Sugar enrichment provides evidence for a role of nitrogen fixation in coral bleaching

The disruption of the coral–algae symbiosis (coral bleaching) due to rising sea surface temperatures has become an unprecedented global threat to coral reefs. Despite decades of research, our ability to manage mass bleaching events remains hampered by an incomplete mechanistic understanding of the processes involved. In this study, we induced a coral bleaching phenotype in the absence of heat and light stress by adding sugars. The sugar addition resulted in coral symbiotic breakdown accompanied by a fourfold increase of coral‐associated microbial nitrogen fixation. Concomitantly, increased N:P ratios by the coral host and algal symbionts suggest excess availability of nitrogen and a disruption of the nitrogen limitation within the coral holobiont. As nitrogen fixation is similarly stimulated in ocean warming scenarios, here we propose a refined coral bleaching model integrating the cascading effects of stimulated microbial nitrogen fixation. This model highlights the putative role of nitrogen‐fixing microbes in coral holobiont functioning and breakdown.


| INTRODUCTION
The symbiosis between reef-building corals and dinoflagellate algae of the genus Symbiodinium provides the foundation for the ecological success of coral reefs over millions of years (Muscatine & Porter, 1977). In this mutualistic association, the coral host provides inorganic nutrients in exchange for photosynthetically fixed carbon (photosynthates) and amino acids from the algal symbiont (Muscatine & Porter, 1977). Coral bleaching, the disruption of this delicate symbiosis by heat and light stress or poor water quality (Fabricius, 2005;Lesser, 1996;Wooldridge & Done, 2009), may ultimately result in the mortality of the coral host. Mass bleaching events have resulted in unprecedented degradation of coral reefs over the past decades and are expected to increase in frequency and severity as global climate change progresses (Hughes et al., 2003).
Even though several decades have passed since the initial observation of large-scale coral bleaching, our understanding of the underlying mechanistic processes remains incomplete. Among the proposed mechanisms, particularly the idea of oxidative stress as a driver of coral bleaching (Oxidative Theory of Bleaching) has found considerable resonance (Downs et al., 2002). This theory posits that the bleaching cascade is initiated by oxidative stress in the algal symbionts (and host tissues) caused by excessive temperature and light conditions (Lesser, 1996). Yet, there is emerging evidence for a more complex mechanistic response, intimately linking bleaching with environmental nitrogen (N) availability (Vega Thurber et al., 2014;Wiedenmann et al., 2012;Wooldridge, 2013;Wooldridge & Done, 2009). As N limitation is required to regulate Symbiodinium cell division rates and to promote the translocation of photosynthates to the coral, N enrichment threatens the persistence of this symbiosis (Dubinsky & Jokiel, 1994;Falkowski, Dubinsky, Muscatine, & McCloskey, 1993). Specifically, it can reduce photosynthate translocation rates by Symbiodinium (Suesc un-Bol ıvar, Traverse, & Thom e, 2016). Accordingly, Wooldridge (2013) proposed that this retention of photosynthates would result in the energy limitation of coral carbon concentrating mechanisms (CCMs). The resulting carbon dioxide (CO 2 ) limitation of photosynthetic "dark reactions" would render Symbiodinium more susceptible to photodamage (i.e., bleaching).
Indeed, the idea of the "selfish symbiont" was recently confirmed by Ezzat, Maguer, Grover, and Ferrier-Pag es (2015), who reported an increased utilization and reduced translocation rates of photosynthetically fixed carbon by Symbiodinium in hospite under nutrient replete growth scenarios. Further, excess N availability can lower the bleaching threshold in corals by shifting Symbiodinium from an N limited to a phosphorus (P)-starved state (Wiedenmann et al., 2012).
Such stoichiometric shifts can cause the substitution of phospholipids with sulpholipids in the chloroplast thylakoid membranes, a common response in photoautotrophs during limited P availability (Frentzen, 2004). As the lipid composition of the thylakoid membrane is closely linked to the assemblage and functioning of the photosynthetic machinery, it can determine bleaching sensitivity in Symbiodinium (Tchernov et al., 2004). Therefore, increased N availability will ultimately increase the bleaching susceptibility of corals (Wiedenmann et al., 2012).
Our understanding of internal N cycling processes in corals during thermal bleaching remains incomplete. This knowledge, however, is critical, as N cycling microbes are ubiquitous associates of corals (R€ adecker, Pogoreutz, Voolstra, Wiedenmann, & Wild, 2015). In particular, diazotrophs, i.e. Bacteria and Archaea capable of reducing dinitrogen (N 2 ) into biologically available N, constitute an important N source for Symbiodinium (Bednarz, Grover Maguer, Fine, & Ferrier-Pag es, 2017;Benavides et al., 2016;Lema, Willis, & Bourne, 2012;Lema et al., 2016;Lesser et al., 2007). Indeed, N 2 fixation can help sustain coral holobiont productivity when nutrients are scarce . Given its functional importance, it is not surprising that N 2 fixation is associated with the majority of investigated coral species R€ adecker et al., 2014;Shashar, Cohen, Loya, & Sar, 1994). Elevated temperatures, however, stimulate the enzymatic activity of nitrogenase and promote the proliferation and activity of coral-associated diazotrophs Compaor e & Stal, 2010;Santos et al., 2014). Consequently, this has led R€ adecker et al. (2015) to propose that excess N availability from increased holobiont-associated N 2 fixation activity may be a major driver of bleaching in heat-stressed corals.
Here, we aimed to gain a better mechanistic understanding of the role of N 2 fixing Bacteria and Archaea during coral bleaching.
For this purpose, we manipulatively stimulated N 2 fixation activity in corals in the absence of heat or light stress. To achieve this, we supplied doses of labile dissolved organic carbon (DOC), more specifically neutral monosaccharides, to stimulate coral-associated N 2 fixation (Shashar et al., 1994). This approach allowed us to identify the effects of increased N 2 fixation activity on the coral-algal symbiosis, while eliminating the confounding effects of temperature and irradiance. We characterized the cascading effects on critical functions of the coral holobiont and three of its main members-the coral host, algal symbionts, and the prokaryotes-in an integrative approach combining physiological and molecular applications.

