Productivity and carbon fluxes depend on species and symbiont density in soft coral symbioses

Soft corals often constitute one of the major benthic groups of coral reefs. Although they have been documented to outcompete reef-building corals following environmental disturbances, their physiological performance and thus their functional importance in reefs are still poorly understood. In particular, the acclimatization to depth of soft corals harboring dinoflagellate symbionts and the metabolic interactions between these two partners have received little attention. We performed stable isotope tracer experiments on two soft coral species (Litophyton sp. and Rhytisma fulvum fulvum) from shallow and upper mesophotic Red Sea coral reefs to quantify the acquisition and allocation of autotrophic carbon within the symbiotic association. Carbon acquisition and respiration measurements distinguish Litophyton sp. as mainly autotrophic and Rhytisma fulvum fulvum as rather heterotrophic species. In both species, carbon acquisition was constant at the two investigated depths. This is a major difference from scleractinian corals, whose carbon acquisition decreases with depth. In addition, carbon acquisition and photosynthate translocation to the host decreased with an increase in symbiont density, suggesting that nutrient provision to octocoral symbionts can quickly become a limiting factor of their productivity. These findings improve our understanding of the biology of soft corals at the organism-scale and further highlight the need to investigate how their nutrition will be affected under changing environmental conditions.

Autotrophic carbon acquisition, estimated using 13 C-bicarbonate incorporation over 5 h under maximal irradiance (P C , expressed in µg C g −1 AFDW h −1 ), was significantly different between the two coral species, with Litophyton sp. exhibiting higher Pc rates than R. f. fulvum (Fig. 1c, Table S1). When normalized to symbiont cells, Pc rates were significantly different between depths, with higher values obtained in mesophotic conditions (Fig. 1d, Table S1). Respiration rates (Rc, expressed in µg C g −1 AFDW h −1 ) were similar between depths for R. f. fulvum (347 ± 45 and 489 ± 47 in shallow and mesophotic conditions, respectively), whereas they were two-fold lower for mesophotic (148 ± 47) than shallow (406 ± 18) nubbins of Litophyton sp. (Table S1, Tukey HSD p < 0.01). While symbiont respiration accounted for less than 1% of the holobiont respiration for Litophyton sp. in both depths, symbiont respiration accounted ca. 6% and 4% to holobiont respiration for mesophotic and shallow R. f. fulvum, respectively. Carbon budget. After the pulse period (5 h), both species exhibited significant differences in their carbon budget between depths ( Fig. 2a-d, Tables S1 and S2). Overall, rates of photosynthate translocation were lower in Litophyton sp. at both depths (36% to 60% of the total fixed carbon) compared to R. f. fulvum (85%), and symbionts of Litophyton sp. retained four to five times more carbon than those of R. f. fulvum ( Fig. 2b-d , Table S1). However, no difference was observed in the rate of carbon assimilated by the host between depths and species (134-150 µg C g −1 AFDW h −1 ). For Litophyton sp. from mesophotic depth, symbionts translocated less carbon to the host (36% versus 60%) and less carbon was lost by the holobiont (19% versus 44%) as compared to nubbins from shallow depth. On the contrary, photosynthate translocation (85%) and carbon loss (64-70%) were comparable between shallow and mesophotic R. f. fulvum corals.
After the chase period (24 h), the carbon budget of both species was clearly species-specific but without depth-specific differences ( Fig. 3a-d, Table S1). In Litophyton sp., symbiont cells assimilated half of the photosynthesized carbon and translocated only 47% to the coral host in both shallow and mesophotic colonies ( Fig. 3a-d, Table S1). On the contrary, symbionts in R. f. fulvum translocated 86-92% of the photosynthetically fixed carbon to the host and assimilated only 4-11% of this carbon for their own energetic needs. Consequently, significant differences were evident between the two species for the rates of carbon 1) translocated (up to one third lower in Litophyton sp. than in R. f. fulvum), 2) retained in the symbionts (ten-fold higher in Litophyton sp. than in R. f. fulvum) and 3) lost from the symbiotic association (two-fold lower in Litophyton sp. than in R. f. fulvum) (Fig. 3a-d, Table S1).

Relationship between Symbiodiniaceae density and carbon flux.
