Linking host morphology and symbiont performance in octocorals

Octocorals represent an important group in reef communities throughout the tropical seas and, like scleractinian corals, they can be found in symbiosis with the dinoflagellate Symbiodinium. However, while there is extensive research on this symbiosis and its benefits in scleractinians, research on octocorals has focused so far mainly on the host without addressing their symbiosis. Here, we characterized and compared the photophysiological features of nine Caribbean octocoral species with different colony morphologies (sea fan, plumes, whips and rods) and related key morphological features with their respective symbiont photobiology. Colony features (branch shape and thickness), as well as micromorphological features (polyp size, density), were found to be significantly correlated with symbiont performance. Sea fans and plumes, with thinner branches and smaller polyps, favor higher metabolic rates, compared to sea rods with thicker branches and larger polyps. Daily integrated photosynthesis to respiration ratios > 1 indicated that the autotrophic contribution to organisms’ energy demands was important in all species, but especially in sea whips. This information represents an important step towards a better understanding of octocoral physiology and its relationship to host morphology, and might also explain to some extent species distribution and susceptibility to environmental stress.

In coral reefs, hexacorals, octocorals and sponges are the most important contributors to the living three-dimensional structures that give complexity to the system. Their diversity and complexity contribute to the correct functioning of the biogeochemical cycles, as well as to the productivity and nursing effect of these complex structures 1 . Octocorallia constitute, among these sessile suspension feeders, a group with significant presence in shallow tropical reefs, where they represent important structural components of the community 2,3 . In these environments, a large proportion of the octocoral species, as with many other reef organisms, are well known for engaging in photosymbiosis with unicellular algae, which provides them with metabolic advantages in nutrition 4 . So far, most of the photobiological approaches and the analysis of the predicted impacts of global change on species fitness in coral reefs have been performed on scleractinians. Octocorals, in this complex panorama, have been so far largely neglected 4 .
The most widespread coral reef photosymbionts contributing to the energy budget of mixotrophic cnidarians are dinoflagellates of the genus Symbiodinium. In scleractinians it has been shown that the symbiont cells produce photosynthates, organic compounds like glycerol and triglycerides that are translocated within and between cells to supplement the host nutrient requirements (up to 95% of that required by the host, e.g. 5 ). The importance of the heterotrophic and autotrophic input in these suspension feeders varies between species of a given taxonomic group. Even though in some corals most of their energetic demands can be fulfilled through the symbiosis, they still depend partially on heterotrophic feeding (mixotrophy) 6 . Among the most important features to

Results
Comparison of species' morphological traits. The species studied here represent a variety of colony morphologies and thus, differences in branch shape and thickness (Fig. 1). The branch thickness varied between species, but in general they could be divided into two groups, species with thin branches, which included the sea fan, plumes and whips, and those with significantly thicker branches, the sea rods (ANOVA, F 3,50 = 25.9, p < 0.0001) (Fig. 2a). The species also differed in branch shapes, with some species, like the sea fan and whips, presenting more flattened branches, resulting in significantly higher SA/V, while the round branches of the sea rods exhibit the lowest SA/V (ANOVA, F 3,96 = 255.3, p < 0.0001) (Fig. 2b). In addition, the studied species differed in micromorphological traits, such as the dimensions of their polyps (see Supplementary Table S1). This trait was related to the species' macro-morphological features (SA/V, branch thickness). With decreasing polyp size, the colony SA/V of the species increased exponentially ( Supplementary Fig. S1a), however, this correlation was not significant (R 2 = 0.34, p = 0.075). In contrast, the branch thickness of the species showed a significant linear relationship with polyp size (Supplementary Fig. S1b; R 2 = 0.94, p < 0.0001).