| Aquarium facilities, coral collection, and maintenance
The experiments were conducted at the wet laboratory facility of the Coastal and Marine Resources Core Lab (CMOR) at the King Abdullah University of Science and Technology (KAUST, KSA). The aquarium system was comprised of two identical units, each consisting of three replicate experimental tanks (i.e., totaling six tanks 100 L each). To stabilize seawater parameters and oxygen (O 2 ) concentrations, untreated Red Sea reef water was circulated in the experimental units, each containing protein skimmer as well as filtration setups. Further, 30% of the water was replaced on a daily basis, assuring close to natural water parameters. Maintenance conditions were kept constant, allowing us to rear corals in the absence of any heat or light stress (seawater temperature at 27°C, salinity at 40.5 PSU, photosynthetic active radiation 100 quanta lmol s À1 m À2 on a 12:12-hr day/light cycle). In three aquaria, labile DOC levels were manipulated by daily additions of a 10 mg/L saccharide mixture (in mg/L; (D+) xylose: 3.82; (D+) glucose: 2.56; (D+) mannose: 1.39; (D+) galactose: 2.22). Respective contribution of each saccharide was based on reports on the neutral monosaccharide composition of sewage and coral reef macroalgae exudates (Huang, Li, & Gu, 2010;Nelson et al., 2013). The other three aquaria were maintained at ambient DOC levels. To avoid drifting effects on the labile DOC concentrations across the replicate tanks, they were supplied from a recirculation reservoir (100 L) according to treatment conditions. The DOC treatment resulted in >10 times enriched conditions (up to 1609 AE 2 lM after 28 days of treatment) compared to the ambient treatment (117 AE 2 lM after POGOREUTZ ET AL. | 3839 28 days; Table S1). The enrichment did not affect dissolved O 2 levels in the treatment tanks (constantly >6 mg/L) or total N and total P concentrations at any time point; for details, see Tables S1 and S2).
Coral fragments from all colonies were distributed evenly among aquaria tanks and acclimated for 28 days. During this acclimation period, corals were moderately fed to assist recovery from fragmentation stress (Reef Roids, PolypLab, USA). Any additional feeding was abandoned 1 week prior to and throughout the experiment to avoid confounding effects from additional nutrient uptake via heterotrophy.