Our results show that in the two coral species at both depths the carbon acquisition per symbiont cell exponentially decreased with increasing symbiont density (Fig. 4a). In addition, we found a positive linear correlation between symbiont cell carbon acquisition and translocation (Fig. 4b).

Discussion
This study shows that the functioning of the coral-dinoflagellate symbiosis in soft corals holds some different characteristics to that in many scleractinian corals, and that the two groups are differently acclimatized to depth. The two investigated soft coral species maintained an equivalent carbon acquisition from shallow to poorly-lit mesophotic habitats, while carbon acquisition of scleractinian corals generally decreases with depth. This autotrophic efficiency, combined with putative high heterotrophic capacities, may provide soft corals with an ecological advantage in the upper Red Sea mesophotic reefs 10 . In addition, the results show that carbon acquisition and translocation rates per symbiont cell were negatively correlated with the symbiont density within the host tissue, with a higher assimilation of carbon in symbiont biomass under high symbiont density. However, the rates at which carbon was translocated to and assimilated by the host were unaffected, indicating that dense symbiont populations can remain mutualistic.
Our findings indicate that the mean autotrophic carbon acquisition in both soft coral species (measured during 5 h and using 13 C-labelling) remained stable from shallow to mesophotic environments, which suggests that soft corals are either photo-limited in shallow waters or well photo-acclimatized to depth (Fig. 5). This is in agreement with previous observations 13 that maximal photosynthetic rates of several soft coral species from the Great Barrier Reef were measured at 20 m depth rather than in shallower waters (Table 1). Such constant carbon acquisition along the depth gradient is surprising in comparison to the general patterns observed in scleractinian corals. Most scleractinian corals experience a significant decrease in rates of carbon acquisition with depth, although this physiological parameter has been poorly investigated in mesophotic environments 5,30,31 . However, a recent study, which investigated the photosynthetic performance of the mesophotic coral Euphyllia paradivisa, found a high photosynthetic capacity of this coral species under low light conditions 32 .
Pigment content, symbiont density and symbiont genus are generally involved in shaping the photo-acclimatization of the coral symbiosis from shallow to mesophotic environments. Our results highlight for the two soft coral species a host-specific Symbiodiniaceae genus that remained stable along the depth profile up to 40 m depth. Litophyton sp. was associated with Symbiodinium sp., as previously reported (formerly Symbiodinium clade A 33,34 ), while R. f. fulvum was rather associated with Cladocopium sp., in agreement with studies performed in the Red Sea (formerly Symbiodinium clade C 33 ), or the Great Barrier Reef 35,36 . Such clade specificity is assumed to be persistent both in space and time in octocorals 37,38 . Thus, acclimatization of the soft corals to low light levels was independent of the symbiont genus, but potentially driven by higher concentrations of total chlorophyll per AFDW and per symbiont cell in mesophotic corals (Fig. 1b), which maximize the light harvesting capacity of the    www.nature.com/scientificreports www.nature.com/scientificreports/ cells under low light levels 5,39 . In addition, both species exhibited a decreased symbiont density in mesophotic as compared to shallow colonies which is likely another adaptation to depth by reducing self-shading under reduced PAR 40,41 . A lower symbiont density also reduces the competition for inorganic nutrient supply 42 . It has been demonstrated that mesophotic scleractinian corals can display several strategies to acclimatize to low light levels, such as modifications in the organization of photosynthetic apparatus (shift to a PSII-based system, including additional photosynthetic antenna 30 ), modifications in the organization of antenna (growth of specialized paracrystalline light-harvesting antenna domains 43 ), or the synthesis of fluorescent pigments by the host (that can perform wavelength transformation to facilitate light penetration 44 ); however, these strategies remain to be further investigated in soft corals.
The autotrophic carbon acquisition to respiration ratios (P C /R C ) of Litophyton sp. ranged from 1 to 2 from shallow to mesophotic depths, the latter value being above the conservative threshold for net autotrophy (P C / R C > 1.5). Considering that the solar radiation level in November is one of the lowest annual levels 45 , the data suggest that Litophyton sp. is mostly an autotrophic species. On the contrary, P C /R C ratios of R. f. fulvum in November    www.nature.com/scientificreports www.nature.com/scientificreports/ were always below compensation (P C /R C = 1) even when considering a maximal autotrophic carbon acquisition rate sustained for 12 h. This strongly suggests that this species relies, at least in winter, on heterotrophy to sustain its daily respiratory needs. The difference in P C /R C between the two species may be due to their different growth shapes and different micro-morphological features of the polyps. Two recent studies on gorgonian octocorals, indeed highlighted a correlation between host morphology, polyp size and productivity 17,18 , thereby corroborating the first observations of Porter (1976) 46 .