The species studied also differed significantly in organic and inorganic matter contents, with consistent patterns among species with similar colony morphologies. The lowest organic matter contents were found in sea rods, followed by the sea whips, while sea plumes and the sea fan exhibited the highest values (ANOVA, F 3,48 = 76.9, p < 0.0001; Fig. 2c). The differences in inorganic matter content, which is mainly composed of sclerites, might be related to the significantly longer and wider sclerites found in sea rods, while the other species had smaller sclerites (Supplementary Table S2). On the other hand, the presence and amount of colored sclerites was species-specific and did not show a relationship with branch morphology (Supplementary Table S2). For example, while no colored sclerites were found in E. tourneforti and P. nutans, they represented a high proportion of the total sclerites in E. mammosa and G. ventalina (Supplementary Table S2).
Additional principal component analyses, including the aforementioned macro-and micro-morphological traits, supported the separation of the studied species into four groups, based on their common names, sea fan, plumes, whips and rods ( Supplementary Fig. S1). Comparison of species' photo-physiological performance. No significant differences in symbiont cell density were found between the different branch morphologies (Fig. 3a), even though there were some differences between species (Supplementary Table S3). On the other hand, differences in branch morphology were accompanied by significant differences in chlorophyll content per ash-free dry weight (AFDW) (ANOVA, F 3,47 = 23.6, p < 0.0001), with the sea whips and the sea fan exhibiting the highest and lowest chlorophyll contents, respectively (Fig. 3b). These differences resulted in the highest chlorophyll content per symbiont cell (C i ) in the sea whips, while the sea fan G. ventalina showed the lowest C i ( Fig. 3c; Supplementary Table S3).
A correlation between branch morphology and the photosynthetic performance of the species was also found. The relationship between photosynthesis and irradiance was exponential in all species, with a coefficient of determination R 2 > 0.95, showing differences between species, but in most cases a similar pattern for species with comparable morphological traits ( Supplementary Fig. S3). The parameters derived from the P-E curves showed significant differences between species and branch morphologies, when either normalized to AFDW or symbiont cell number (Table 1) or surface area and chlorophyll content (Supplementary Table S4).
When comparing the photosynthetic performance of the species, we found slight differences, depending on the normalization (AFDW or symbiont cell density) in the cases of the sea plumes and rods. When normalized to AFDW, the sea fan G. ventalina had the highest photosynthetic capacity (P max ) and efficiency (α), as well as the highest respiration among the studied species, while the sea rods exhibited the lowest values (Fig. 4). The sea plumes, however, showed intermediate values, statistically different from sea fan when normalized by AFDW, while they were not statistically different from the sea fan when normalized by symbiont cell numbers (Fig. 4). In the case of the sea rods, they expressed significantly lower metabolic rates when normalized to AFDW; however, per symbiont cell they were no different from the sea whips.
When plotting the symbiont cell density per polyp against the photosynthetic performance per symbiont cell, maximum photosynthesis and efficiency decreased exponentially with increased symbiont cell density (P max : R 2 = 0.82, p = 0.0062; α: R 2 = 0.92, p = 0.0008), with sea whips and rods having the highest cell density per polyp, also exhibiting the lowest P max and α values (Fig. 5).
The daily integrated photosynthesis and respiration, when normalized by AFDW showed a significant linear relationship with the polyp density per surface area (Fig. 8a,b), both increasing with polyp density (P int : R 2 = 0.91, p = 0.0018; Respiration: R 2 = 0.93, p = 0.0012). In addition, daily photosynthetic and respiratory rate of the species were directly related, showing that species that exhibited the highest photosynthetic rate, such as the sea fan G. ventalina, also had the highest respiratory rate (Fig. 8c¸ R 2 = 0.99, p < 0.0001). The resulting photosynthesis to respiration ratio (P/R), with values >1 (from 1.1 to 1.8) in all species, showed significant differences between groups (ANOVA, F 3,45 = 4.2, p = 0.01134). The sea whips reached significantly higher ratios compared to the other groups (1.5-1.8; Table 1), while there were no significant differences between sea fan, plumes and rods (Fig. 8d).
