In situ growth and bioerosion rates of Lophelia pertusa in a Norwegian fjord and open shelf cold-water coral habitat

Studied reef sites

For a 1 year in situ growth and bioerosion rate assessment, two Norwegian Lophelia reef sites with different environmental characteristics were chosen for collection and re-deployment of live corals and dead erect coral framework. The approximately 13 km long and 700 m wide Sula Reef Complex on the Sula Ridge off the coast of Sør-Trøndelag is the second largest known Lophelia reef on the Norwegian Shelf (Freiwald et al., 2002; Hovland et al., 2005). This offshore location comprises a relatively constant habitat in terms of environmental parameters such as temperature, salinity, pH, and currents, while the selected inshore location, a reef near the island Nord-Leksa in the outer Trondheimsfjord (henceforth referred to as Leksa Reef), is exposed to a highly variable environment due to strong tidal and compensatory currents (Form et al., 2015 Cruise Report POS473). At this location the in situ experiments were placed both in the living area of the reef and in the zone of dead coral debris a few tens of metres downslope.

Collection of Lophelia pertusa and maintenance on board

Sampling of coral specimens of the species L. pertusa was conducted with kind permission of the Norwegian Directorate of Fisheries (Fiskeridirektoratet) under permit number 12/17918. Corals from the Leksa Reef were collected on 29^th and 30^th June 2013 at 63°36.46′N and 9°22.76′E and 157 m water depth (white specimens) and 63°36.43′N and 9°22.45′E and 152 m (orange specimens) during research cruise POS455 with RV POSEIDON (GEOMAR Helmholtz-Zentrum für Ozeanforschung, 2015). At the Sula Reef, corals of both colourmorphs were collected on 4^th July 2013 at 64°06.62′N, 8°07.1′E in 303 m water depth. At both sites, dead erect coral framework, bearing established bioeroder communities (chiefly bioeroding fungi, bacteria, bryozoans, and sponges), was sampled from the reef basis. All samples were collected by means of the manned submersible JAGO with its sensitive claw for non-destructive sampling (GEOMAR Helmholtz-Zentrum für Ozeanforschung, 2017). On board, corals and dead coral framework were placed in large holding tanks (120 × 110 × 80 cm) filled with 500 L natural seawater obtained from ~70 m depth. Four of those holding tanks were connected in order to create a recirculating system. An interconnected cooling unit (Titan 4000; Aqua Medic GmbH, Bissendorf, Germany) kept the water temperature in the tanks at ambient seabed temperature of 7.5–8.5 °C.

Preparation of the corals and dead erect coral framework for re-deployment

Live coral colonies as well as dead erect coral framework were fragmented into fist-sized pieces soon after collection. Afterwards, live corals were stained with Alizarin Red S. For this purpose, live coral branches were placed in separate staining tanks (2 × 30 L plastic containers) mounted within the holding tanks for temperature equilibration. The dye, pre-dissolved in ethanol, was slowly added until a concentration of 10–15 mg L⁻¹ was reached according to the protocol applied in Brooke & Young (2009) for L. pertusa. Live specimens were incubated in the staining tanks for 2–3 days, as cold-water corals incorporate the dye more slowly than faster growing warm-water corals (Brooke & Young, 2009; Form & Riebesell, 2012). The dead framework was examined for calcifying epibionts, which were carefully removed with tweezers and a scalpel in order to eliminate weight gain due to their ongoing calcification during the experiment. All living and dead coral fragments were weighed under water following the buoyant weighing technique described by Davies (1989), employing a Sartorius BP 310 P (Göttingen, Germany; d = 0.001 g) with a purpose built free hanging weighing gondola to enable weighing on board the vessel and to reduce transmission of vibrations onto the balance. Weighing on board was performed at very calm sea, but was nevertheless unsteady. Therefore, the average of 10 consecutive values was taken for each fragment to improve precision. Shortly before deployment of the fragmented and stained corals, one white and one orange coral fragment were attached with cable ties inside a ‘coral cage’ (168 × 178 × 156 mm PP Nalgene^® autoclave baskets). Four such coral cages were prepared for each study site.