Symbiont cell carbon acquisition was negatively correlated with symbiont density in both investigated soft coral species and decreased exponentially as soon as the symbiont density increased. Although this pattern is first described on soft corals in this study, similar observations were reported with gorgonian octocorals 18 and scleractinian corals 47 , and might be due to a host-dependent regulation of light and nutrient ressources 47 , or a self-shading of the symbionts 42 .
Our results also highlight a positive correlation between carbon acquisition and carbon translocation rates per symbiont cell, similarly to a previous study on scleractinian corals 48 . In Litophyton sp., a low carbon acquisition per cell corresponded to a high symbiont density and consequently, symbiont cells assimilated more carbon for www.nature.com/scientificreports www.nature.com/scientificreports/ their own metabolism rather than translocating it to the host (50% translocation). In contrast, the symbionts in R. f. fulvum translocated significantly more carbon (80 to 90% translocation) to the host, likely due to their lower abundancy and increased rates of carbon acquisition per cell. Carbon translocation rates of >80% have also been reported for other coral species [2][3][4] . For example, Scheufen et al. 47 measured highest productivity rates in corals during the very oligotrophic summer season despite seasonally reduced symbiont density. This may also imply a high carbon translocation to the host that is essential when exogenous food sources are scarce.
Differences in cell-specific carbon assimilation and translocation rates may not only be related to the symbiont density but also to the symbiont genus. Indeed, Symbiodinium sp. (associated with Litophyton sp. here) is known to translocate less carbon to its host as compared to Cladocopium sp. (associated with R. fulvum fulvum here) [48][49][50] . However, an important observation is that the lower percentage of carbon translocation in Litophyton sp. is not indicative of a shift towards symbiont parasitism 50 , as the carbon translocation to the host still exceeded the metabolic costs for holobiont respiration. In R. f. fulvum, due to the low symbiont density, an increased translocation rate per cell was needed to transfer the same amount of carbon to the host. In addition, the percentage of carbon lost by the host was lower in Litophyton sp. than in R. f. fulvum, compensating for the lower percentage of carbon translocated from the symbionts. As a consequence, the rates at which carbon was retained in the host were constant between species and depths (between 13% and 19% of the photosynthesized carbon, or 93 to 138 µg C g −1 AFDW h −1 ). Assimilation rates of 10 to 20% of the photosynthesized carbon by the animal host seems to be a general feature in corals 4,5 .
Soft and scleractinian corals therefore seem to exhibit a different trend of carbon acquisition along the depth gradient, although more investigations with different scleractinian and soft coral species are still needed. In order to compare carbon fixation per biomass between soft and scleractinian corals, we used data previously obtained on the scleractinian coral Stylophora pistillata sampled at the exact same location, depths and month 5 (Table S3) and normalized the rates to AFDW. The estimates we obtained show that soft corals fix less carbon per biomass as compared to S. pistillata at shallow depth in November (estimation of 2426 µg C g −1 AFDW h −1 for S. pistillata versus 554-958 µg C g −1 AFDW h −1 ), although soft corals contain comparable or higher dinoflagellate density 51 and chlorophyll content per biomass (Table S3) than scleractinian corals. However, in deeper environments, soft corals tend to fix carbon per biomass at similar or even higher rates (estimation of 555 µg C g −1 AFDW h −1 for S. pistillata versus 710-829 µg C g −1 AFDW h −1 for R. f. fulvum and Litophyton sp.). This may indicate that in shallow waters, soft corals have particularly less fixed carbon available to maintain or grow their biomass as compared to scleractinian corals. Morphological characteristics of the host, which have an impact on the diffusion and transmission of light to the symbionts, may explain differences in carbon acquisition in shallow water corals. Soft corals lack a calcium carbonate skeleton, contrary to scleractinian corals, and only possess sclerites as calcified structures. However, skeletons present light scattering abilities, which enhance light absorption efficiency of the symbionts (e.g., Enríquez et al.) 52 . Although characteristics of sclerites and tissues remain to be investigated to understand the potential light amplification and dispersion in soft corals, we hypothesized that they would affect the light environment of the symbionts to a lesser extent than skeletons do in scleractinian corals. In addition, soft corals have a thick coenenchyme and exhibit a low ratio of colony surface area to volume, which does not favor light exposure and gas or nutrient exchange through the epidermal tissue 53 . Finally, the variability of their hydroskeleton allows soft corals to contract and expand their tissue. Tissue contraction removes symbionts from the tissue surface, thereby higher irradiance is required to reach the symbionts and to achieve photosynthetic compensation and saturation.
