Testing assumptions of nitrogen cycling between a temperate, model coral host and its facultative symbiont: symbiotic contributions to dissolved inorganic nitrogen assimilation

Coral symbioses are predicated on the need for mutual nutrient acquisition and translocation between partners. Carbon translocation is well-studied in this classic mutualism, while nitrogen (N) has received comparatively less attention. Quantifying the mechanisms and dynamics of N assimilation is critical to understanding the functional ecology of coral organisms. Given the importance of symbiosis to the coral holobiont, it is important to determine what role photosynthetic symbionts play in N acquisition. We used the facultatively symbiotic temperate coral Astrangia poculata and 15N labeling to test the effects of symbiotic state and trophic status on N acquisition. We tracked assimilation of 2 forms of isotopically labeled dissolved inorganic N (DIN: ammonium, NH4 and nitrate, NO3) by fed and starved colonies of both symbiotic and aposymbiotic A. poculata. Coral holobiont tissue was subsequently analyzed for δ15N and changes in photosynthetic efficiency. Results suggest that corals acquired the most N from DIN via their symbiont Breviolum psygmophilum and that NH4 is more readily assimilated than NO3. Photosynthetic efficiency increased with the addition of NH4, but only for fed, symbiotic treatments. NO3 adversely affected photosynthetic efficiency among starved corals. Our results suggest that symbiosis is advantageous for DIN acquisition, that dysbiosis inhibits corals’ mixotrophic strategy of nutrient acquisition, and that either feeding or symbiosis alone does not fully provide the energetic advantage of both. This study lends support to the emerging hypothesis that symbionts are mutualists in optimal conditions but shift to a parasitic paradigm when resources or energy are scarce.