Stable isotopes. The δ 13 C values of the nine studied species varied between −13.8‰ in P. anceps and −17.7‰ in A. americana ( Supplementary Fig. S4). The values of the δ 15 N stable isotope oscillated between a minimum of 4.2‰ in P. nutans and a maximum of 7.5‰ in E. mammosa ( Supplementary Fig. S4).

Discussion
The present study shows that octocoral symbiont performance is either directly or indirectly correlated to certain host morphological traits and that higher symbiont performance does not necessarily translate to a higher autotrophic input to the hosts' energy budget.
On one hand, symbionts are concentrated in the polyps (see Fig. 6 16,17 ); and thus, our findings of direct correlations between polyp size and density and symbiont photosynthetic performance are not surprising. On the other hand, polyp size of the studied species is directly related to branch morphology ( Supplementary Fig. S1), which is why we also found an indirect correlation between symbiont photosynthetic performance and this morphological trait.
The direct correlation between symbiont photosynthesis and polyp size was most likely related to the fact that the latter defines polyp SA/V, and hence, light absorption efficiency and gas exchange. For example, species like the sea fan G. ventalina and the sea plumes Antillogorgia spp. possess very small polyps compared with sea rods, whose polyps are more than an order of magnitude larger (Supplementary Table S1). As a consequence, the polyp density and SA/V of each polyp, relative to the overall colony size, is much higher in sea fans and plumes. This fact, together with lower symbiont cell densities per polyp, increases the amount of light reaching the symbionts in the polyp tissues, and hence, their photosynthetic performance (Fig. 5). The converse appears to be true for most sea rods, where larger and fewer polyps, higher symbiont density per polyp, and the thick, rod-shaped branch morphology seems to limit productivity, both in terms of per unit of symbiont cell and per AFDW. Larger polyps might result in more efficient feeding structures; however, several studies on non-symbiotic species also reported that gorgonians bearing the smaller polyps did not show smaller prey size or lower capture rates 18,19 . The sea whips represent an exceptional case, as their polyps, which are of intermediate size, have a very low density per surface area due to their arrangement in lateral series, only along the elevated branch ridges 20 . This, in addition with a very high symbiont cell density per polyp (1.5-2.4 and 10-29 times higher compared to sea rods and sea fan and plumes, respectively), seems to limit the photosynthetic performance of these species, most likely due to self-shading ("package effect"; sensu 21 ) between symbiont cells.
In the species studied here, polyp size showed a strong association with branch thickness ( Supplementary  Fig. S1b), which seems to be a general feature in octocorals 22 . Hence, symbiont photosynthesis was also found to be correlated with branch morphology, resulting in higher photosynthetic capacity per symbiont cell in sea fans and plumes, compared to sea whips and rods (see Fig. 4d). This might be in part due to the alternating and sometimes very flexible branches 10 of sea plumes that could aid in reducing self-shading. Sea fans, such as G. ventalina, have a net-like planar morphology with high surface area, which is positioned perpendicular to water flow 23 . This orientation, together with the flexibility of the colony due to relatively low sclerite content, facilitates the oscillation of the colony from full illumination to shade in synchrony with the wave period, thus maximizing symbiont performance, a feature that they share with the sea plumes. The movement of the bio-structures (like the leaves of marine phanerogams) has been demonstrated to be a key point to understand the optimization of light harvesting within a canopy 24 , and is a parameter that may also be important in octocorals gathered in dense patch structures.