| Sampling
N 2 fixation activity, diazotroph abundance, maximum quantum yield, and Symbiodinium density were measured at days 0, 7, 12, and 28 of the experiment. Remaining response parameters were measured at the first and last day of the experiment. Noninvasive parameters (pulse amplitude fluorometry, rate functions) were applied in a repeated measures design to increase statistical power. For the remaining (invasive) response parameters, single fragments originating from all mother colonies and treatments were rinsed with filtersterilized seawater (FSW; 0.22 lm), flash-frozen in liquid N 2 , and stored at À80°C until further processing. Seawater samples were collected every 7 days, filtered, and frozen for subsequent analysis of DOC, total dissolved N (TN), and P (TP) content. A brief overview of all measured response parameters is provided in the following; please refer to Supplementary Methods for a more detailed description.

| O 2 and N 2 fixation measurements
Photosynthesis and respiration rates were derived from O 2 evolution/depletion incubations. For this, net photosynthesis and respirations rates were quantified from start and endpoint O 2 measurements of corals incubated in gastight chambers for 2 hr during dark and light conditions, respectively (R€ adecker et al., 2014).
Gross photosynthesis rates (P G ) were calculated as the combination of net photosynthesis (P N ) and respiration rates (R). Similarly, gross N 2 fixation rates were indirectly quantified via measurements of ethylene (C 2 H 4 ) evolution using the acetylene (C 2 H 2 ) reduction assay (Wilson et al., 2012). Specifically, corals were incubated for 24 hr in gastight chambers containing seawater as well as an airfilled headspace both enriched in C 2 H 2 . N 2 fixation rates were inferred from differences in C 2 H 4 concentrations of gas samples collected at the start and end of the light as well as the dark phase of incubation.

| Symbiodinium response and elemental analyses
Photosynthetic performance of Symbiodinium cells in hospite was confirmed by measuring PSII maximum quantum yield (F v /F m ) of darkadapted coral fragments (n = 12 per treatment) 1 hr into the 12-hr dark phase. Measurements were carried out using a pulse amplitude modulation (PAM) fluorometer (DIVING-PAM, Walz, Germany). To assess symbiont density, Symbiodinium cells were freshly isolated from coral tissue by NaOH extraction (Zamoum & Furla, 2012). Symbiodinium cell counts were determined using flow cytometry and normalized to coral fragment surface areas (Lavy et al., 2015).
Isotopic d 15 N signatures and N:P ratios were determined with an isotope ratio mass spectrometer and a photometer, respectively, from dried coral tissue and extracted Symbiodinium cells previously separated by centrifugation and collected on filters. d 15 N signatures of dried material relative to atmospheric N were analyzed with an isotope ratio mass spectrometer (Lesser et al., 2007). Further, TN-to-TP ratios were measured photometrically following Hansen and Koroleff (2007).

| Seawater nutrient measurements
Treatment water samples for nutrient analysis were collected at all sampling points. Treatment water was sampled in 30 and 50 ml triplicates for organic and inorganic nutrients, respectively, filtered (0.45 lm) and preserved with 100 ll of 35% phosphoric acid or frozen at À20°C, respectively. Analysis of DOC was performed with an Apollo 9000 Total Organic Carbon (TOC) Analyzer TM (Teledyne Instruments Tekmar, USA), and TN and TP concentrations were simultaneously measured according to standard method (SM) 4500-P J (Valderrama, 1981). Samples were analyzed by the Analytical Core Lab (ACL) at KAUST, Saudi Arabia, and the Marine Chemistry Lab at the University of Washington, USA, respectively.

| Statistical analysis of physiological parameters and bacterial communities
All statistical analyses of physiological response parameters were conducted in R v3.2.2 (R Development Core Team, 2015). N 2 fixation, P G, and R rates, as well as maximum quantum yield, were tested for individual and interactive effects of treatment and time by 2-factorial generalized estimation equations generalized linear models for repeated measures (GEEGLMs) in the R package GEEPACK (Højsgaard, Halekoh, & Yan, 2006). Similarly, Symbiodinium density, seawater nutrient concentrations, and nifH gene copy numbers were tested with 2-factorial generalized linear models (GLMs). Stable isotope composition and N:P ratios were analyzed in 3-factorial GLMs accounting for individual and interactive effects of treatment, time, and compartment. All models were based on a Gamma distribution with best fitting link function to account for skewing of data. To illustrate significant differences between manipulations, treatment effects of individual time points were compared using unpaired Welch's unequal variances t-test. Bacterial community composition was compared using analysis of molecular variance (AMOVA) as implemented in MOTHUR. All data are reported as mean AE SE; asterisks indicate statistically significant differences (*p < .05, **p < .01; for details, see Table S3).

