ABSTRACT
The cold-water scleractinian coral Lophelia pertusa (Linneaus 1758) is the main ecosystem engineer in the northeast Atlantic (Freiwald et al� 2004) building important habitats for fish and invertebrates (Costello et al� 2005, Mortensen et al� 1995, Reed 2002, Ross and Nizinski 2007)� Coral presence increases diversity threefold compared to surrounding soft-sediment habitat (Gage and Roberts 2003, Henry and Roberts 2007, Husebø et al� 2002, Jensen and Frederiksen 1992, Jonsson et al� 2004, Jonsson and Lundälv 2006), and loss of coral habitat adversely affects local fisheries (Fosså et al� 2002, Koenig et al� 2000)� It is estimated that 30% to 50% of the Norwegian reefs have been damaged by bottom-trawling (Fosså et al� 2002)� Likewise, the coral coverage
CONTENTS
Introduction ������������������������������������������������������������������������������������������������������������������������������������ 113 Materials and Methods ������������������������������������������������������������������������������������������������������������������ 115
Experimental Design ����������������������������������������������������������������������������������������������������������������� 115 Tanks and Substrates ����������������������������������������������������������������������������������������������������������������� 116 Feeding and Monitoring ������������������������������������������������������������������������������������������������������������ 116 Growth Measurements �������������������������������������������������������������������������������������������������������������� 118 Statistics ������������������������������������������������������������������������������������������������������������������������������������ 118
Results �������������������������������������������������������������������������������������������������������������������������������������������� 119 Mineral Accretions�������������������������������������������������������������������������������������������������������������������� 119 Coral Response to the Treatments �������������������������������������������������������������������������������������������� 120
Discussion ��������������������������������������������������������������������������������������������������������������������������������������124 Comparisons with Previous Studies ������������������������������������������������������������������������������������������ 125
Summary ���������������������������������������������������������������������������������������������������������������������������������������� 127 Acknowledgments �������������������������������������������������������������������������������������������������������������������������� 128 References �������������������������������������������������������������������������������������������������������������������������������������� 128
and habitat complexity of the L. pertusa reefs in the Skagerrak have been severely reduced� In the Swedish part of Skagerrak, only one small live reef, consisting of small detached and scattered colonies, remains, whereas six hitherto known reefs are extinct and consist only of dead coral rubble� Rubble fields are known to give a poor substrate for recolonization, with little or no recovery in both high and low wave-energy environments, and, rather than recovery, a further deterioration has been the case (Brooke et al� 2006, Clark and Edwards 1994)� Furthermore, due to geographic isolation of reef sites in the area, the natural recolonization by coral larva from neighboring reefs is unlikely� For example, dispersion probabilities using a Lagrangian model shows a mere 4-9% probability of larval recruits from the nearest Norwegian reef (Tisler) reaching Saekken (Ericson and Ljunghager 2006)� These results are corroborated by genetic data showing high genetic differentiation between reefs in Skagerrak (Broberg 2006) and high levels of clonality within reefs (Dahl 2006)� Hence, it seems that rehabilitation efforts by means of deployment of artificial reefs with coral transplants are necessary to restore cold-water coral cover in the area�
The method of mineral accretion through electrolysis in seawater for the purpose of coral reef rehabilitation was developed by Wolf Hilbertz and Thomas Goreau (1996) and has been experimentally tested by several workers (Borell et al� 2009, Eisinger 2005, Goreau et al� 2004, Romatzki 2009, Sabater and Yap 2004, Schuhmacher et al� 2000, van Treeck and Schuhmacher 1997)� The results have widely varied between experiments and between species of corals, and although measurements from rehabilitation projects around the tropics find growth rates two to six times higher than natural coral colonies, the scientific reports published so far have more ambiguous results to present� Few of the studies from rehabilitation projects have yet been published except as theses in Indonesian� It has nevertheless been substantiated that there is an increased survival rate of coral transplants growing on cathodes (Sabater and Yap 2002, van Treeck and Schuhmacher 1997)� The method has thus far been used in the rehabilitation of shallow tropical coral habitats, and has never been tested in deeper temperate habitats�
The occurrence of L. pertusa on oil rigs (Gass and Roberts 2006) and the use of electrical stimulation in the form of microcurrent electrical therapy (MET) in human and animal medicine with several positive effects observed in different studies (Cheng et al� 1982, Kloth 2005) suggest the possibility of using metal structures and trickle currents to enhance survival and growth in transplanted corals� Gass and Roberts (2006) reported high growth rates of corals growing on oil rigs; trickle currents are used to protect the metal from corrosion and could explain the positive effects on growth� Cheng et al� (1982) observed an increase in adenosine triphosphate (ATP) production in skin-tissue samples exposed to microcurrents ranging from 50 to 1000 μA; thus there is a substantiated positive effect of microcurrents on the ATP production� Furthermore, the application mimics the natural mode of mineralization� Corals produce an alkaline environment in an enclosed space between the calicoblastic cells and the substrate that promote the spontaneous precipitation of aragonite crystals, with the deposited organic matrix working as a primer� Electrodeposition produces the same type of orthorhombic aragonite crystals, thus growing a seminatural substrate on the metal conductor� The proposed benefits for corals (and other calcifying organisms) growing on the cathodes are as follows: (1) supersaturation of calcium and carbonate ions in the vicinity of the cathode, (2) increased efficiency of cation uptake and transport due to the availability of electrons, and (3) increased metabolic efficiency because free electrons are available for ATP production (Hilbertz and Goreau 1996)�
The aim of this study was to evaluate the potential of mineral accretion for the rehabilitation of cold-water coral habitat� More specifically, a predeposited mineral layer was tested with galvanic elements (GEs) (~0�4 V) and three levels of applied electrode potential (2�0 V, 2�5 V, and 3�0 V) without applying electric current to find the optimal level of current density, considering survival and growth of the coral transplants, or if the substrate per se without applied electrical current would have a positive effect� In addition, the response in polyp behavior (degree of extension and activity) and budding frequencies were evaluated�
MATERIALS AND METHODS
The experiment was conducted at the marine research station in Tjärnö (Department of Biological and Environmental Sciences, University of Gothenburg), at the Swedish west coast facing the Skagerrak� The research facility has a complete flow-through saltwater system with deepwater of similar composition (chemistry, salinity, etc�) as the ambient water of the nearby reefs of L. pertusa� The experiment was launched on September 3, 2008, and ran for six months�
Experimental Design
Twenty-four aquarium tanks were set up in a constant temperature room with the inflowing water temperature set at 8°C to imitate the in situ conditions for local coral populations (natural range: 4-10°C)� The water intake for the deepwater flow-through system is at 40-45 m depth in the adjacent Koster Fjord� The tanks were assigned either one of the six different treatments to be tested, four replicate tanks per treatment with four coral pieces in each, making a total of 96 coral fragments (mean length ± SD: 47�2 ± 12�6 mm and size range: 22�2-85�4 mm)� The coral pieces were randomly distributed between tanks� The number of calices on each fragment ranged from 3 to 20� Three different morphological types were recognized: white slender (45 pieces), compact white corals (31 pieces), and a red morph (20 pieces) with smaller calices and a thin, richly branched skeleton� The red morph was spread out to have not more than one representative in each tank� The corals were collected with a remotely operated vehicle (ROV) (Sperre subfighter 7500 DC) equipped with a manipulator arm at ~100 m depth at the Tisler reef (58°59�70′N and 10°58�00′)� Six different treatments were applied as follows:
The AS treatment could be viewed both as a procedural control and a zero-voltage treatment, testing the substrate per se� The GEs with steel cathodes and zinc anodes have a galvanic potential of ~0�4 V in seawater, and thereby offers a lower current alternative to the three treatments with applied DC current� The different voltage levels were applied by three EP-613 DC sources with three-digit LCD displays (0-30 V and 0-3 A, Manson Engineering Industrial Ltd)� Connections were made with the four replicates of each treatment in parallel from one DC source� The voltage was monitored and kept constant throughout the experiment, while the amperage decreased over time� The different treatments were randomly distributed between the tanks as far as possible to avoid artifact effects of position in the room� The cables and connections, however, made it impossible to fully randomize, and therefore all the AS treatments were assembled on the same bench together with one control and one GE tank� The current density (A m−2) was calculated for comparisons with other studies where A m−2 has been used�
Tanks and Substrates
The tanks were constructed with two separate main chambers: one that would contain the corals and one that would hold the anodes or left empty� Perforated partition walls guaranteed that there was no back-mixing between the chambers to avoid exposure to the lower pH or chlorine gas produced at the anodes� All the tanks were given the same design to have equal flow regimes, and all anode chambers were covered with plastic and sealed with duct tape� The sealing of the anode chambers were done to retain chlorine gas within the water to purify and neutralize the outflowing water in filter trays containing activated carbon and oyster and mussel shells prior to release�
The cathodes were cut to 110 × 160 mm pieces from a steel mesh with a metal surface area calculated to 0�04 m2� Average weight of the cathodes was 253 g� The anodes were made of a 1 × 1 mm angular wire mesh of coated titanium (150 × 200 mm and average weight 12 g)� The GEs were produced by attaching zinc anodes to one end of the cathodes, ~43�8 g of zinc on each GE� In the controls, the corals were placed on a plastic mesh of the same size as the cathodes; mounted on a small, flat stone; and affixed with aquarium silicon�
All materials were conditioned in seawater for five weeks before mounting the corals� The AS treatment was run on electrolysis (4�0 V) during these weeks to get a mineral cover; however, the deposited minerals consisted mainly of brucite and were very porous� Additional short periods of DC currents (3�0 V) were given on six occasions during the first three weeks of the experiment, to counteract corrosion on cathode surfaces that had its mineral layer scraped off while mounting the corals� When the experiment was terminated, the mineral layers were measured with a pair of vernier calipers, with five replicate measurements at each end of the cathode (close or far end relative to the anode)�
Feeding and Monitoring
During the experiment, the corals were fed with Artemia salina nauplii (brine shrimp) on three occasions per week� The Artemia were hatched over 48 hours and reared for another 48 hours while fed with microalgae (Isochrysis sp�) to the increase nutrient value� The food was supplied via the water flow-through system, to be evenly distributed� Four plastic containers were used to buffer water, so that the water levels in the containers and thereby pressure and flow rate to the tanks could be kept equal� Each container supplied six tanks each with water� Flow rates were measured on three occasions during the experiment with three replicate measurements from each tank�
Water temperature and pH values were monitored weekly during the experiment with a handheld digital pH meter (Waterproof pH Testr 30, accuracy; ± 0�001 pH units and ± 0�5°C)� Measurements were taken 1 cm above the substrates and randomized between the tanks� The pH were between 8�06 and 8�10 (see Table 10�1 and Figure 10�1)� Water temperature was 7�07°C ± 1�45°C (mean °C ± SD), starting around 10�0°C and leveling off at around 5�5°C-6�0°C over several weeks in the second half of the experimental period� During the four last months of 2008, the measured salinity at 30 m depth (close to the depth of water intake) in the Kosterfjord was 32�5-33�8 psu (SMHI Report No� 2009-6)�
The health status of the corals was likewise monitored on a weekly basis; the corals were given points based on the degree of extension of tentacles� Fully withdrawn = 0, partially withdrawn = 1, extended but slack = 2, intermediate = 3, extended and vivid (stiff) = 4, and actively moving tentacles = 5 points� The health-status observations were not randomized due to the corals’ sensitivity to vibrations that caused them to withdraw at the slightest disturbance� Instead, the observations were made in the same order, from tank 1 to 24, as swiftly and quietly as possible, and repeated in the opposite direction for another day of the week to elucidate whether there would be a difference in extension rates due to procedures�
Growth Measurements
Photographs of the coral fragments were taken at the point of start and end of the experiment, with the pieces placed on a gray perforated panel to provide a reference for measurements (see Figure 10�2)� Measurements were made in the free software ImageJ (version 1�42a, Wayne Rasband, National Institutes of Health, USA)� Several measurements on each coral fragment were made, and mean growth and maximum observed growth were recalculated to growth in mm yr−1 for analysis�
Some of the corals had broken calices after the handling during collection, and these had either died or healed to different degrees� The mostly one-sided growth of healing was measured and noted separately in the protocol� Furthermore, the degree of healing was assigned a symbol; one plus sign (+) for a moderate healing (0�8 ≤ × < 2�0 mm), two plus signs (++) for a good healing (≥ 2�0 mm), and a minus sign (−) for broken calices that did not heal� A zero (0) denotes that there were no broken calices�
The numbers of new buds were counted, and partial or complete mortality-that is, the number of dead polyps (not counting those that were already dead at the beginning) divided by the total number of polyps on each fragment-was noted�
Statistics
The frequency distributions of the number of new buds and the calices displaying a growth of 0�8 ≤ × < 2�0 mm yr−1 or ≥ 2�0 mm yr−1 were analyzed by Chi-square (χ2) tests� A one-way ANOVA was performed on growth rate data, and Pearson Correlations were performed to test whether the different water-flow rates had an effect on the variables’ mean and maximum growth or the frequencies of new buds and calices with a growth of ≥ 2�0 mm yr−1�
All statistical analyses were performed with the statistical software SPSS 17�0 (2008)�
RESULTS
Mineral Accretions
The accreted mineral layer varied in thickness between the treatments and with distance to the anode� Higher current density produced a thicker layer, and subsequently, the end of the cathode that was closer to the anode received a thicker layer of precipitated minerals (Table 10�2)� On the LI cathodes, the measured layer was 1�53 ± 0�74 mm (mean ± SD) closest to the anode and 0�99 ± 0�60 mm on the end furthest away� The LII treatment had 5�07 ± 0�85 mm and 2�48 ± 0�62 mm
thick layers on the close and far end relative to the anode, respectively� Both the treatments had produced a hard crust� The AS treatment had layers between 7�90 ± 2�03 and 1�98 ± 0�30 mm� Noteworthy, the rapid deposition of minerals on the AS treatment produced a very porous layer initially, which subsequently was strengthened, ending up a hard crust� The efforts to counteract corrosion on the AS cathodes failed� The LIII treatment gave a rapid accretion of porous minerals that crumbled while dismounting the corals and thus could not be properly measured; however, layers were up to 10 mm thick on one side only� There were no visible mineral accretions on the GE cathodes� The mineral layer on LI barely covered the cathode, whereas full accretion was accomplished on the LII cathodes� Coral fragments were firmly attached to the cathode surfaces by the mineral accretions�
The amperage decreased rapidly from the initial values during the first week, and after the initial drop, the amperage leveled out and decreased only slightly over the remaining period (Table 10�2)� The mineral accretion rates were equal within the treatments, indicating an equal distribution of electrical currents over the four replicate electrode pairs�
Coral Response to the Treatments
There were significant differences between the treatments in the frequency distribution of new buds developing during the experimental period (χ2(df 5) = 12�3, exact p = 0�03, and χ2crit = 11�07 at the 0�05 level)� There were 17 new buds in the LI treatment, 13 new buds in the GE treatment, 10 in the LII treatment, and the controls had 8 new buds (see Tables 10�3 and 10�4 and Figure 10�3a)� Pairwise Chi-square tests between the controls and the different treatments revealed no significant effects� Testing the substrate per se (AS) against the treatments with an electrode potential (GE, LI, LII, and LIII) also turned out nonsignificant; however, the difference between AS versus LI was close to significant (χ2(df 1) = 4�17, χ2(crit) = 3�84, and exact p = 0�06)� The largest difference was found between LI and LIII (χ2(df 1) = 9�80, χ2(crit) = 3�84, and exact p = 0�003)�
The number of calices displaying a growth of ≥ 2�0 mm yr−1 (see Figure 10�3c) differed significantly between the treatments and were higher in the controls, LI, and GE with 10, 11, and 9 calices, respectively (χ2(df 5) = 12�5, and exact p = 0�03)� Pairwise tests were nonsignificant, except controls versus LII (χ2(df 1) = 7�36, χ2(crit) = 3�84, and exact p = 0�01)� The number of buds displaying a growth rate of 0�8 ≥ × < 2�0 mm yr−1 was similar in all treatments except GE and LIII, which had lower numbers, albeit nonsignificant (Figure 10�3b)�
The one-way ANOVA performed on growth-rate data turned out nonsignificant, and only diagrams with means and standard error bars are presented to show the trends in the effects on growth (Figure 10�3d through 10�3f)� The observed mean growth was slightly higher in the lowest applied current density treatment (LI) than in the controls, that is, 1�27 ± 1�22 mm yr−1 and 1�20 ± 0�97 mm yr−1, respectively (mean ± SD), while all other treatments had lower growth rates (see Table 10�4 and Figure 10�3d)� The same pattern was seen in the observed maximum growth where corals in the LI treatment had an average maximum growth of 3�26 ± 2�66 mm yr−1 compared to 3�05 ± 1�73 mm yr−1 in the controls (Figure 10�3e)�
All Pearson Correlations testing whether there were any effects on growth rates or frequencies due to different water flow rates (mL s−1) in the tanks were nonsignificant (Table 10�5)� The r2 value of the maximum growth indicates a lower-range medium-effect size (Kinnear and Gray 2008), but since there was a nonsignificant correlation, this is interpreted as no effect of water-flow rate�
Considering the general health status, it seemed as all corals fared as well; the corals in the controls were extending their polyps to a slightly higher degree, whereas the polyps on the GEs were slightly less extended and vivid, albeit no significant results were found (Figure 10�3f)� Looking at the health status derived from checks in one direction only (tanks 1 through 24), it seemed like the corals on the zero-voltage treatment (AS) fared less well; however, this pattern disappeared when pooling the checks from both directions, as presented in Figure 10�3f�
In the summary table (Table 10�4), there are some additional information presented; heal growth (mm) together with degree of healing (ranks) and mortalities� The healing of broken calices was highest in the controls with an average one-sided growth of healed calices of 1�29 ± 1�94 mm followed by GE, AS, and LI� The controls displayed a high degree of healing, whereas AS, GE, and LI displayed cases of good healing simultaneous with some cases of no healing� Because the number of broken calices differed among treatments, the degree of healing is not entirely comparable; for example, the higher current density treatments (LII and LIII) had very few broken calices to begin with�
There were zero mortalities in the controls and LI, and a few cases of partial mortality in AS and LII (see Table 10�4)� In the GE treatment, there were four cases of partial mortality ranging from 29% to 67%; two of these are explained by an infestation by bacteria in one of the tanks� The bacteria were not analyzed, but red crystalline metabolites on the surface of the cathode indicated chemoautotrophic iron-oxidizing bacteria� The highest voltage treatment (LIII) produced a high number of partial mortality-eight cases, ranging from 14% to 29%—and furthermore, two complete mortalities� These could all be explained by the treatment itself, as the rate of mineral accretion was so high that it covered the calices close to the cathode surface and entirely overgrew two small coral
fragments� In addition, some of the corals in the LIII treatment had skeletons so fragile that when the experiment was terminated, they fell apart while being dismounted from the cathodes�
DISCUSSION
This study is the first to test the method of mineral accretion on cold-water corals� The most striking result of this experiment is the number of new buds that developed in the low current density treatments (Table 10�4 and Figure 10�3a)� Although no significant effects were found in the pairwise comparisons between the controls and the different treatments, there was a significant difference (p = 0�03) between the treatments in the overall Chi-square test and testing the substrate per se against the charged treatments; the difference between AS versus LI was found to be close to significant (p = 0�06)� The overall difference between the treatments could largely be attributed to the difference between the LI and LIII (lowest vs� highest applied current density treatment and p = 0�003)� Looking at the frequency distributions of new buds (Figure 10�3a), the increase lies over the low current density treatments (GE, LI, and LII), and this result is congruent with the previous studies, which is discussed later� The highest current density treatment (LIII) appears to be overcharged and detrimental to the corals, with very few new buds developing as a result�
Also in the frequency distribution of calices with a growth of ≥ 2�0 mm yr−1, there were significant differences between the treatments, although the controls displayed an equally good growth at the lower current-densities (LI and GE), and this effect could therefore not be linked to trickle currents (Figure 10�3c)� The pairwise comparisons revealed a significantly lower frequency of calices with a growth of ≥ 2�0 mm yr−1 in the intermediate current density treatment (LII), and, rather than a positive effect of applied direct current, there seems to be mainly a negative impact in AS, LII, and LIII� Also in AS and LIII, the frequencies were lower, albeit nonsignificant� The low frequencies of calices with a growth of ≥ 2�0 mm yr−1 in AS and LII is somewhat compensated by the higher frequencies of calices with the lower range of growth rate (0�8 ≥ × < 2�0 mm yr−1)�
The growth rates observed in this study were below or in the lower range of the reported rates for the species, 4-25 mm yr−1 (Freiwald et al� 2004)� Only a few calices on the GEs and the lowest applied current density (LI) showed the growth rates of around 10 mm yr−1, as seen in Table 10�4, where maximum observed growth is presented� Possible causes for this are poor nutritional values
of the chosen food (Artemia) or effects of stress� The trend in the growth rates seen both in mean and maximum observed growth (Figure 10�3d and e) was a slight increase in the growth rate in LI as compared to the controls, whereas all other treatments had lower rates, the differences being more pronounced in mean growth�
Zero mortality was observed in the lowest applied current density treatment (LI) as well as in the controls, showing that a trickle current has no negative impact on coral transplants if the level is optimized, while overcharging could be detrimental, as seen in the higher current density treatments (LII and LIII)� The partial mortalities and overall low performance of the corals in the zero-voltage treatment (AS) and GE could, however, be due to leaking metal ions� Considering the AS treatment, the failure of counteracting corrosion on the spots that had the mineral layer scraped off while mounting the coral fragments could have led to detrimental levels of metal oxides� Rust has been seen to be avoided by settling organisms, leaving bare patches on oxidized metal surfaces (Fitzhardinge and Bailey-Brock 1989)� Leaking ions could also explain the low performance of corals on the GEs, as the zinc anodes were oxidized� Interestingly, the partial mortalities caused by the iron-oxidizing bacteria found in one of the GE tanks did not concur with the impaired growth� Noteworthy, three of the corals growing in this specific tank had the highest growth rates observed within the GE treatment� The bacterial growth was restricted to one side of the cathode and thus left the two corals unaffected by direct bacterial overgrowth� Despite some partial mortality, the growth rate of the living calices in one of the affected corals was relatively high� Chemical reactions mediated by the bacteria could have mitigated the effect of leaking zinc ions, thus giving the positive response in growth in this specific tank�
The measured pH differed slightly between the treatments, with highest pH observed in the LII treatment� The pH was initially high also in LIII but dropped after the 12th week (Figure 10�1)� This drop could be driven by the fast accretion that filled the voids in the mesh, thus trapping hydrogen gas underneath the cathodes, with equilibrium reactions between the gas and water interface decreasing pH as a result� The fragility of coral skeletons in the LIII treatment could be explained by the lower pH toward the end of the experiment� An additional explanation of the observed fragility could be that the corals incorporate more porous brucite minerals rather than aragonite at high current densities�
The observed drop in pH for all the treatments during the first weeks was probably caused by an early autumn down-mixing of surface waters, as an intense low pressure occurred in August, with winds of storm strength (mean wind velocity 24 m s−1) accompanied by heavy rains (SMHI Report No� 2009-6) that could cause salinity to drop during the following period and affect pH� The water temperature at 30 m depth had its annual maxima in September (temporarily warmer than the surface waters), and the stratification of water masses was less pronounced during the period�
Comparisons with Previous Studies
The stimulatory effect in bud production seen in this study is congruent with the results of the study by Sabater and Yap (2004), where treated nubbins of the branching coral Porites cylindrica had significantly higher densities of corallites� Although the results of the present study were nonsignificant in the pairwise Chi-square test between the controls and LI, the concordance with the results of Sabater’s and Yap’s study strengthens the present results� The stimulated budding is an interesting effect that could prove valuable in transplantation programs, as coral transplants that bud off richly initially will have more growing tips and thus a more rapid outgrowth into a complex matrix of coral branches� The morphology of P. cylindrica is very different from that of L. pertusa, and perhaps the latter species is better served by an effect of this nature because budding is directly affecting branching� The mechanism for this, however, can only be speculated� Although Sabater and Yap (2004) ascribes this effect to the increased mineral ion concentration, an alternative explanation could be an increase in ATP production at an optimal current density�
As shown by Cheng et al� (1982) in a clinical in vitro study on the effects of electric currents on skin tissue samples (from rats), currents ranging from 50 to 1000 μA had positive effects on ATP generation, with a threefold to fivefold increase in ATP levels� ATP concentrations leveled when applying currents exceeding 1000 μA and were reduced at currents of 5000 μA� This current-density-dependent ATP stimulatory effect could explain why positive effects are restricted to low current density treatments, while higher levels have a negative impact despite the theoretically higher availability of calcium ions with increasing current densities� ATP is needed for the active transport of calcium ions into the calcifying compartment, as well as for removing the hydrogen ions from the same to maintain the alkalinity necessary for precipitation of aragonite within the compartment (Allemand et al� 2004; McConnaughey and Whelan 1997; Tambutté et al� 1996)� It is also known that feeding increases calcification rates (Houlbrèque et al� 2003 and 2004), and that respiration rates and ATP concentrations are elevated in the growing tips of coral branches, thus supporting the higher calcification rates (Fang et al� 1989; Gladfelter et al� 1989)� Corals use mainly metabolic CO2 for calcification (Furla et al� 2000), explaining the positive effect of feeding on calcification rates� ATP is thus the limiting factor for calcification in saturated environments� High food availability can support a denser coral colony; could the corals be fooled to branching by artificially elevated ATP levels generated by the electrode potential?