Our findings of higher photosynthetic performance when polyps are smaller partially agrees with the hypothesis put forth by Porter 8 . He reported an inverse relationship between SA/V of the colony and polyp size in Caribbean scleractinian corals and interpreted it in terms of the corals' dependence on light-capturing ability versus plankton-capturing ability, respectively, assuming that small polyp forms are ineffective at prey capture and better suited for light capture. In the octocorals studied here, a similar relationship between SA/V and polyp size was found as reported by Porter 8 (see Supplementary Fig. S1a) and species with smaller polyps had a higher photosynthetic capacity. Interestingly, in contrast to Porter's conclusion, it did not translate into a more or less autotrophic nature of the species. In all species, the P/R values, derived from the daily integrated photosynthetic and respiratory rates of the species, were >1, indicative of at least partial reliance on autotrophic input for their energy demands. Despite its high photosynthetic rates, the sea fan G. ventalina did not exhibit higher daily net production, as it also had the highest respiratory rates. This seems to be a common feature of sea fans, also reported previously by Lewis and Post 25 , whom reported two times higher respiration rates in sea fans compared to sea rod species (Supplementary Table S5), most likely related to their high polyp density. Thus, the higher respiratory rates in species with smaller polyps and higher polyp density (sea fan, plumes) translated to similar P/R ratios as in species with larger polyps, such as the sea rods (see Fig. 8d). On the other hand, in sea whips, with polyps of intermediate sizes, the contribution of symbiont photosynthates to the hosts' energy budget seemed to be potentially more important than in the other species (see Fig. 8d).    This inconsistency with Porters' model might be related to: (1) not considering that smaller polyp size usually relates to higher polyp density, which might result in higher respiratory rates and (2) his assumption of a relatively uniform distribution of polyps over the surface area, which does not apply for the studied sea whips. In addition, previous studies showed that having smaller polyps per unit biomass of coral increases surface area for feeding or light capture 26,27 . The results of the δ 15 N analyses ( Supplementary Fig. S4) indicate a species-specific, but nontrivial input of energy from a heterotrophic source 28 , which shows that there is not a strong relationship between polyp size and potential prey capture (see below).
The higher potential autotrophic contribution in sea whips seems reasonable, as they have reduced polyp numbers, a trait that reduces self-shading between polyps but also diminishes the effectiveness of zooplankton feeding. This result also agrees with the reported shallower depth limit for the sea whips (≤15 m), ensuring high light availability, while sea fan, plumes and rods that potentially rely less on autotrophic input are able to extend their range deeper (Fig. 9).
In general, the comparison of P/R ratios between studies is difficult since photosynthesis and respiration determinations, as well as the way the ratios are calculated, differ greatly. In Red Sea octocorals, P/R values ranging from 1.0 to 1.4 have been found 29 , while they were slightly higher (1.07-1.61) in octocorals from the Great Barrier Reef 13 . The study of Baker et al. 15 reported P/R of 0.71-1.87 in Caribbean octocorals, with the lowest values found in sea rod species, while sea fans achieved the highest P/R. In the shallow-water Caribbean octocorals studied here, the P/R values were slightly higher, ranging from 1.1 to 1.8 (see Table 1). The discrepancy with the results of Baker et al. 15 might be due to differences in the way the photosynthetic and respiratory rates were measured and calculated. The P/R values reported in the former study derived from significantly lower metabolic rates, specifically lower photosynthetic rates, when compared to our study (Supplementary Table S5). This might have been the result of using freshly fragmented colonies and very long incubation times (7.5-8 h) in a small water volume (0.5 L), without stirring. This could have caused an underestimation of photosynthetic rates due to increased respiration of stressed samples, oxygen oversaturation and carbon limitation. Also, our calculations for P/R were done differently. We used an approach to obtain accurate and comparable daily integrated P/R ratios without the need of performing in situ measurements. For this, we used (a) light data from the field and the maximum photosynthetic capacity (P max ) and the photosynthetic efficiency (α) of the species (derived from photosynthesis-irradiance curves) to calculate their daily integrated photosynthesis, and (b) light respiration values to account for the higher respiration during the day. We believe that the data obtained from our approach are comparable to those derived from in situ measurements over a 24-h cycle.
Our results of P/R >1 indicate that the autotrophic contribution to the octocorals' energy budget is potentially important. However, the stable isotope values of the studied species (Supplementary Fig. S4) indicate that in neither of them the heterotrophic input is negligible 6,28 . Baker et al. 28 showed that those areas in which eutrophic disturbances were detected, the δ 15 N values were as high as 7.7‰ in G. ventalina (very similar to those found in the present study), suggesting sewage-derived N inputs. Thus, it is possible that the conditions in the Puerto Morelos reef lagoon (no longer oligotrophic due to the high tourist pressure) 30 allow a substantial N input from different heterotrophic sources 28 , thus coupling the autotrophic contribution with significant heterotrophic input. The heterotrophic input in Caribbean gorgonians is still a pending issue, being a potential factor that would help to understand the trophic flexibility of some species.