INTRODUCTION
Mixotrophy, the derivation of nourishment from both autotrophy and heterotrophy, is common among symbiotic marine organisms. Mixotrophy is most typically represented by nutritional mutualisms, enabling multimodal nutrient acquisition in different environmental contexts (Porter 1976, Muscatine & scale due to rising seawater temperatures and other consequences of global change (Gates et al. 1992, Hoegh-Guldberg 1999.
Coral symbioses are predicated on the need for mutual nutrient acquisition and translocation between partners (Muscatine & Porter 1977, Tremblay et al. 2012, 2014. Much like nitrogen (N), the heterotrophic contribution of carbon (C) to the holobiont is also highly variable, ranging from 0−100% with averages in the 40−60% range, and acquisition dynamics can be influenced by species, depth, irradiance, trophic plasticity, food availability, and symbiont density (Palardy et al. 2008, Tremblay et al. 2014, Radice et al. 2019. When dysbiosis occurs, coral hosts become stressed and enter a period of functional starvation (Hoegh-Guldberg 1999). In response, symbiotic hosts either draw upon nutrient reserves (Fitt et al. 1993, Schoepf et al. 2015, Levas et al. 2018 or shift to increased heterotrophy during periods of stress (Palardy et al. 2005, Grottoli et al. 2006, reviewed by Houlbrèque & Ferrier-Pagès 2009). However, these triage behaviors are not sustainable in the long term (Grottoli et al. 2006) as most tropical corals have an obligate relationship with their symbionts wherein up to 95% of autotrophically fixed C is translocated from the symbiont to the host (McCloskey & Muscatine 1984, Davies 1991, Davy et al. 1996, Alamaru et al. 2009, Bednarz et al. 2017.
N is one of the most important nutrients for tissue growth. Corals need N for DNA and protein synthesis (Raymond et al. 2004, Wang & Douglas 1999, yet N is often the limiting nutrient in marine ecosystems (Vitousek & Howarth 1991). Acquisition of N through N fixation by diazotrophs within the coral holobiont is an energetically costly process that is likely fueled by C fixed photosynthetically by Symbiodiniaceae (Cardini et al. 2015). Additionally, the acquisition of dissolved inorganic N (DIN) (both ammonium, NH 4 + , and nitrate, NO 3 − ) is fueled by photosynthates within the symbiont and translocated photosynthates within the host (Kopp et al. 2015, Ezzat et al. 2017. Most evidence suggests NH 4 + is preferred over NO 3 − for both coral host and algal symbiont (Grover et al. 2002, Pernice et al. 2012. Unlike NO 3 − , which requires a minimum concentration to be met for uptake to occur, NH 4 + uptake is independent of concentration, though assimilation and concentration are positively correlated (D'Elia & Webb 1977, Burris 1983, Bythell 1990, Szmant et al. 1990). DIN is assimilated via the glutamine synthetase/glutamine:2 oxoglutarate aminotransferase cycle. Though this enzymatic cycle is commonly thought to be present in both the coral host and symbiont tissue (Yellowlees et al. 1994, Catmull et al. 1987, Wang & Douglas 1999, it appears that symbionts have the greater capability for assimilating DIN (Kopp et al. 2013, Leal et al. 2015, Tanaka et al. 2015, Aranda et al. 2016). However, due to the species-specific nature of coral host−symbiont ecology (Ezzat et al. 2016, Biquand et al. 2017, there is no consensus as to the type of DIN symbiotic corals prefer. The question of what DIN acquisition pathways remain viable against starvation and mortality in corals experiencing dysbiosis as a result of environmental stress is an area of active investigation (e.g. Reynaud et al. 2009, Krueger et al. 2018, Ezzat et al. 2019. There is growing interest in symbiont ecology as to how N is acquired under aposymbiotic (primarily heterotrophic) vs. symbiotic (autotrophic + heterotrophic) states in corals. Facultatively symbiotic systems can offer unique opportunities (Cui et al. 2019) to probe the nature of N acquisition by both autotrophic and heterotrophic pathways independently, in a nonstressed state (Szmant-Froelich & Pilson 1977. Recent work has suggested that N dynamics regulate symbiont growth (Cui et al. 2019) and that N dynamics are context-dependent by depth and heterotrophic activity (Tilstra et al. 2019). Though these studies hint at the mechanism behind N use in corals, experimental manipulation of DIN uptake by autotrophic and heterotrophic corals is needed to fully test the hypothesis that coral symbionts control DIN availability or that DIN availability may influence symbiont performance (photosynthetic efficiency).
In this study, we used the facultatively symbiotic coral Astrangia poculata to examine interactions between coral symbiotic state, food availability, and DIN uptake. A. poculata facultatively associates with the endosymbiotic dinoflagellate Breviolum psygmophilum, formerly genus Symbiodinium, Clade B2 (LaJeunesse et al. 2012(LaJeunesse et al. , 2018. Symbiotic states have previously been assigned according to color, approximate chlorophyll concentration (Dimond & Carrington 2008, DeFilippo et al. 2016, Sharp et al. 2017, Burmester et al. 2018, and cell density of Symbiodinium in summer months, with brown-colony Symbiodinium densities exceeding 106 cells cm −2 coral tissue, and white-colony Symbiodinium densities ranging from 104−106 cells cm −2 (Sharp et al. 2017). Aposymbiotic colonies are likely more reliant on heterotrophy for energy and nutrition than their symbiotic counterparts. We used a 15 N tracer experiment to identify the forms of DIN (NH 4 + or NO 3 − ) that can be taken up by symbiotic and aposymbiotic A. poculata colonies in both starved and fed treatments to manip-ulate holobiont need for DIN. Additionally, we measured related impacts on symbiont performance (i.e. photosynthetic efficiency) and host performance (i.e. protein production). This experiment answers 4 key mechanistic, ecologically relevant functional questions: (1) Does the natural δ 15 N signature differ between symbiotic and aposymbiotic colonies of wildcollected A. poculata, indicative of different sources of colony N? (2) Is NH 4 + or NO 3 − more readily assimilated by the host, symbiont, or both? (3) Is there a compensatory nutrient-uptake mechanism in the absence of heterotrophy? (4) Is enhanced DIN assimilation associated with more photosynthesis?

Experimental overview
Five trials of the same experiment were conducted, with 36 corals trial −1 in August and September 2018 for a total of 180 corals tested. Within each trial, corals were placed into one of 3 possible DIN treatments: 15 NH 4 + enriched artificial seawater (ASW), 15 NO 3 − enriched ASW, or unenriched ASW. Within each treatment, 6 colonies were fed (3 symbiotic and 3 aposymbiotic colonies) and 6 were starved (Fig. 1).