Bioeroded coral fragments were attached in smaller baskets (123 × 154 × 105 mm, PP, Nalgene^® autoclave baskets). For each location six ‘bioerosion cages’ filled with dead coral framework material were weighed under water using the buoyant weighing technique (Davies, 1989) and attached as a cluster for facilitating deployment and recovery. In bioeroded dead L. pertusa framework, a number of successive stages of bioerosion, characterised by certain bioerosion trace assemblages, have been identified (Beuck, Freiwald & Taviani, 2010). For the purpose of our experiment, we attempted to distribute skeletons of these different bioerosion stages (stages 3 to 6 sensu Beuck, Freiwald & Taviani, 2010) as evenly as possible to the different replicates and locations, though the amount of bioeroders per fragment can still vary considerably. As in Sula only little and relatively young dead coral framework could be collected, the material in the Sula cages comprised a mixture of dead erect coral framework from both collection sites in Sula and Nord-Leksa. Apart from this, we attempted to distribute skeletons as evenly as possible to the different replicates and locations with regard to their appearance of bioerosion stages.

Deployment and recovery of coral and bioerosion cages

Four coral cages and six bioerosion cages were deployed simultaneously at each of the three deployment locations, two at the inshore reef south of Nord-Leksa and one at the offshore Sula Reef in July 2013. To assure constant submersion in seawater, the cages were immersed in a sampling box installed in front of JAGO filled with seawater before lifting the submersible into water. On the ground, the baskets were positioned at the desired locations one by one with the submersible’s manipulator arm. At the first Leksa station (Leksa on-reef), the cluster and the coral cages were placed into a living reef area at 180 m water depth (Fig. 1A). At the second Leksa location (Leksa off-reef), baskets were placed on the bare sediment off the live reef zone at 218 m (Fig. 1B) to determine whether coral survival and growth is also supported in areas of no coral growth in the vicinity of the living reef. At the Sula Reef Complex (Sula), coral cages and the bioerosion cluster were deployed at the southernmost third of the reef chain in a small depression almost completely engulfed by the live reef at 304 m, about 50–100 m away from the nearest living Lophelia colonies. Unfortunately, in Sula no coral baskets could be deployed into the living reef structures because of limited dive possibilities due to rough weather conditions.

After 1 year, in July/August 2014, all locations were revisited with RV POSEIDON (POS473) and all coral cages and bioerosion clusters were recollected by means of JAGO and brought back aboard (permit number 14/1781 of Directorate of Fisheries for cruise POS473), where they were immediately transferred to large holding tanks. Except for one coral cage in the Leksa off-reef location, all coral cages and clusters could be retrieved. Soon after recovery, the coral fragments and bioerosion cages were weighed on board following the same protocol and using the same equipment as outlined above. Afterwards, all samples were dried at 60 °C on board and packed cushioned for later laboratory analyses.

Post-cruise analyses

After the cruise, all samples were dried at 70 °C for several days until constant weight was reached. Dry weights of empty baskets, cable ties, and corals were separately measured. Calcifying epibionts grown on the dead erect coral framework as well as on the live coral fragments (carbonate accretion) over the year of exposure were removed with a scalpel and weighed separately after drying to constant weight. Then, all samples were scanned via computed tomography (CT) for volume and surface area analysis (see detailed description of the methods below), before linear extension rates of the ‘live’ coral fragments were determined by measuring the distance from the Alizarin Red S band of each corallite to the rim of the calyx using a digital calliper. As growth of the calices is sometimes more pronounced on one side of the calyx than on the other, this measurement was done on two opposing sides, with the lowest and the highest distance between stain bands and rim of each corallite. In addition, the numbers of newly grown (completely unstained) corallites, as well as the number of died polyps (calices without tissue) of the ‘live’ coral fragments were counted. When being deployed in 2013 it was made sure that only intact corallites remained on the fragments, while all empty corallites were removed. Counted newly grown corallites and dead polyps were compared to the total number of corallites of each branch to assess budding rate as well as mortality as a percentage over the course of the experiment. Finally, all samples were dried again to constant weight and combusted at 500 °C for 5 h for differentiation of organic vs. inorganic content.