| RESULTS
3.1 | Stimulated N 2 fixation, diazotroph abundance, and elemental changes Within 7 days of manipulation, DOC additions caused a significant fourfold increase in holobiont gross N 2 fixation activity (assessed via acetylene reduction assay) compared to controls during both light and dark phase (Figure 1a and b; for all model statistics, see Table S3).
Light N 2 fixation activity was higher and more variable compared to dark conditions in both treatments at all times. In contrast to N 2 fixation activity, the relative abundance of diazotrophs (relative number of nifH gene copies as quantified by qPCR) did not exhibit significant changes until day 28, but then experienced a 23-fold increase under high DOC compared to same-day ambient controls (Figure 1c). Stimulated N 2 fixation activity concurred with a significant depletion of the isotopic d 15 N signature and a 40% increase in the N:P ratio over time for Symbiodinium under high DOC (Figure 2). In contrast, coral tissues maintained stable d 15 N signatures over time, but exhibited a doubling in the N:P ratio under stimulated N 2 fixation over the course of the experiment.

| Overall bacterial community
While diazotroph populations proliferated in the coral tissues, the overall bacterial community did not exhibit any significant compositional changes under high DOC over the course of the experiment ( Fig. S1; Table S4). The overall community was dominated by Gammaproteobacteria (78%-85% of all sequences) for both treatments and across all time points. While an overall decrease in bacterial diversity was observed over time in both treatments (Table S5), no enrichment in potentially pathogenic bacterial families, such as the opportunistic Vibrionaceae (class Gammaproteobacteria) as previously reported from corals under DOC enrichment and coral bleaching and disease (Rosenberg & Falkovitz, 2004;Vega Thurber et al., 2009), was found (see Table S6 for abundances of operational taxonomic units (OTUs)).  Table S3 [Colour figure can be viewed at wileyonlinelibrary.com] Strong reductions in photosynthetic efficiency during heatinduced bleaching are well documented (Wiedenmann et al., 2012).

| Coral bleaching
Even though the photosynthetic efficiency experienced a significant reduction, it remained at an overall high level. Hence, this response is not comparable to heat stress responses and not indicative of photodamage and the associated accumulation of reactive oxygen species (ROS). As the upregulation of photosynthetic ROS production is a central mechanism of current bleaching theories, the apparent absence of oxidative stress raises the question about the exact trigger of bleaching in the present study (Weis, 2008  In contrast to the overall stable bacterial community composition, the pronounced response of diazotrophs highlights that the DOC enrichment stimulated this functional group and potentially affected the activity of other holobiont-associated microbes. Stimulated diazotroph proliferation and N 2 fixation activity of similar magnitude have previously been reported for heat-stressed corals Santos et al., 2014), thereby substantiating the proposition of a linkage between N 2 fixation and coral bleaching. Previous studies suggested that N 2 fixation in corals is energy-limited (R€ adecker et al., 2014;Shashar et al., 1994 Noteworthy, N 2 fixation activity in the DOC treatment increased before a relative proliferation of diazotrophs was observed in the tissue. This implies that diazotroph proliferation may have occurred elsewhere earlier in the experiment (e.g., in the mucus or coral skeleton). Further, this suggests that N 2 fixation activity in the holobiont may be limited by energy and environmental conditions rather than diazotroph abundance.

| The role of the microbiome
Importantly, due to the absence of potential pathogen propagation in the overall stable bacterial community, we can effectively rule out pathogenicity as suggested in previous studies (Kuntz et al., 2005;Smith et al., 2006). Consequently, the observation of increased N 2 fixation under these conditions may provide a mechanistic insight into the processes leading to symbiosis breakdown.