Understanding nutritional ecology of octocorals is still in its infancy. The common belief is that octocorals are more heterotrophic than scleractinian corals, because of their low carbon acquisition in surface waters, and because their morphology is more suited for heterotrophy 7 . Two recent studies on Caribbean gorgonian octocorals have observed a possible correlation between host morphology and symbiont performance 17,18 . Octocorals with thin branches and small polyps can be more autotrophic as compared to those with a massive shape and large polyps. In our work, a different degree of autotrophy was demonstrated for the two investigated soft coral species, which might be linked to morphological traits. The arborescent shape of Litophyton sp. may enhance the exposure of symbionts to light and thus favour autotrophy compared to the mat-forming shape of R. f. fulvum. In addition, our results strongly support the view that high symbiont density results in lower cell-specific carbon translocation and acquisition. However, soft corals with high symbiont densities (Litophyton sp.) reached similar carbon translocation than those with low symbiont density. Further work involving various soft coral species, growth-shapes and symbiont-specificities will be essential to provide more insight into the physiology and trophic ecology of octocorals thriving along environmental gradients.

Material and Methods
Soft coral collection and experimental setup. The study was conducted in November 2017 at the Inter-University Institute for Marine Science (IUI), Gulf of Eilat. Experiments were performed with two soft coral species harboring dinoflagellate symbionts, Litophyton sp. (Forskål, 1775) and Rhytisma fulvum fulvum (Forskål, 1775) belonging to the Alcyoniidae family.
Ten colonies per species were sampled on the reef adjacent to the IUI (29°51′N, 34°94′E) at 8 m and 40 m depth, respectively, to generate a total of 40 nubbins per species (two nubbins per colony). Nubbins were allowed to recover for ten days in open-water cages placed at the original collection site before being brought back to the Red Sea Simulator facility 54 . There, nubbins were allocated to three different outside aquaria per species and depth. All aquaria were continuously supplied with water directly pumped from the reef and were exposed to the natural diurnal light cycle. Furthermore, aquaria were shaded with several layers of mesh clothes to adjust daily maximum light levels to the maximum irradiance measured at the corresponding depth. Thus, all nubbins were maintained under natural diurnal variation in irradiance and received the depth-corresponding in situ maximum irradiance at noon. Levels of photosynthetically active radiation (PAR) were obtained from the Israel National Monitoring program of the Gulf of Eilat (http://www.iui-eilat.ac.il/Research/NMPMeteoData.aspx). During the five days of our experiment, surface water PAR reached 1200 µmoles photon m −2 s −1 at midday (Fig. S1). The PAR received at 10 m and 40 m depth peaked at ca. 550 µmoles photon m −2 s −1 and 50 µmoles photon m −2 s −1 respectively, considering attenuation coefficients of 0.072-0.1 m −1 at this precise location 55,56 . These values are in agreement with one-off measurements performed with a data logger during coral collection. The in situ seawater temperature (24 °C) and nutrient concentrations (<0.5 µM dissolved nitrogen, and <0.2 µM dissolved phosphorus) were stable throughout the investigated depth gradient and period of time (http://www.iui-eilat.ac.il/Research/NMPMeteoData.aspx). Coral nubbins were maintained for one day under these conditions before the following measurements were performed in the outside aquaria under exactly the same environmental conditions. naH 13 co 3 incubations. The rates at which carbon was fixed by the shallow and mesophotic coral colonies under their natural PAR levels were estimated using 13 C-labelled bicarbonate according to Tremblay et al. 4 . To take into account variation in PAR levels during the day, corals were incubated for 5 h, between 10 am and 3 pm, to cover the maximal daily irradiance 45 (Fig. S1). For each species and depth condition, 10 coral nubbins, from 10 different colonies were placed in individual beakers filled with 200 mL FSW enriched with 0.6 mM NaH 13 CO 3 (98 atom % 13 C, #372382, Sigma-Aldrich, St-Louis, MO, USA). After this "pulse" period of 5 h, half of the corals were sampled and stored frozen at −20 °C until further analysis (T 0 ). The other half was transferred into 200 mL of non-enriched FSW for a chase period of 19 h (T 24 ). The determination of % 13 C enrichment, and total carbon content in the symbionts and host compartments were performed with a Delta plus Mass spectrometer coupled to a C/N analyser (Thermo Fischer Scientific, Bremen, Germany). The natural isotopic abundance of each species at each depth was determined from the corals used for the respiration measurements. Detailed calculations are described in Tremblay et al. 4 . Data were normalized to ash-free dry weight of the organisms 27 .