All fragments were dried to constant weight at 68 °C before tissue residuals were removed with chlorine bleach according to the method described by Davies (1989). Afterwards, the buoyant weight of the fragments without tissue residuals was measured. In order to get rid of all accumulated air bubbles within the skeletal structures, fragments were treated in a vacuum drying cabinet in beakers filled with seawater, so that the weights were not falsified by additional buoyancy. After being washed in distilled water, the fragments were dried and weighed again until constant weight was reached and the skeletal densities were calculated from the following equation following the method by Davies (1989): $${\delta }_{\mathrm{S}\mathrm{k}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{o}\mathrm{n}}=\frac{{\delta }_{\mathrm{s}\mathrm{w}}}{\left(1-\frac{\mathrm{B}\mathrm{W}}{\mathrm{D}\mathrm{W}}\right)},$$ with δ_sw = density of the seawater, BW = buoyant weight of the coral fragments without tissue, and DW = the dry weight of the fragments without tissue.

Calculations and statistical analyses

Growth and bioerosion rates were calculated according to descriptions in Davies (1989) based on buoyant-weight gain or loss of the coral skeleton over time. The gained rates were normalised to weight change per day as a percentage of the initial weight of the coral (G % d⁻¹) as parameterisation predominantly used in experimental studies with live corals applying the buoyant weighing technique. In addition, rates were normalised to weight change in grams per square metre coral surface (gained from the CT measurements) per year (g m² yr⁻¹), which represents the most common unit in bioerosion studies. Data are depicted as mean ± standard deviation (SD). Statistical analyses were performed using SigmaPlot© (version 12.0; Systat Software, Inc., San Jose, CA, USA) and MS Excel Redmond, WA, USA. For statistically comparing the results between the three locations of white as well as orange coral colourmorphs, One-way analysis of variance tests (ANOVAs) were carried out with n = 4 replicates per location, except for the Leksa off-reef location at which one basket and therewith one orange and one white replicate were missing. In case of statistical differences, a post-hoc test for pairwise multiple comparisons following the Holm-Sidak method was carried out to distinguish differences among groups/locations. Whenever data were pooled to increase the sample size and statistical power, this was done upon confirmation that there were no significant differences in the statistical tests among pooled groups. For direct comparisons of white and orange corals or only two locations, t-tests were performed. In order to obtain more accurate and reliable means for conversion factor calculations, outlier tests were carried out in MS Excel (Excel QUARTILE and OR functions).

Coral structural analyses

Coral surface area, volume, corallite number, and skeletal density of live and dead coral fragments (Table 1) were gathered post-experiment after recovery of the coral and bioerosion baskets. Volume and surface area were significantly different between live corals from both Leksa sites and the Sula Reef in both colourmorphs (p ≤ 0.001; One-way ANOVAs), with significantly less bulky coral fragments deployed in Sula compared to both Leksa locations. While Leksa corals had 119 ± 44 polyps/corallites per fragment on average, the Sula corals had only 35 ± 11 polyps per branch (Table 1). This is attributable to the different morphology of the offshore corals. While fjord colony growth is more compact, offshore coral growth tends to be more extended and branched, which corresponds to lower polyp numbers as well as surface area and volume in Sula despite similar fragment sizes like the Leksa fragments. Both surface area and volume of the coral branches correlated well with polyp count (R² = 0.7) with a slightly better correlation of surface area with total polyp count than volume. Figure 2 shows exemplary CT scan images of a live coral fragment (A), and dead coral framework (B) from one basket of the cluster.