| The fate of microbially fixed nitrogen
Although the exact localization of diazotrophs within P. verrucosa remains yet to be determined, the stimulated N 2 fixation activity likely provided excess N to the coral holobiont. Indeed, the depletion in d 15 N in Symbiodinium suggests the direct utilization of N 2 fixation products at significant rates in the Pocillopora verrucosa holobiont, as reported previously for other corals (Lesser et al., 2007). While the underlying mechanism(s) of the transfer of N 2 fixation products to Symbiodinium remain(s) elusive, Benavides et al. (2016) recently showed that the direct transfer of fixed N and heterotrophic ingestion of diazotrophs provides a non-negligible and important N source for Symbiodinium.
This uptake of additional N from N 2 fixation can explain the observed 40% increase in the N:P ratio in Symbiodinium cells in the current study, which are in general constant (Ferrier-Pag es, Godinot, D'Angelo, Wiedenmann, & Grover, 2016). Further, the shift in algal symbiont nutrient stoichiometry suggests that excess N uptake released Symbiodinium from their N-limited state, an important regulatory mechanism maintaining the coral-algae symbiosis (Falkowski et al., 1993).
As d 15 N signatures in coral tissue did not exhibit depletion, we can effectively rule out that increased N 2 fixation provided a significant source of N to the coral host within the experimental time frame. Still, coral tissue N:P ratios experienced an increase steeper than that of the algal symbionts. Although speculative at this point, | 3843 this may hint toward buffering mechanisms. Specifically, the coral host likely "sanctions" N supply to Symbiodinium by withholding N from its own metabolism, possibly to restore N limitation and prevent P starvation. These mechanisms may involve the storage of N derivatives in specialized host cells or organelles as previously suggested (Aranda et al., 2016;Pernice et al., 2012).
Taken together, our findings suggest that stimulated N 2 fixation altered the internal nutrient stoichiometry in the coral holobiont and disrupted the N-limited state of Symbiodinium. In this context, Godinot, Ferrier-Pag es, and Grover (2009) and Ezzat et al. (2016) previously reported shifts toward net release of dissolved inorganic N coupled with increased P and decreased N uptake in heat-stressed coral holobionts. This implies that shifts in internal nutrient stoichiometry may not be exclusive to DOC-induced bleaching, suggesting similar underlying processes may be involved during heat stressinduced (thermal) bleaching.

| A putative role of microbial N 2 fixation in coral bleaching
While the exact mechanism triggering symbiont expulsion requires further clarification, our findings do not contradict the prevailing bleaching theories, but rather extend our current understanding.
Hence, we here propose a mechanistic concept integrating the observed detrimental role of stimulated N 2 fixation activity into the existing model(s) of (thermal) bleaching (Figure 3). This extended model posits that high temperatures (heat stress) or elevated DOC levels both stimulate nitrogenase activity and diazotroph proliferation, thereby increasing N 2 fixation activity Santos et al., 2014). The increased and preferential uptake of excess (microbially) fixed N releases the resident Symbiodinium population from N limitation, subsequently stimulating nutrientbalanced growth or even shifting algal symbionts to relative P depletion (P starvation). Although the mechanism of symbiont expulsion manifested as coral bleaching remains unknown at this point, here we demonstrate that the disruption of N limitation alone can rapidly result in the loss of algal symbionts. As the present study was not confounded by heat and light stress, the reported effects will likely be dramatically pronounced under these conditions.
Excess N supply from stimulated N 2 fixation could ultimately induce P starvation in Symbiodinium. Such shifts in the N:P ratio promote alterations in the symbiont's thylakoid membrane composition, increasing its susceptibility to photodamage. Simultaneously, the disruption of N limitation of Symbiodinium decouples the tight nutrient exchange relationship with the coral host (Dubinsky & Jokiel, 1994).
As Symbiodinium will subsequently retain and channel most of their photosynthates into their own cell growth and repair, the coral host would be deprived of its main energy source. The resulting energy limitation of host CCMs would cause CO 2 limitation of photosynthetic dark reactions in Symbiodinium, thereby increasing their susceptibility to photodamage (Wooldridge, 2013). The consequential photosynthetic impairment and subsequent overproduction of ROS would cause further damage to the PSII and result in oxidative stress of both Symbiodinium and host cells (Weis, 2008).
Based on the strong increase in N:P ratios in the coral tissue in spite of the increase in d 15 N, we hypothesize that the coral host simultaneously attempts to restore a stable nutrient exchange relationship by altering the nutrient supply to Symbiodinium. This could be achieved either by removal, assimilation, or storage of N derivatives in host cells or organelles, or by the upregulation of other microbial N cycling pathways (nitrification, denitrification) coupled with increased P uptake and translocation to Symbiodinium (Ezzat et al., 2016;Pernice et al., 2012;R€ adecker et al., 2015).
Noteworthy, the present study was conducted in the absence of additional light or heat stress. Hence, the consequences of altered nutrient cycling would have likely resulted in a more pronounced stress response under these conditions. Ultimately, the threshold at which coral bleaching occurs likely depends on whether the intensity and duration of environmental stress exceed F I G U R E 3 Conceptual model of the role of coral-associated dinitrogen (N 2 ) fixation in coral bleaching. High seawater temperatures or dissolved organic carbon (DOC) levels stimulate N 2 fixation activity in the holobiont. Excess fixed nitrogen (N) is rapidly taken up by Symbiodinium, inducing nutrient-balanced growth or even phosphorus (P) starvation. This P starvation would alter the composition of algal photosynthetic membranes, causing the photosystem II (PSII) of the photosynthetic apparatus to malfunction. Under heat stress, damage to PSII would promote the increased production of reactive oxygen species (ROS) in Symbiodinium. Simultaneously, the disruption of N limitation would increase the retention of fixed carbon by algal symbionts, forcing the coral host to deplete its own energy reserves. This could increase the susceptibility to photodamage due to a failure of host carbon concentration mechanisms (CCMs) causing CO 2 limitation of photosynthetic dark reactions. To restore control over the symbiosis, we hypothesize the coral host would attempt to increase its P uptake and/or to retain N from Symbiodinium. [Colour figure can be viewed at wileyonlinelibrary.com] the energetic capability of the coral host to maintain the N limitation of Symbiodinium.

