.53°C) was recorded in February 2014, the lowest in August 2014 (9.46°C). The largest monthly temperature fluctuations of Δ0.93°C were recorded in February and December 2014 at Lilihuapi.

The salinity-driven pycnocline was very pronounced at XHuinay (Figure 2C; for full dataset see Laudien et al., 2014), extending between 5 and 12 m water depth, with an average salinity of 25.3 ± 0.07 in surface waters (<5 m) and 32.5 ± 0.2 in waters below 12 m. At Lilihuapi, the pycnocline was less pronounced but also present between 5 and approximately 12 m water depth, with a salinity of 30.1 ± 0.4 in shallower waters (<5 m) and 32.5 ± 0.1 below 12 m (Figure 2C). Consequently, the salinity at 20 m, the respective depth at which the T. endesa specimens were mounted on the coral holders, was similar at the two experimental sites, with 32.7 ± 0.1 at XHuinay and 32.6 ± 0.1 at Lilihuapi (Table 1).

In January 2015, all experimental specimens were alive and none showed any obvious visible sign of degradation (e.g., tissue damage/loss or bioerosion). Overall, there was no significant difference between the skeletal growth rates of T. endesa specimens at the two experimental sites XHuinay and Lilihuapi (Figure 3 and Supplementary Table 1). Neither between specimens of the control groups (i.e., “X” and “L”), which remained at their site of origin, nor between the cross-transplanted individuals (i.e., “X to L” and “L to X”). Therefore, growth rates of T. endesa from Comau Fjord were averaged for all specimens, indicating an increase of skeletal mass of 46 ± 28 milligrams per year and individaul, or 0.03 ± 0.02 % per day. Out of the 40 corals in the experimental setup, only two transplanted specimens showed a decrease or stagnation in mass over the experimental period, one from the group “X to L” with an average decrease of –0.02% per day, and one specimen of the group “L to X,” where the mass remained almost identical (< –0.002% d^–1) between 2014 and 2015.

Patagonian fjords are known for their low seawater pH, which is assumed to result from the high geological activity and seeps of volcanic origin in the region (Pantoja et al., 2011; Jantzen et al., 2013b; Muñoz et al., 2014). The emerging gradient from the mouth to the end of these fjords is also reflected in the carbonate chemistry of the seawater in Comau Fjord, characterized by an overall comparatively low pH and a horizontal pH gradient with the lowest values at the end of the fjord (Station 1 with 7.57 ± 0.01). In the center (Station XHuinay), a pH of 7.70 ± 0.02 was measured, and the highest pH of 7.90 ± 0.01 at its mouth (Station Lilihuapi). Here, the influence of the Pacific Ocean increases (Pickard, 1971), and the fjord water mixes with that of the Gulf of Ancud. Combined with the low total alkalinity (TA) of the fjord water, the carbon equilibrium shifts toward CO₂, leading to a reduction in the overall aragonite saturation state (Kleypas et al., 1999; Zeebe and Wolf-Gladrow, 2001), with occasional undersaturation of the waters, for example at XHuinay (Ω_arag 0.98 ± 0.05). Seawater that is undersaturated (<1) in aragonite is assumed to be corrosive and thus inhibits calcification as it may promote the dissolution of unprotected aragonite structures (Feely et al., 1988, 2008). However, it has also been suggested that the increased availability of bicarbonate, due to an enhanced pCO₂ may facilitate calcification, although it simultaneously decreases the Ω_arag (Marubini et al., 2008; Jury et al., 2010). Comparing the seawater chemistry at the two experimental sites in 2014 and 2015 with the previous reports from 2010 and 2011 (Jantzen et al., 2013b) also shows that it is subjected to temporal variations. While the seawater pH and salinity at the experimental sites remained stable over these five years, the TA in 2010/11 at both sites were slightly lower than the ones measured in the present study. Salinity mainly influences TA of the ocean and thus the corresponding Cl^– and Na⁺ concentrations, as well as biogeochemical processes such as the production of particulate organic matter by microalgae and the precipitation of calcium carbonate (Wolf-Gladrow et al., 2007). Since salinity did not change during the respective years, the amount of organic matter and the precipitation of calcium carbonate may be the cause of these TA differences. This observation may also be an indication of the dynamics of the abiotic environmental conditions in the region.

Given this, and the fact that the present study only includes the assessment at two time points (i.e., in February 2014 and February 2015), it remains unclear whether the differences in carbonate chemistry at the two experimental sites (i.e., XHuinay and Lilihuapi) persist throughout the year and/or if they are subject to fluctuations. Since the assessed skeletal growth rates of T. endesa at the two experimental sites (for both, cross-transplanted and control corals) were not significantly different, we report below on the general in situ skeletal growth rates of T. endesa in Comau Fjord.