Mean skeletal density of all live corals was 2.734 ± 0.043 g cm⁻³. Orange corals had a slightly lower skeletal density (<1%) by trend than white corals. Note that the skeletal density of the orange coral fragments from the Sula Reef was considerably lower than the densities of all other fragments and was identified as outliers. The outlier values were therefore omitted from the average skeletal density of live corals. Bioeroded skeleton material had slightly higher densities than live corals averaging 2.758 ± 0.031 g cm^-3. Both, white vs. orange live as well as bioeroded vs. live coral skeletons were not significantly different in densities (t-tests). For the calculation of growth and bioerosion rates the specific density means of live or bioeroded skeleton material was used.

Mortality of live Lophelia fragments

Polyp mortality of the branches was quite variable between the replicates within and among locations, ranging from 0–86% dead polyps per branch. The highest variability was found in the Leksa on-reef location (Fig. 3). There was no statistically significant difference in mortality between locations (white and orange live coral branches separately tested or pooled). However, lowest mortality was found in Sula with only half as many dead polyps as a percentage of the total polyp count of a branch (10 ± 14%) as in the Leksa off-reef location (21 ± 19%) and one third of the percentage of the Leksa on-reef group (30 ± 27%). Comparison of polyp mortality between white and orange fragments (Fig. 4) revealed a statistically significant difference when white and orange corals were pooled over all three locations (p = 0.002; Mann–Whitney Rank Sum Test). While the orange coral fragments had on average 8 ± 9% dead polyps per replicate, the white corals had 33 ± 23% (Table 2).

Linear extension rates

The overall mean extension rate of all stained corallites of the living coral branches (not all corallites incorporated the dye) from all sites was 2.12 ± 0.86 mm yr⁻¹ (n = 18; Table 3). Examples of coral branches with corallites showing the Alizarin Red S band are depicted in the photographs in Fig. 5. There were no statistically significant differences in average linear extension rates of the replicates, neither between the three locations (One-Way ANOVA) nor between colourmorphs (t-test) (Fig. 6). Nevertheless, the orange specimens tended to have ~15% higher linear extensions than the white ones (pooled over all locations: 2.31 ± 0.90 mm yr⁻¹; n = 8 (orange) vs. 1.96 ± 0.84 mm yr⁻¹; n = 10 (white); Table 3). However, especially the orange corals showed high variances between replicates, which was pronounced most strongly in the on-reef replicates of Leksa, similarly to weight gain. Moreover, Sula corals showed considerably lower growth (~46% less average linear extension) compared to inshore sites. Within the orange coral group this is, however, based on only one replicate of the Sula location as the dye was visibly incorporated in only one of the four replicates at Sula. Thus, averaging all Leksa corals regardless of white or orange from both Leksa sites and comparing Leksa and Sula extension rates revealed a statistically significant difference with ~44% higher extension rates (p = 0.03; t-test; 2.34 (n = 14) vs. 1.31 mm yr⁻¹ (n = 4); Fig. 7). Average linear extension rates correlated well with weight gain in percent per day (R² = 0.83).

The amount of newly grown corallites that developed after staining was similar in all locations and averaged 47.4 ± 12.5%. New corallites alone had significantly higher extension rates (p = 0.043; t-test) than all stained polyps of the branches (total extensions; including newly grown polyps and all corallites where staining bands could be determined). Mean linear extension of newly grown polyps/corallites over all locations and specimens was 2.84 ± 1.04 mm yr⁻¹ compared to 2.11 ± 0.86 mm yr⁻¹ total extension on average (Fig. 8), and compared to 1.87 ± 0.59 mm yr⁻¹ when taking only the ‘old’ stained corallites alone (p = 0.003; t-test comparing newly grown and stained corallites excluding newly grown). Comparing new vs. total extensions (averaged over all stained and newly grown corallites) of the different groups shows that the greatest effect of new growth took place in the Leksa on-reef location. On-reef, 60–75% higher linear extension rates of new corallites of white and orange specimens were gained, while in the Leksa off-reef location it was less than half as much (20–33%). In Sula, growth rates of newly grown corallites were not different or even lower than total stained corallites, although the percentage of newly grown corallites per branch was similar to the percentage of newly grown polyps on the Leksa branches. As newly grown corallites make up for almost half of all stained corallites the pattern of the slight differences between location and/or colourmorph is similar to average extensions of all (old and young) corallites.