Coral colony collection and husbandry
Astrangia poculata coral colonies were collected at Fort Wetherill State Park in Jamestown, RI (41.478°N, 71.359°W) on 16 July 2018. For collection, divers targeted corals that were clearly symbiotic or aposymbiotic corals (not mixed); 105 colonies of each were selected for the experiments (210 colonies total).
Corals were acclimated and housed in a recirculating closed tank system filled with Instant Ocean ® ASW (Spectrum Brands) and kept on a 12 h light:12 h dark cycle with no nutrient addition. The life support components of the systems include a fluidized biofilter, UV sterilizer, heat exchanger, protein skimmer, and 2 mechanical filter bags to remove detritus. Tank conditions were monitored for light and temperature using a Seneye Reef Monitor; system-wide conditions were monitored by a CA Digital Aquatics Reefkeeper Elite V2. Temperature was maintained at an average of 21.6 ± 0.9°C, mimicking August temperatures in Jamestown, RI (average: 20.9°C; https:// www. seatemperature.org/), which correspond to the time-frame of the experiment. Photosynthetically active radiation (PAR) was maintained at an average of 89.6 ± 10.8 μmol m −2 s −1 during daylight conditions. Prior to experimentation, corals were fed frozen nauplii shrimp twice weekly. Tanks and corals were cleaned as needed to prevent algal overgrowth. Each colony was individually labeled, photographed, and later randomly assigned to a condition (starved, fed), treatment ( 15 NH 4 + , 15 NO 3 − , untreated/negative control), or trial (1−5 temporal replicates).

Experimental preparation
Prior to experimentation and treatment allocation for each trial, 12 corals treatment −1 were each preorganized into 3 groups of 4 (quads) and then allocated across conditions (symbiotic fed, symbiotic starved, aposymbiotic fed, aposymbiotic starved) within each treatment. In this way, corals were selected to be morphologically similar in terms of polyp size and interstitial distance between polyps. Using a Dremel rotary saw with a diamond wheel blade, colonies were cut to have 15−25 polyps quad −1 and allowed at least 12 h of recovery time. All cuts were made between polyps, where there is little/no coenenchyme. For each quad, one of each symbiotic state was assigned to be fed or starved, and the entire group was allocated to a N treatment (control, 15 NH 4 + , or 15 NO 3 − treated seawater; Fig. 1). There were 5 temporal replicates (trials), and each trial consisted of 3 colony replicates for each symbiotic state Three days prior to DIN treatment, corals were starved to minimize interference with metabolic cycles during the experiment. To avoid pseudoreplication, each colony was placed in its own 250 ml capacity jar containing 200 ml ASW. Each jar was randomly assigned to a location within a tank that served as a bath with common temperature and light conditions. Jars remained open on top and were given their own individual air stones for aeration and water circulation. All glassware was pre-cleaned in a 10% v/v HCl acid bath and rinsed thoroughly with deionized water. At the onset of each experiment, each jar was enriched with either ≥ 99% 15 NH 4 Cl or Na 15 NO 3 (Sigma Aldrich) for a final concentration of 5 μM of respective enriched DIN jar −1 or left unenriched. We elected to amend to 5 μM as this concentration is typical of DIN concentrations observed near the ocean−estuary interface in Narragansett Bay (Oviatt et al. 2017 . Air stones were turned on for 5 min to circulate treatments and then were briefly turned off for feeding (including for starved corals). Fed corals were fed live Artemia brine shrimp once daily in still water for 10 min and then the air stones were turned back on. The δ 15 N value of our Artemia was 20.86 ‰. Corals underwent a complete water change (thereby refreshing the labeled DIN treatment) every 24 h over a total treatment time of 3 d, with new labeled DIN added each day to ensure sufficient supply. The DIN enrichment and experimental duration used was similar in length and intensity to other DIN pulse studies in corals (Grover et al. 2003, Tanaka et al. 2006, Pernice et al. 2012, Kopp et al. 2013, Ezzat et al. 2015. After 3 d of treatment, coral tissue was harvested for analysis. A subset of each colony (2−3 polyps each) was flash-frozen in liquid N for protein analysis. The remaining coral tissue was rinsed thoroughly with standard seawater and removed from the skeleton with a Waterpik ® yielding coral tissue slurry, which was frozen at −4°C. Samples were later defrosted and centrifuged at 10 000 RCF for 5 min to separate tissue from seawater. The supernatant was decanted, and the coral tissue pellet was dried overnight or until a constant weight was achieved. Dry holobiont tissue was then crushed into a powder and used for stable isotope analysis.