On average, the skeletal mass of T. endesa specimens increased by 46 ± 28 milligrams per year, corresponding to 0.03 ± 0.02% per day. These growth rates are comparable to those reported for other CWC species, such as L. pertusa from the North-Atlantic (0.01 ± 0.01% per day; Büscher et al., 2019), or Dendrophyllia cornigera (0.04 ± 0.02% per day; Orejas et al., 2011).

However, the growth of CWC species can be influenced by various environmental factors, such as temperature (Grigg, 1974; Matsumoto, 2007; Silverman et al., 2007), current speed (Purser et al., 2010), prey abundance (Silverman et al., 2007; Purser et al., 2010; Naumann et al., 2011), and the aragonite saturation status (Gattuso et al., 1998; Silverman et al., 2007; Jury et al., 2010; Form and Riebesell, 2012; Büscher et al., 2019). The relatively harsh environmental conditions of Comau Fjord (with particularly low pH and Ω_arag) may thus challenge the calcium precipitation in T. endesa, resulting in reduced growth rates than would be the case under more calcification-favorable conditions. The variety of growth responses of marine calcifiers to a low seawater pH may thus reflect the actual range and plasticity of acclimatization potentials of some species to maintain their skeletal growth rates, over a range of seawater Ω_arag (McCulloch et al., 2012a,b; Cohen et al., 2017). For instance, some solitary and temperate coral species, such as Caryophyllia smithii (McCulloch et al., 2012a) and D. dianthus (Wall et al., 2016), are suspected being capable of an active physiological adaption to the low pH of ambient seawater by regulating their internal pH through an increased Ca²⁺-ATPase activity (McCulloch et al., 2012a,b). Consequently, these species are capable to regulate the pH in their tissues up to 0.78 units above the pH of the surrounding seawater (McCulloch et al., 2012a,b). As T. endesa shows skeletal growth rates comparable to those of other CWC species from less corrosive seas (e.g., Mediterranean Sea), this species may also be capable of such an up-regulation of the internal pH, when seawater Ω_arag is below 1 as it was the case in January 2015 in XHuinay (Table 1). However, the physiological responses of this species to its environment are still unknown and need future investigation. Such a pH elevation would also be accompanied by an increased energetic expenditure, leading to a reduction of energy for other physiologically important processes. A 0.1 unit decrease in seawater pH may result in a 10% increase in energy budget (McCulloch et al., 2012b). The finding that the majority of the corals in this experiment (95%) sustain net positive calcification rates under the prevalent, harsh environmental conditions thus suggests that such an potentially increased energy demand would be met by the relatively large zooplankton abundances and amounts of particulate and dissolved organic matter that the local fjord systems receive, either from marine or terrestrial primary production (Montero et al., 2011; Sánchez et al., 2011), or anthropogenic sources, such as detritus from terrestrial plants, urban, and aquaculture waste (Rojas and Silva, 2005). However, further research, including extension of the experiments reported here (e.g., over several years), but also on the physiological mechanisms involved, is needed to understand how T. endesa can sustain its growth rates and calcification over long term under the harsh local environmental conditions.

Skeletal growth of tropical and photosymbiotic scleractinians has been extensively studied in the past, both in situ and under a variety of experimental conditions in the laboratory, for example under different light intensities (e.g., Rodolfo-Metalpa et al., 2008; Schutter et al., 2012), feeding regimes (e.g., Houlbrèque et al., 2003; HoulbreÌque et al., 2004), water temperatures (e.g., Carricart-Ganivet, 2004; Rodolfo-Metalpa et al., 2008), and nutrient availabilities (e.g., Ferrier-Pagès et al., 2000; Tanaka et al., 2007). However, data on growth rates of cold-water scleractinians are scarce, mainly because CWC ecosystems have only recently gained attention (Freiwald et al., 2004). This is particularly true for CWC that are part of benthic communities in remote areas, such as the solitary coral species T. endesa from the Chilean fjord region. Like other benthic species in this region (e.g., other CWC species such as C. huinayensis), T. endesa was only recently described (Cairns et al., 2005) and thus information on its ecology and physiology is still extremely scarce. The general lack of data for CWC is also due to the fact that there are only a few locations worldwide where these organisms are relatively easy to access, such as the Patagonian fjord system and some Norwegian fjords. As a consequence, only few studies have assessed long-term growth rates of CWCs (e.g., Brooke and Young, 2009; Orejas et al., 2011; Larcom et al., 2014; Büscher et al., 2019). However, these mainly focused on the most common framework forming cold-water scleractinian (Roberts et al., 2009) L. pertusa, and the solitary CWC species D. dianthus (Hassenrück et al., 2013; Jantzen et al., 2013a; Martínez-Dios et al., 2020), and their growth rates were assessed both, under different laboratory conditions, e.g., different food supply (Büscher et al., 2017) or different pCO₂ levels (Hennige et al., 2014; Büscher et al., 2017), and in situ (e.g., Brooke and Young, 2009; Larcom et al., 2014; Büscher et al., 2019). These studies also used a range of different techniques, such as geochemical analysis of skeletal isotopes (Mikkelsen et al., 1982; Mortensen et al., 1998), visual growth banding patterns (Brooke and Young, 2009), estimates from in situ video recordings (Larcom et al., 2014), or buoyant weighing (Orejas et al., 2011; Jantzen et al., 2013a; Büscher et al., 2019). The results of this study are therefore not only the first scientific report on skeletal growth rates of the CWC T. endesa, but also one of the first in situ assessments of a solitary CWC species in general.