Calcification rates of live corals

Overall net carbonate production rate of all observed live coral fragments based on buoyant weight measurements (SD of the 10 consecutive measurements of each fragment of the Leksa weighing session = 0.052 g and Sula weighing session = 0.082 g) was 0.0122 ± 0.0103% d⁻¹ or 61.7 ± 48.2 g m⁻² yr⁻¹ (n = 22; Fig. 9; Table 3). Mean values of white and orange coral fragments averaged over all locations were 0.0107 ± 0.0057% d⁻¹ or 56.02 ± 32.01 g m⁻² yr⁻¹ (n = 11) and 0.0138 ± 0.0137% d⁻¹ corresponding to 67.42 ± 61.51 g m⁻² yr⁻¹ (n = 11), respectively. Calcification rates did not differ statistically significantly between white and orange corals (averaged over all locations; t-test), nor between the different sites (One-way ANOVAs). However, corals from Sula Reef generally showed lower calcification rates with only half as much CaCO₃ precipitation on average (48%) as the Leksa off-reef corals when comparing all live corals (white and orange pooled) of the single sites (p = 0.029; t-test). Although growth rates in the top reef replicates in Leksa on-reef were even higher on average than at the Leksa off-reef site, the top reef corals were not significantly different from the Sula corals. A similar comparison of all inshore vs. offshore replicates as for the linear extension rates (compare Fig. 7) revealed a similar picture with 50% higher growth rates in weight gain in % d⁻¹, though not statistically significant due to the high variability in the Leksa on-reef location (0.015% d⁻¹ (n = 14) vs. 0.0074% d⁻¹(n = 8)). Differences between the different locations and between the two colourmorphs of L. pertusa are shown in Fig. 10.

Bioerosion rates and epibiont carbonate accretion rates of dead erect coral framework

Carbonate degradation rates presumably resulted primarily from bioerosion processes, as the aragonite saturation of the seawater was supersaturated at all locations at the time of deployment as well as recovery of the cages (Ω > 1.7; Table 4) and seasonal undersaturation (Ω < 1) is expected very unlikely. Thus, physicochemical dissolution of the corals’ skeleton is considered negligible here and degradation rates are referred to as bioerosion rates in the following. Bioerosion rates of the dead erect framework integrated over all locations (expressed here as negative values for indicating a loss in weight as opposed to the gain in weight by coral calcification) was −0.0020 ± 0.0015% d⁻¹, corresponding to −12.37 ± 9.40 g m⁻² coral surface yr⁻¹ (Fig. 9). Highest degree of bioerosion took place in the Leksa on-reef location with −0.0036 ± 0.0012% d⁻¹ or −23.20 ± 7.87 g m⁻² yr⁻¹, which was 74% higher than the off-reef site in Leksa with −0.0009 ± 0.0007% d⁻¹ or −5.88 ± 4.42 g m⁻² yr⁻¹ and 64% higher than the offshore location Sula with −0.0013 ± 0.0003% d⁻¹ or −8.03 ± 2.24 g m⁻² yr⁻¹. Values from the on-reef Leksa location were statistically different from both off-reef placements (p ≤ 0.001; One-Way ANOVA; Fig. 10; Table 5), despite highest variation between replicates in the on-reef site. The off-reef locations (in Leksa and Sula) were not significantly different from one another.

Carbonate accretion by calcifying epibionts that grew during the 1 year of exposure was 0.0029 ± 0.0013% d⁻¹ (18.48 ± 8.54 g m⁻² yr⁻¹), accounting for about one-fourth (23.7%) of the growth in percent per day of the living corals (Fig. 9; Table 5). However, this number has to be taken with caution, as particularly accretion might be subject to estimation errors. Intensity of carbonate accretion in the dead framework was found to covary with the observed bioerosion rates, with highest accretion in the Leksa on-reef location and lower accretion rates in both off-reef sites. Carbonate accretion differed significantly only between the Leksa on-reef location and the Sula Reef (p = 0.006; One-Way ANOVA; Table 5; Fig. 10).