| Ecological relevance of elevated DOC levels on coral reefs
Coral reefs can be regionally exposed to periodically changing levels of TOC/DOC and may range from low (~30 to 70 lM; Haas et al., 2016; on average 130 lM for ambient Red Sea water in the present study) to periodically high levels (as observed for some parts of the Caribbean and the Florida Keys; >1,000 lM; Kline et al., 2006;up to 1,600 lM;Boyer, Fourqurean, & Jones, 1997). The DOC enrichment in the present study achieved a more than tenfold increase (868-1,609 lM) relative to the untreated ambient control (117-154 lM) and therefore constitutes an ecologically relevant enrichment level at an order of magnitude increased in comparison with levels reported for coral reefs. It is worthwhile to note that biological replicates were supplied from a common reservoir of DOC enriched water, in order to exclude confounding effects from differences in DOC enrichment or degradation. This was necessary as labile DOC is rapidly consumed in an aquaria setup (Haas, Al-Zibdah, & Wild, 2009). A potential carry-over effect between corals and water coming from any of the aquaria was minimized via filtration, the use of protein skimmers, and a high renewal rate of seawater in the tanks. While we cannot positively exclude that coral microbiomes were affected by surrounding colonies (something also possible in the reef environment; see Roder, Bayer, Aranda, Kruse, & Voolstra, 2015), possible effects are assumingly minor in relation to the treatment effect. This is supported by the notion that microbial community compositions were maintained throughout the course of the experiment on the level of (1) replicate coral colonies, (2) between the control and treatment, and (3) that the seawater N and P content showed no differences between control and treatment over time. Nonetheless, the DOC enrichment caused a rapid significant shift in the N:P ratios of the two main eukaryotic departments of the P. verrucosa holobiont: the host and the algal symbiont. These changes were likely facilitated by the oligotrophic conditions of the Red Sea water used in this experiment. In a naturally less oligotrophic system, such as the Caribbean, higher DOC levels would likely be necessary to evoke equivalent responses (Kline et al., 2006). Nevertheless, DOC additions in the same order of magnitude as employed in the current experiment induced coral bleaching and mortality in corals from Panama and the Northern Gulf of Aqaba (Haas et al., 2009;Kline et al., 2006). Apart from these environmental factors, the effects of DOC enrichment on coral holobionts also depend on its quality and composition. While a large fraction of DOC in the Caribbean is refractory and of terrestrial origin (Lirman & Fong, 2007;Nebbioso & Piccolo, 2013), the current and previous manipulative studies employed mostly labile DOC sources. As labile DOC is readily available for microbial utilization, its overall effects on the coral holobiont are different from those of refractory DOC. Labile DOC is introduced onto coral reefs from various sources. Municipal sewage and algal exudates, for instance, contain significant proportions of labile DOC of similar saccharide composition as in the present study (Huang et al., 2010;Nelson et al., 2013). Indeed, macroalgae exudates differentially enrich and stimulate bacterial cell growth, favoring the prevalence of opportunistic and potentially pathogenic bacteria, induce coral mortality, and cause shifts toward less efficient copiotrophic reef bacterial communities (Haas et al., 2016;Nelson et al., 2013;Smith et al., 2006).