Positive control
A positive control test used the green algae Ulva lactuca, collected at low tide in September 2018 in Salem, MA (42.537°N, 70.867°W) to confirm that bio logical uptake of the enriched DIN would be successful in our experimental setup. Algal controls were manually cleaned of epiphytes, laid out flat, and cut into 45 pieces (6.5 cm 2 each) from undamaged areas. Each square was placed into a treatment jar with 200 ml of ASW; 15 jars were treated with 5 μM 15 NH 4 Cl, 15 with 5 μM Na 15 NO 3 , and 15 received no amendment. All samples were treated with the same procedure as previous trials, with 2 exceptions: air stones were not used and samples were not fed Artemia. After 3 d, samples were rinsed thoroughly, dried, ground, and submitted for stable isotope analysis.

Measuring photosynthetic efficiency
Relative rates of photosynthesis for each treatment at the beginning and end of each experiment were estimated using pulse-amplitude modulated (PAM) fluorometry. A Walz Junior PAM meter was used to measure chlorophyll activity as a proxy for photosynthetic efficiency (F v /F m ). Corals were darkacclimated for 1 h prior to measurements being taken. Three polyps per colony were measured and the mean F v /F m for that quad was calculated. Initial PAM measurements were taken immediately before DIN treatments were administered. Final PAM measurements were taken after the second full day of δ 15 N exposure (61.10 h).

Stable isotope and elemental analyses
Tissue from Trials 3, 4, and 5 was analyzed for δ 15 N (the ratio of heavy to light isotopes) in the Boston University Stable Isotope Lab (https:// www. bu. edu/ sil/). This analysis also generated δ 13 C, %N (by mass), %C (by mass), and C:N ratios (by mass). Samples were dried to a constant weight, encapsulated in 5 × 9 mm tin capsules (Costech Analytical Technologies), and then combusted using a Eurovector CN analyzer connected to a continuous flow GV Instruments Isoprime Isotope Ratio Mass Spectrometer. External precision standards were 0.2 ‰ for 15 N and 13 C, and all data is reported in per mille (‰) values. Isotope ratios are calibrated to international standards. For 13 CV-PDB the gas is calibrated against NBS 20 (Solnhofen Limestone). For 15 N air the gas is calibrated against atmospheric N 2 and IAEA standards N-1, N-2, and N-3 (all are ammonium sulfate standards). Randomly selected technical replicates were conducted on 5 samples to ensure precision, and all were within the min./max. range of other individuals in the same treatment with one exception, which was subsequently excluded from the analysis. Two 15 NO 3 − treated samples with unusually high δ 15 N values due to carryover from previous samples (a known potential issue when processing enriched samples) were considered methodological artefacts and thus eliminated from data analysis.

Protein analyses
Protein concentrations were determined using Brad ford protein assay on previously frozen coral tissue. Thawed tissue was removed from the skeleton with a Waterpik ® , standardized by volume, and homogenized via sonification. Absorbance at 595 nm for 3 replicates of 24 samples as well as 3 replicates of a serial dilution of BSA (a known protein concentrate) were measured on a single 96 well microplate with a BioTek Synergy H1 microplate reader. A total of 5 plates were used to run all 205 samples. The mean of each of the 3 replicates was calculated via a standard curve based on the albumin standard. The skeleton of each coral fragment was labeled and photographed with a size standard, then used to determine polyp surface area using ImageJ. Final protein concentration was determined as a unit of volume polyp surface area (μl cm −2 ).