The comparison of skeletal growth rates of T. endesa from this study to previous reports on other heterotrophic cold-water, or temperate coral species (Table 2) reveals that growth rates can vary considerably depending on species, location, and experimental approach (i.e., in situ or laboratory; long-term or short-term measurements). This highlights the importance of further studies on CWC species to gain a better understanding of their ecology, physiology, and response to a changing ocean. For instance, the 1-year in situ skeletal growth of T. endesa (0.03 ± 0.02% d^–1) was comparable to rates previously reported for L. pertusa from the North Atlantic (with a long-term in situ skeletal growth rate of 0.01 ± 0.01% per day; Büscher et al., 2019), and the Mediterranean (long-term rate of 0.02 ± 0.01% per day assessed in laboratory experiment; Orejas et al., 2011), as well as from Dendrophyllia cornigera (long-term growth rate of 0.04 ± 0.02% per day assessed in the laboratory; Orejas et al., 2011). However, the short-term (2 weeks) skeletal growth rates of D. dianthus, also from Comau Fjord, were between three times (when assessed in situ) and eight times (when assessed in a laboratory experiment) higher (Jantzen et al., 2013a) than those of T. endesa. However, it is noteworthy that the reported short-term growth rates of D. dianthus of 0.25 ± 0.18% per day (Jantzen et al., 2013a) is also one of the rare occasions when a heterotrophic CWC species exhibits growth rates comparable to photosymbiotic, tropical conspecifics, such as Pocillopora damicornis (0.2–0.3% d^–1; Piniak and Brown, 2008) or Stylophora pistillata (0.35 ± 0.07% d^–1; Ferrier-Pagès et al., 2000). These comparatively high growth rates, reported by Jantzen et al. (2013a) might also be biased by seasonal effects, as it has been shown that D. dianthus increases its feeding response linearly with prey abundance (Höfer et al., 2018). As the short-term growth rates of D. dianthus were assessed during austral summer, the potential oversupply of food during this season may have led to a temporary peak in skeletal growth, leading to an overestimation of the overall growth rates of this species.

When comparing the skeletal growth rates of different CWC species obtained in different studies, the limitations of the respective experimental approaches must also be taken into account. While insitu measurements are limited in determining all potential influencing factors responsible for coral metabolism and growth rates (e.g., food availability, as well as daily or seasonal variations of various water parameters), it has been shown that in laboratory experiments, the quality of the cultivation of the CWCs may have a significant impact (Jantzen et al., 2013a; Lartaud et al., 2014). Direct comparison of coral growth rates between different studies should thus be considered with some caution, as variations in the cultivation setups, divergent measurement methods, and time frames, as well as other factors (e.g., temperature, feeding regime, or ambient carbonate chemistry), may influence growth.

Although we did not observe any obvious traces of bioeroding organisms (e.g., boring sponges, polychaetes or gastropods) on the experimental specimens, it cannot be excluded that the latter may not have influenced the long-term in situ skeletal growth rates on a micro scale (Försterra et al., 2005), thus potentially leading to an underestimation of growth rates (Jantzen et al., 2013a). The possible colonization by bioeroding organisms also provides another potential explanation for the stagnation, or even decrease in skeletal mass in two of the experimental specimens. Though the T. endesa specimens in this study thrive under the comparatively harsh conditions of Comau Fjord, the exponentially growing economic interest in the fjord region increases anthropogenic pressure. The input of nutrients, sediment due to mariculture and road construction, and the input of pharmaceutical and anti-fouling substances (introduced by the salmon-farming industry) is already a severe threat to vulnerable benthic communities of the Patagonian fjord region (Bustamante, 2009; Försterra et al., 2014, 2016).