| Coral reef resilience in a changing ocean
Coral-associated N 2 fixation is increasingly being recognized as beneficial for coral health (R€ adecker et al., 2015) and fundamental for sustaining primary productivity under (seasonally) changing environmental conditions R€ adecker et al., 2015). On the other hand, we show here that diazotroph activity can destabilize the coral-algae symbiosis and thus may pose a threat to overall holobiont functioning. While the current study used DOC enrichment to induce bleaching, our findings may be applicable to thermal bleaching as similar responses of the diazotroph community appear to be in place. Thermal bleaching has long been recognized as one of the most severe threats to modern coral reefs (Hughes et al., 2003). Our findings imply that the ubiquitous presence of diazotrophs in most coral holobionts may pose a threat to corals in a warming ocean. However, similar as in thermal bleaching, changes in the coral-associated N 2 fixation activity and its impact on holobiont functioning will be largely dependent on the environmental (i.e., holobiont) context (Grottoli, Rodrigues, & Palardy, 2006). Reshef, Koren, Loya, Zilber-Rosenberg, and Rosenberg (2006) suggested that a restructuring of the coral microbiome may facilitate the rapid adaptation of coral holobionts to changing environmental conditions (see also Ziegler, Seneca, Yum, Palumbi, & Voolstra, 2017). Therefore, a reduction in diazotroph abundance or activity could potentially enhance the thermal tolerance of corals in a warming ocean. In the long term, however, the coral's ability to evolve may be hampered by its complex mutualistic relationship with Symbiodinium, rendering scenarios likely in which rapid global climate change outpaces the coral's capacity for adaptation (Pandolfi, Connolly, Marshall, & Cohen, 2011).
There may be no rapid solution to reduce the effects of global climate change in the near future. All the more important becomes mitigation, for example, by reducing local anthropogenic stressors, in future conservation efforts. Here, we show that DOC enrichment can rapidly stimulate the N 2 fixation pathway in the coral P. verrucosa. Based on this, we argue that the stimulation of N 2 fixation via sewage and wastewater may be a possible mechanism rendering reef-building corals more susceptible to the effects of global environmental change, particularly heat stress. At the same time, the role of N 2 fixing bacteria for holobiont functioning largely depends on environmental N availability. While helping to sustain productivity during low N availability, stimulated N 2 fixation together with environmental N enrichment may destabilize holobiont functioning. To provide a better understanding of the mechanism POGOREUTZ ET AL.
| 3845 proposed in this study, follow-up work will have to validate and extend the experiments conducted, including other coral species that cover different coral functional groups as well as a range of environmental stressors causing bleaching. Further, the mechanism proposed here may not be applicable to coral bleaching responses induced by factors other than DOC or heat stress (e.g., cold stress-induced bleaching).
The detrimental effects of labile DOC on reef-building corals, however, remain non-negligible (Kuntz et al., 2005;Kline et al., 2006;Smith et al., 2006;Haas et al., 2009;this study). Thus, a priority in local management efforts should be the reduction of DOC input and loading on coral reefs. Sources of DOC enrichment on coral reefs include sewage, wastewater, and excessive algal abundance (Smith et al., 2006;Wear & Thurber, 2015). Consequently, to diminish microbially driven reef degradation processes, management measures would benefit best from combined efforts (Haas et al., 2016;Zaneveld et al., 2016). Specifically, improved wastewater facilities to effectively retain inorganic and organic nutrients coupled with the restoration of herbivorous fish stocks to control for harmful algal growth would likely increase the resilience of corals to ocean warming (Vega Thurber et al., 2012, 2014Wear & Thurber, 2015).
Dieter Peterke for support during sample analysis. Finally, we would like to thank the editor and three anonymous reviewers, whose comments greatly improved the manuscript. The contribution of C.P. was grant Wi 2677/9-1 to C.W.

AUTHOR CONTRIBU TI ONS
CP, NR, AC, CRV, and CW designed research. AG, CRV, CW contributed reagents and tools. CP, NR, AC performed research, and CP and NR analyzed data. CP, NR, CRV, and CW interpreted data. CP, NR, AC, CRV, and CW wrote the manuscript. All authors read and approved the final manuscript.

COMPETI NG FIN ANCIAL INTERESTS
None declared.