Statistical approach
To compare between treatment groups, we used a mixed model approach using R statistical software (R Core Team 2014) and the 'lme4' package (Bates et al. 2015). To begin, we determined whether the δ 15 N and protein content data best fit a normal, lognormal, or gamma distribution using the 'fitdistrplus' package (Delignette-Muller & Dutang 2015, Table S1 in the Supplement at www.int-res.com/ articles/ suppl/ m670 p061_supp.pdf). Following determination of the distribution that best described each data set, we created 2 models. One model contained feeding treatment, symbiotic state, and DIN addition as interacting fixed effects, and the other contained these same fixed effects with the addition of trial as a random effect. In cases where data was not normally dis-tributed, we either log transformed it or shifted it so all points were positive and informed the model the data had a gamma distribution. We then compared the 2 models with and without trial as a random effect using Akaike's information criterion (AIC). In all cases, the simpler model that did not include the random effect had a lower AIC score, indicating trial was not an important effect. We continued the rest of our analysis without including trial as a random effect in our models. We used least square means tests via the 'emmeans' package (Lenth 2018) to investigate pair-wise comparisons between treatments, allowing us to incorporate potential interactive effects in our comparisons ( Table S2 in the Supplement). We re peated this approach with data subset by 15 N ad dition type, creating models that included only feeding treatment and symbiotic state in order to compare how these processes changed within each N addition treatment (Tables S3 & S4).
Paired t-tests were conducted using Systat (v.11) and GraphPad Prism (v.8.1.0) to test whether photosynthetic efficiency changed throughout the course of the experiment for each treatment type and were Bonferroni-corrected to avoid potential Type II statistical error. See Table S5 in the Supplement for significance test results. We used linear regressions to test for relationships between DIN assimilation of each DIN treatment and change in photosynthetic efficiency.

Data availability
All data used in this study is available in the Dryad digital repository and can be accessed via https:// doi. org/ 10.5061/dryad.brv15dv9q.

Coral protein content
There was a significant difference between symbiotic and aposymbiotic corals across all treatments (p = 0.013; Table S2), most likely driven by the increased protein observed in symbi-otic starved corals compared to aposymbiotic fed corals in the 15 NH 4 + enriched seawater treatment (p = 0.018; Fig. S1, Table S4 in the Supplement).

Other nutritional variables
Measurements of C:N, δ 13 C, and %N showed no significant differences for any variables (Figs. S2, S3 & S5 in the Supplement). Symbiotic fed corals in the 15 NH 4 + treatment showed a greater %C than aposymbiotic fed corals in the same treatment (Fig. S4). Means and standard errors are shown in Table S6 in the Supplement.

Coral photosynthetic efficiency
Within the 15 NH 4 + treatment ( Fig. 4 top row), symbiotic fed corals showed a significant increase in photosynthetic efficiency over the course of the 3 d treatment (p = 0.004; Table S5 in the Supplement). Symbiotic starved corals showed a trend to wards increased photosynthetic efficiency (p = 0.060; Table S5). There were no significant differences in 15 NH 4 + treated aposymbiotic fed (p = 0.548; Table S5) and 15 NH 4 + treated aposymbiotic starved corals (p = 0.702; Table S5).
In the NO 3 − treatment (Fig. 4 middle row), symbiotic starved corals showed a significant decrease in photosynthetic efficiency over the course of the 3 d treatment (p = 0.034; Table S5). Aposymbiotic starved corals showed a declining trend in photosynthetic efficiency (p = 0.088; Table S5), whereas symbiotic (p = 0.978; Table S5) and aposymbiotic fed corals showed no significant changes over the course of treatment (p = 0.168; Untreated/negative control starved corals (Fig. 4  bottom row) showed a significant increase in photosynthetic efficiency over the course of the 3 d treatment in both symbiotic (p < 0.001; Table S5) and aposymbiotic (p = 0.008; Table S5) corals. Untreated symbiotic fed corals (p = 0.649; Table S5) and untreated aposymbiotic fed corals (p = 0.981; Table S5) showed no significant changes over the course of the 3 d treatment time.