Similar to the growth rates of living corals, bioerosion as well as accretion rates showed the highest variability of rates (highest SD of the mean) at the on-reef location.

Conversion factors

We used various methods for growth rate measurements as well as for the normalisation of the different variables, and are thereby able to provide conversion factors for coral growth in size and weight and for the standardisation of these data (Table 6). Since there were no statistically significant differences between colourmorphs and locations, conversion factors for growth rates based on differences in buoyant weight or linear extension rates as well as buoyant weight vs. dry weight, dry weight vs. volume and surface area, and weight, volume or surface area vs. number of polyps were averaged across all samples of live corals. Weight, size, and polyp number correlated well (R² ranging from 0.616 to 0.999).

Growth and mortality of living corals

Mortality

Compared to recent laboratory studies with L. pertusa specimens from the same sites (Büscher, Form & Riebesell, 2017), polyp mortality in this in situ experiment was relatively high with 10–30% on average, depending on location. While the staining approach might have contributed to the mortality, this was not observed during the recovery days in the tanks on board directly after staining. Moreover, growth studies on L. pertusa using Alizarin as staining method did not show enhanced mortality over prolonged experiments and no lethal effects were detected related to the dye (Brooke & Young, 2009; Form & Riebesell, 2012; Lartaud et al., 2017).

A comparison of different sites revealed no significant differences in mortality, but a general trend towards lower mortality in off-reef sites with lowest average percentage of dead polyps in Sula in white and orange corals (65–67% lower average percentage of dead polyps than at the Leksa sites). Although levels of sedimentation were expected to be highest in Sula, where the coral baskets were placed in relatively soft sediment, sediment particles as well as overgrowth by epibionts were more pronounced on Leksa specimens and less on Sula corals. L. pertusa was found to be fairly resilient to sediment loading, since this species efficiently cleans itself through ciliary action and mucus shedding, and its survival is at risk only when completely buried for several days (Brooke, Holmes & Young, 2009; Larsson & Purser, 2011). Moreover, Brooke, Holmes & Young (2009) tested two different morphotypes of L. pertusa, the heavily calcified form with thick branches and the more fragile form with smaller branches and corallites, with regard to their tolerance towards sedimentation and burial and found no difference between morphotypes. Thus, differences in mortality among locations due to sedimentation in the different habitats and sites or due to the different morphotypes between inshore and offshore branches are rather unlikely.

Instead, the higher mortality at the inshore sites may be related to stronger environmental fluctuations with regard to abiotic factors such as temperature, salinity, and currents, as the hydrodynamic conditions inshore are more variable than offshore. Moreover, elevated concentrations of nutrients or pollutants due to aquacultures in the Trondheimsfjord, for example, may be a possible explanation for higher polyp mortality inshore. Higher nutrient levels inshore must yet be verified.

Orange specimens showed far lower mortality (less than a third) at all locations compared to white specimens. In addition to slightly higher growth rates (18–30%), this implies that another underlying mechanism than environmental differences or handling stress causes coral mortality to a different extent in the two colourmorphs.

Bioerosion and accretion of dead coral framework

Conversion factors

In our study we compared two established methods (buoyant weighing and linear extension measurements) to directly assess natural growth rates of living corals and measured different parameters to describe the coral fragments physically (e.g. weight, volume, polyp count). These different approaches allowed for computing conversion factors of the corresponding parameters to transform growth estimates based on buoyant weight measurements to linear extension rates and vice versa, for example, and to convert standardisation parameters such as dry weight, buoyant weight, volume, and surface area of L. pertusa. This might be helpful in future studies for a better comparability of different normalisations of physiological data of this species and for broadening assumptions to non-measured parameters.