Physiological and tissue descriptor measurements. Estimation of the dark respiration rates of the whole colony, as well as of the isolated symbionts, was needed for the establishment of the daily carbon budget. Dark respiration rates were measured on four nubbins per species and depth (from four different colonies), as well as on freshly isolated symbionts from four other nubbins according to Tremblay et al. 4 . Dark-acclimated nubbins/symbionts were individually placed in stirred incubation chambers filled with 0.45 µM filtered seawater and maintained at 24 °C. Changes in dissolved oxygen were monitored over 30 minutes using optodes connected to an Oxy-4 (PreSense, Regensburg, Germany). Oxygen fluxes were converted into carbon equivalents (R C ) 4 . The autotrophic carbon acquisition to respiration ratios (P C /R C ) were estimated considering a maximal autotrophic carbon acquisition rate for 6 h, and a half rate for the remaining 6 h. At the end of the incubations, nubbins were flash-frozen in liquid nitrogen, freeze-dried, weighed for the total DW determination and processed as described in Pupier et al. 27 (Supplementary information) for the further determination of the symbiont density, total chlorophyll concentration and ash-free dry weight (AFDW). The genus of the symbionts hosted by the shallow and mesophotic populations of Litophyton sp. and R. f. fulvum was investigated following the protocol of Santos et al. 57 .

Interspecies comparisons.
To compare our data on soft corals, normalized to AFDW, with those found in the literature on scleractinian corals, often normalized to skeletal surface area, it was necessary to use the same normalization metric. As surface area could not be measured with accuracy on soft corals (since they have a highly variable hydroskeleton which expands and shrinks the colony size by several folds depending on the environmental conditions), we normalized all data to AFDW 27 . For this purpose, we chose eight nubbins (skeletal surface area ranging from 4 to 29 cm²) of the scleractinian coral Stylophora pistillata originating from the Red Sea and grown at the Monaco Scientific Center, as it is one of the dominant species in the Gulf of Eilat and many data have been acquired on this species. AFDW was determined as described above after separating the tissue from the skeleton and combusting it. Surface area was determined using the wax technique 58 . A conversion factor (255.65 ± 8.98, Table S4), estimated from the ratio between skeletal surface area (in cm²) and AFDW (in g) of these nubbins, was used to transform data normalized to surface area (Table S3) into data normalized to AFDW. Statistical analyses. Analyses were performed using R software (R Foundation for Statistical Computing). All data were expressed as mean ± standard error. Prior to analyses, outlier values were identified using Grubb's test and were excluded when p-values were significant (p < 0.05). Assumptions of normality and homoscedasticity of variance were evaluated through Shapiro's and Bartlett's tests along with graphical analyses of residuals. A two-way analysis of variance (ANOVA) was performed to test the effect of species and depth on symbiont density, chlorophyll concentrations and carbon acquisition. The rates at which carbon was assimilated, translocated and lost were analysed separately between the two time points (i.e., pulse and chase periods) using 2-way ANOVAs with species and depth as fixed effects. When the interaction was significant pairwise, Tukey tests were performed as a posteriori testing.