DISCUSSION
The aim of this study was to explore nutrient acquisition pathways in corals and their symbionts under several N-availability, symbiotic state, and heterotrophy/autotrophy scenarios. This experiment answers 4 key mechanistic, ecologically relevant functional questions: (1) natural abundance δ 15 N values are higher in aposymbiotic colonies of wild-collected Astrangia poculata than in symbiotic colonies; (2) NH 4 + is more readily assimilated than NO 3 − by both symbiotic and aposymbiotic colonies; (3) there is no evidence of compensatory nutrient uptake in the absence of heterotrophy, instead, fed corals are more able to absorb and utilize N; and (4) enhanced DIN assimilation is associated with greater photosynthetic efficiency for NH 4 + . Our analysis of A. poculata suggests that NH 4 + is the preferred DIN source and the symbiont, Breviolum psygmophilum, is the major N-assimilator, similar to other isotope enrichment studies. The mechanism underlying this response could be the symbiont mediating the host or direct symbiont assimilation, but the holobiont clearly takes up more NH 4 + regardless of trophic condition. The major contribution of our study is the insight into heterotrophic versus autotrophic dynamics. We observed that food availability influenced DIN uptake in the 15 NH 4 + treatment where fed corals showed greater DIN uptake than their starved counterparts regardless of symbiotic state, and in the 15 NO 3 − treatment where symbiotic fed corals showed greater DIN uptake than symbiotic starved corals. The behavior of the symbiont in response to different DIN and food availability treatments suggests different energetic and metabolic tradeoffs whereby symbiosis can provide mutualistic benefits, or parasitic costs, to hosts under different scenarios. Our study lends support to the emerging hypothesis that symbionts are mutualists in optimal conditions in which photosynthates derived by the symbiont are shared with the host but shift to para-   (Tables S2 &S3). n = 15 for each treatment, except n = 14 for fed symbiotic colonies and starved aposymbiotic colonies treated with nitrate sitic nutrition when scarcity of resources or energy causes the symbionts to retain their photosynthates, forcing the host to rely on stored lipids for energy in order to maintain their symbiont population (Szmant-Froelich & Pilson 1980, Baker et al. 2018). The dynamics of DIN assimilation during enrichment is fundamentally different than what is found in the natural environment. Experimental corals that were not treated with enriched DIN were used to approximate natural abundances of 15 N in wild facultatively symbiotic corals. We observed that aposymbiotic corals had a higher ‰ δ 15 N value than symbiotic corals. This supports either the assumption that aposymbiotic corals are more dependent on heterotrophy than their symbiotic counterparts due to the preferential storage of 15 N in tissue resulting in a bioaccumulation effect (DeNiro & Epstein 1978, Peterson & Fry 1987 or the theory that internal recycling of 14 N back and forth between the coral host tissue and symbiont makes heterotrophically derived N less necessary and drives the δ 15 N value down (Muscatine & D'Elia 1978, Reynaud et al. 2009, Bednarz et al. 2017. It is reasonable to expect that temperate corals exposed to seasonal shifts in temperature, light, and nutrient availability may obtain N differently than corals found in warm, stable, oligotrophic tropical waters. For example, Cladocora caespitosa is a symbiotic temperate Mediterranean coral with a high rate of heterotrophic feeding that is exposed to shifting seasonal conditions with corresponding variability of δ 15 N and δ 13 C suggesting reliance on heterotrophy in the winter (Ferrier-Pagès et al. 2011). Oculina arbuscula, a temperate facultatively  . The box extends from the 25th to the 75th percentiles; centerline: median; whiskers: min. to max. values. Individual colony response is plotted as points overlaying the box plots, showing the difference between initial and final time points for symbiotic or aposymbiotic, fed or starved corals. Heavy-outlined shaded boxes denote significant differences (p ≤ 0.05) following a paired t-test (Table S5) symbiotic coral, exhibited δ 15 N levels indicative of autotrophic assimilation of DIN regardless of season for both symbiotic and aposymbiotic colonies (Leal et al. 2014). Thus, differences in the assimilation of DIN by temperate corals over seasons is not fully reconciled and may be more complex than previously appreciated. Although our experimental A. poculata were able to assimilate DIN from the environment in both the 15 NH 4 + and 15 NO 3 − treatments, 15 NH 4 + was incorporated