A final consideration to bear in mind is that despite the strong correlation between symbiont performance and species' morphological traits and the related differences in the potential autotrophic contribution to hosts' energy demands, other factors might have had an influence on our results. One potentially important factor is related to the symbiont phylotype hosted by the species. Differences in the associated phylotype can affect symbiont performance and P/R, as it is well known in scleractinian corals that photosynthetic performance and the proportion of photosynthetate translocated to the host can differ greatly between phylotypes 31,32 . In the studied species, the degree of symbiont specificity varies from sea fan (B1) > sea plumes and whips (B1, B1i, B1m) > sea rods (B1, B1a, B9, B19, B41) 33,34 , which may have some influence on the species-specific differences in symbiont performance and P/R found here. In this context, Baker et al. 15 pointed out that there might be a correlation between phylogeny and the species' reliance on autotrophic input, as they found a strong negative correlation between polyp size and carbon translocation from the symbiont to the host. Thus, they suggested that selection might have favored more obligate associations with highly productive Symbiodinium in species that in turn have reduced feeding structures but increased SA/V for light capture (smaller polyps).

Conclusions
Physiological studies in octocoral are scarce. The works investigating their physiological response to climate change 4 are also very few and still pose more questions than answers. This study demonstrates that there exists a pattern between hosts' morphological traits and symbiont performance and shows that there might be potential differences in the importance of the symbiosis for species' energy budget. It also demonstrates that both heterotrophy and autotrophy have a clear role in the energy inputs, although there is still work to be done on their respective contributions in the octocorals' mixotrophic strategy. respiration (sea whips are not included here, due to their polyp arrangement in lateral series only along the elevated branch ridges). (c) Linear relationship between daily photosynthesis and respiration and resulting P/R ratios (d) for octocorals with different morphological traits (sea fan-dark grey, sea plumes-black, sea whips-grey, sea rods-white). For calculation of the daily integrated respiration and P/R ratio, light respiration (R L ) to R D over a 12 h: 12 h cycle was considered. Results of one-way ANOVA and posthoc test are shown and significant differences between groups (ANOVA, Newman-Keuls test, p < 0.05) are indicated by superscript letters. Data represent means ± SE. The general assumption of high resistance of this group to ocean warming, compared to scleractinian corals, has been attributed to their ability to lower the dependency on autotrophic input and increase the contribution of the organisms' energetic demands through heterotrophy 13,35 . This might be the case in some of the studied species. But the potential higher dependency on autotrophy found here in sea whips, together with their low polyp density (suggesting inefficiency for heterotrophic feeding) leads us to hypothesize that these species might be more susceptible to increases in temperature in oligotrophic (non-perturbed) zones. The hypothesis is supported by a study from Puerto Rico, reporting the proportion of bleached colonies of different species in octocoral communities 36 . It showed that Pterogorgia sp. was one of the most affected species (84.5% bleached colonies), while species, such as Gorgonia sp., Eunicea sp. and Antillogorgia sp., were unaffected. The high resilience of Caribbean sea rod species to increased temperature has also been reported recently by Goulet et al. 34 , suggesting that these species are possibly more resistant to fast changing environmental factors. Thus, further work on the importance and efficiency of symbiont-host translocation in octocorals, as well as their flexibility in the relative contribution of autotrophy versus heterotrophy to octocoral nutrition, will be essential to shed light on the potential costs of the symbiosis, and hence, species' resilience in the face of stressful conditions and associated possible future seascape transformations.