noticeably more than 15 NO 3 − in symbiotic fed corals. Like existing literature (Grover et al. 2002, Pernice et al. 2012, Kopp et al. 2013, Tanaka et al. 2015, our 15 NH 4 + trials indicate that symbionts are most efficient at NH 4 + assimilation, further substantiating the important role they play in DIN acquisition (Tanaka et al. 2015, Bednarz et al. 2017. Additional studies have shown that Symbiodinium microadriaticum have high numbers of NH 4 + transporter domains (genes that promote C and N transport and provisioning), which likely help in the translocation of nutrients (Aranda et al. 2016). In contrast, Acropora palmata and its associated symbionts assimilate both NH 4 + and NO 3 − and exhibit diurnal variation in NO 3 − uptake, with daytime NO 3 − assimilation rates higher than NH 4 + (Bythell 1990). Together, these studies suggest that DIN uptake and DIN assimilation dynamics are not universal and may be highly dependent on coral and symbiont species.
We found that NO 3 − uptake differed from NH 4 + uptake in 2 ways: it was lower in general, and only symbiotic fed corals showed any significant uptake. These results indicate either energetic costs or time delays associated with reducing NO 3 − to NH 4 + so that it may be taken up by glutamine synthetase enzymes. This observation is similar to other studies in which energetically costly NO 3 − reduction and uptake has been shown to happen only after available NH 4 + has been depleted (Berges at al. 1995) and that time delays of 15−30 min have been shown to occur between NO 3 − introduction and uptake by coral symbionts (Kopp et al. 2015). Furthermore, it has been suggested that only symbionts, and not the coral host, have the ability to reduce NO 3 − to NH 4 + ( Grover et al. 2003). If NO 3 − reduction costs the coral holobiont energy, colonies not able to acquire energy from both heterotrophy (i.e. Artemia) and autotrophy (photosynthates) might not have had enough available energy for that process. As such, symbiont assimilation of NO 3 − may only occur under nearly ideal conditions when the host is able to fully meet the energetic requirements of the symbiont for photosynthesis.
If DIN uptake does benefit the symbiont, an increase in photosynthetic efficiency would be expected. Our results show direct evidence for increased photosynthetic efficiency in symbiotic, fed corals in response to 15 NH 4 + enriched water, suggesting that symbionts react positively to NH 4 + availability with an immediate photosynthetic payoff. Correspondingly, our results also clearly show a decrease in photosynthetic efficiency when exposed to 15 NO 3 − enriched water in symbiotic starved corals, which provides insight into the energetic costs of NO 3 − assimilation by the symbiont. Starved corals may not be able to provide their symbionts with the energy necessary to reduce NO 3 − into a usable form (i.e. NH 4 + ) (Ezzat et al. 2019). The coral host may even start to parasitize energy from the symbiont when starved due to the lack of mutualistic benefit as well as the extra energy expenditure of NO 3 − reduction. While it is possible that endolithic algae within the coral skeleton may also contribute to holobiont C fixation, the extent of this contribution is still relatively unknown and likely minor compared to the contribution from Symbiodiniaceae (Shashar & Stambler 1992, Schlichter et al. 1995, Shashar et al. 1997, Fine & Loya 2002, Sangsawang et al. 2017. Further evidence for energetic limitation in A. poculata was observed in the 15 NH 4 + trials, where aposymbiotic corals with few symbionts were only successful at assimilating 15 NH 4 + when fed. In previous studies, aposymbiotic A. poculata corals were able to heal wounded surface areas more effectively when fed, and starved aposymbiotic corals were never able to fully recover from wounding (Burmester et al. 2018). We found that starved, aposymbiotic corals were unable to assimilate NH 4 + , suggesting that environmental nutrients are inaccessible by the holobiont when there is no autotrophic and heterotrophic energy. This suggestion is further supported by our observations that there were no significant differences in protein levels in the control or 15 NO 3 − treatments, likely because excess N was not available (control) or not accessible ( 15 NO 3 − ). However, in the 15 NH 4 + treatment, symbiotic starved colonies had the highest protein levels across all treatments. As such, A. poculata holobionts may upregulate their protein production via symbiont assimilation of DIN in response to host starvation stress. This effect could be compounded by increased protein retention via host−symbiont nutrient translocation, and future studies could further