Material and Methods
Study site and sample collection. During September 2014, nine species of octocorals, representing a variety of colony morphologies, including a sea fan (Gorgonia ventalina), plumes (Antillogorgia bipinnata, A. americana), whips (Pterogorgia anceps, P. citrina) and sea rods (Eunicea sp., Eunicea mammosa, E. tourneforti, Plexaurella nutans) (Fig. 1) were sampled at ∼2 m depth in the Puerto Morelos reef lagoon, Mexican Caribbean (see 37 for a detailed description of the study site). These species are considered abundant, key components of Caribbean coral reefs 3,10 . For each species, branches of six different colonies (5-8 cm height) were collected the day before the physiological measurements. The time elapsed between sampling and transport to the mesocosm facility at Universidad Nacional Autónoma de México (UNAM) did not exceed 15 minutes. The branches were fixed with non-toxic rubber in seawater flow-through tanks to allow recovery from handling. Preliminary tests showed that one day of recovery from sampling ensured adequate physiological conditions of the gorgonians, as the obtained photosynthetic rates were not different from samples kept to recover for two and three days before photophysiological measurements.

Photosynthesis measurements.
To determine oxygen fluxes of the octocorals, branches were incubated in a sealed temperature-controlled acrylic dual-chamber at 29 °C (temperature in the reef lagoon during collection period). The incubations were performed by exposing the samples to increasing light intensities and recording the changes in oxygen concentration through an optode system (FireStingO 2 , Pyroscience, Aachen, Germany), which in turn was connected to a computer running the Pyro Oxygen Logger (FireSting Pyroscience, Aachen, Germany) (see Supplementary Material for further details). Gross photosynthetic rates were calculated by adding respiration (average between dark and light respiration) to net photosynthesis. The highest photosynthetic rate was considered as P max and the photosynthetic efficiency (α) was estimated from the initial slope of the light-response curve by linear least-square regression analysis. The irradiance at the onset of saturated photosynthesis (E k ) was obtained from the ratio P max /α. For comparison with other studies, the photosynthetic parameters were normalized by ash-free dry weight (AFDW), symbiont cell number, surface area and chlorophyll a concentration of the samples. Using the photosynthetic efficiency (α) and daily variation of the irradiance at collection depth, daily integrated gross photosynthesis and respiration ( Supplementary Fig. S5), as well as the photosynthesis to respiration ratios (P/R) were calculated (see Supplementary Materials for further details).
The branches used in the incubations were photographed next to a ruler (a small transversal cut was also made to obtain the diameter of the branches) to calculate the total surface (for further details on surface area and volume calculation see Supplementary Materials). Immediately afterwards, the samples were frozen at −80 °C for subsequently determination of symbiont density and chlorophyll concentrations and ash-free dry weight per area (see Supplementary Figure S6).

Determination of chlorophyll concentrations and symbiont cell density.
Symbiodinium cell densities and chlorophyll (Chl a and c 2 ) concentrations were determined by adding filtered seawater to the tissue samples to a final volume of 5 mL, homogenizing them with an Ultra-homogenizer, after which the samples were centrifuged at 1500 x g for 15 min. The supernatant was discarded and the pellet resuspended in 5 mL filtered seawater, while subsamples were taken for subsequent cell density and chlorophyll determinations.
To determine chlorophyll concentrations, the pigment extraction was performed as outlined by Iglesias-Prieto et al. 38 and chlorophyll a (Chl a) and c (Chl c 2 ) contents were calculated using the equations of Jeffrey and Humphrey 39 . Chlorophyll concentrations were standardized to ash-free dry weight, branch surface area (see Supplementary Material) and Symbiodinium cell density.
A subsample of 1 mL was taken from the resuspended tissue pellet, fixed with Lugol's solution and subsequently, the symbiont cells were counted using an improved Neubauer haemocytometer and normalized to ash-free dry weight and surface area.