investigate the role of N availability in protein production in this and other facultatively symbiotic systems. It is likely that in the absence of any food source (heterotrophic or autotrophic), aposymbiotic corals undergo initial stages of nutritional quiescence (dormancy), experiencing polyp withdrawal and a general cessation of movement and behavior (R. D. Rotjan & A. Breef-Pilz pers. obs.). As such, these corals would likely be unable to perform the enzymatic processes to assimilate DIN. Nutritional quiescence may be similar to temperature-induced, cold-water quiescence (Grace 2017, Sharp et al. 2017, wherein corals stop exhibiting polyp extension behavior in winter temperatures. Although our enriched DIN experiments were shortterm (3 d), the observed changes in protein might reflect an organismal response to starvation stress in the presence of available N, but more research is needed.
Measurements of photosynthetic efficiency are often used as a proxy for symbiont population growth and health (Beer et al. 1998, Ralph et al. 1999, Warner et al. 1999). In the context of coral bleaching, there is some evidence that stress can cause one mutualistic partner to turn parasitic towards another, further contributing to dysbiosis events (Baker et al. 2018). Except for our 15 NH 4 + enrichment treatments, where DIN was readily assimilated, photosynthetic efficiency differed the most between symbiotic and aposymbiotic corals in starved treatments (for both 15 NO 3 − and control), which may indicate that the stress of starvation catalyzes a switch from mutualism to parasitism. In A. poculata, the hosts rely heavily on heterotrophy for nutrient acquisition (Szmant-Froelich & Pilson 1980). Under increased metabolic demand and starvation stress, symbionts may be unable to photosynthesize as efficiently (as in the case of energetically costly NO 3 − reduction), functionally depriving the host of energetic resources to meet its own energetic requirements. Symbionts may also have the ability to increase their photosynthetic efficiency to compensate for a lack of food/nutrition even when no additional DIN is available as long as there is no energetically prohibitive process occurring. This can be seen in the control groups for symbiotic starved and aposymbiotic starved corals, which each saw a significant increase in photosyn-thetic efficiency. Too much DIN in the environment can cause excessive symbiont cell growth, which may deprive corals of nutrition (Lesser et al. 2013), and can lead to phosphorus limitation within corals, reducing stress tolerance and photosynthetic efficiency (D'Angelo & Wiedenmann 2014, Ezzat et al. 2015). Increased symbiont growth can also lead to a decrease in calcification as more C is kept in the symbiont and not shared with the host, reducing the availability of C to make the calcium carbonate skeleton (Reynaud et al. 2003). To prevent this from happening, corals have mechanisms to self-regulate symbiont growth rates and concentrations by withholding excess nutrients from their symbionts (Rees 1991, D'Angelo & Wiedenmann 2014. Global sea surface temperature changes are driving a major ecological crisis for coral reefs, causing bleaching events and mortality. It is well established that coral photosynthetic endosymbionts can fix C and are the driving force behind C acquisition for corals (Muscatine & Cernichiari 1969, Muscatine & Porter 1977, Erez 1978, Gattuso et al. 1999). However, dysbiosis not only deprives corals of C but also of N and other essential nutrients (Miller & Yellowlees 1989). The emerging finding of this study is that symbionts, either directly or indirectly, are the driving force behind successful N acquisition. In nature, aposymbiotic temperate corals likely only thrive because they exist in a nutrient-rich environment with abundant particulate and dissolved organic matter for consumption. Despite being mixotrophic, our study suggests that within the coral holobiont, the autotrophic contribution is extremely valuable even in highly heterotrophic species. Further studies into host regulation of N cycling could be key to understanding the full extent of the effects of bleaching on coral nutrient cycling.
Acknowledgements. The authors are grateful to the Boston University Initiative on Cities for seed funding that led to this work. Authors are additionally thankful to Robert Michener and the BU stable isotope facility for analysis. Authors are also grateful to Sean Mullen and Adrien Finzi for helpful discussions related to this work, and to Cara Johnson for husbandry assistance. Authors are further grateful to members of the Rotjan Lab, past and present, for their assistance, helpful discussions, and support.