Octocoral sclerites and organic matter content. Determinations of the sclerite size and shape, as well as the proportion of colored sclerites were performed on sub-samples (n = 3 per species) that were placed in 1.5 ml Eppendorf tubes to which 5% sodium hypochlorite solution (household bleach) were added to remove the organic matter. After removing the supernatant, distilled water was added to the samples, and after stirring, an aliquot was placed on a slide and photos were taken from the sclerites (see Supplementary Fig. S7), using a binocular microscope (AxioImager Zeiss Microscope). Based on the photos, the length of the sclerites (n = 10 of each photo), as well as their maximum and minimum width were determined. In addition, the proportion of colored (pink to dark-brown) sclerites relative to the total amount of sclerites was determined from the photos.
For ash-free dry weight determination, branch fragments of 1 cm length from the samples used for photosynthetic measurements were subsampled from each species and photographed to calculate their area. Afterwards, the proteinaceous axis of the fragments was removed and ash-free dry weight was determined as described in Rossi et al. 40 . Based upon the ratio of AFDW per area, a conversion factor was determined for each species, which allowed the normalization of the photosynthetic parameters to metabolically active tissue (see Supplementary Figure S6).

Symbiont distribution in octocoral tissue.
For histological analysis, octocoral branches were fixed in formalin (10%) and treated in a series of rinsing and decalcification steps (for further details see Supplementary Materials), after which the tissue sections were stained with Meyer's hematoxylin and eosin procedures, and coverslipped with Organol/Limonene (Sigma O8015) mounting medium. Subsequently, photographs were obtained using an AxioImager Zeiss Microscope.

Stable isotope analysis.
Proportions of carbon (δ 13 C) and nitrogen (δ 15 N) stable isotopes can vary with nutrient source and trophic level of consumers. Thus, stable isotope analysis has been successfully used to elucidate food source partitioning, and food web dynamics 41 . Samples of octocoral coenenchyma (n = 5 for each species) were oven-dried at 60 °C and homogenized using mortar and pestle. To remove carbonate structures, the sample was acidified by adding 1 M HCl drop-by-drop 41 . Afterwards, the samples were re-dried at 60 °C for 24 h.
One to five milligrams per sample were weighed in tin capsules, combusted and the resultant gases were analyzed in a Thermo Electron Isotope Ratio Mass Spectrometer for stable isotope abundance. Internal laboratory standards were used after every five samples to calibrate the system. Isotopic values were expressed in δ notation as parts per thousand (‰) differences from international standards (Vienna Pee Dee Belemnite and atmospheric N 2 for carbon and nitrogen respectively): δX = [(R sample /R standard ) − 1] * 10 3 , where X is equal to 15 N or 13 C and R is the corresponding ratio 13 C/ 12 C or 15 N/ 14 N. Based on replicates of laboratory standards, analytical precision was ±0.2 and ±0.1‰ for δ 15 N and δ 13 C, respectively.
A posteriori tests. Since most of the symbionts reside in the polyps (this study 16,17 ), we tested the correlation between symbiont photosynthesis and cell density per polyp. To that end, we used the data on intercalyx distance reported by Velásquez and Sánchez 11 for the studied species to calculate the number of polyps per surface area (Supplementary Table S1), which allowed us to calculate the number of symbiont cells per polyp based on our data of symbiont cell density per surface area. To test our hypothesis that symbiont physiology correlates to the species' morphological features, we also used macro-morphological data (branch thickness) and micro-morphological data (polyp size) to show the relationship between these morphological traits and symbiont photosynthesis.
As proxy for polyp size, the volume of the calyx (structure into which the polyp retracts) was used, calculated from data on calyx depth and diameter, reported by Velásquez and Sánchez 11 for the studied species (Supplementary Table S1). Either calyx/polyp diameter or depth (distance from the base where polyp is anchored to surface edge of aperture) have been used in previous studies as a proxy for polyp size. However, there is no clear relationship between calyx diameter and depth, as similar diameters could be found in sea whips and rods, but when considering also the depth of the calyx, the calculated space that the polyp occupies, and therefore polyp size (here calculated as calyx volume), differed greatly between these two groups (Supplementary Table S1). Thus, we consider that including both features, as a proxy for the anatomical space the polyp is occupying, represents a more exact proxy for polyp size.