Ecophysiology of mesophotic reef‐building corals in Hawai‘i is influenced by symbiont–host associations, photoacclimatization, trophic plasticity, and adaptation

Mesophotic reef corals remain largely unexplored in terms of the genetic adaptations and physiological mechanisms to acquire, allocate, and use energy for survival and reproduction. In the Hawaiian Archipelago, the Leptoseris species complex form the most spatially extensive mesophotic coral ecosystem known and provide habitat for a unique community. To study how the ecophysiology of Leptoseris species relates to symbiont–host specialization and understand the mechanisms responsible for coral energy acquisition in extreme low light environments, we examined Symbiodinium (endosymbiotic dinoflagellate) photobiological characteristics and the lipids and isotopic signatures from Symbiodinium and coral hosts over a depth‐dependent light gradient (55–7 μmol photons m−2 s−1, 60–132 m). Clear performance differences demonstrate different photoadaptation and photoacclimatization across this genus. Our results also show that flexibility in photoacclimatization depends primarily on Symbiodinium type. Colonies harboring Symbiodinium sp. COI‐2 showed significant increases in photosynthetic pigment content with increasing depth, whereas colonies harboring Symbiodinium spp. COI‐1 and COI‐3 showed variability in pigment composition, yield measurements for photosystem II, as well as size and density of Symbiodinium cells. Despite remarkable differences in photosynthetic adaptive strategies, there were no significant differences among lipids of Leptoseris species with depth. Finally, isotopic signatures of both host and Symbiodinium changed with depth, indicating that coral colonies acquired energy from different sources depending on depth. This study highlights the complexity in physiological adaptations within this symbiosis and the different strategies used by closely related mesophotic species to diversify energy acquisition and to successfully establish and compete in extreme light‐limited environments.

Light availability is a crucial factor affecting the physiology, productivity, distribution, and abundance of life in the ocean (Gattuso et al. 2006;Falkowski and Raven 2007). Both quantity and quality (spectral composition) of photosynthetically active radiation (PAR) can influence the recruitment, physiological performance, and survival of different life stages of marine organisms (Mundy and Babcock 1998;Falkowski and Raven 2007;Roth 2014). As depth increases, light intensity not only decreases exponentially but also ultraviolet and red wavelengths decrease faster causing a spectral enrichment in the blue and blue-green wavelengths (Kirk 2011). Scleractinian corals are ecologically and economically important organisms that live in a mutualistic symbiosis with intracellular algae (dinoflagellates of the genus Symbiodinium) and depend on PAR to obtain energy for photosynthesis. Symbiodinium translocate fixed carbon to their coral host to meet their host's metabolic demands as well as contribute to coral growth and calcification (Goreau and Goreau 1959;Muscatine 1990). Most symbiotic *Correspondence: jpgamino@uw.edu This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Additional Supporting Information may be found in the online version of this article.
This manuscript is dedicated to the memory of our dearest Ruth Gates, an incredible scientist, mentor, and friend. We deeply miss you. corals cannot photosynthesize and survive at depths greater than 60 m as PAR availability is limited in spectral features and quantity (Fricke et al. 1987;Lesser et al. 2009;Loya et al. 2016). However, some species persist and even flourish in dimly lit environments such as the mesophotic zone, where downwelling irradiance can be as low as 1% of surface irradiance (Kahng and Kelley 2007;Lesser et al. 2009;Bridge et al. 2011;Pyle et al. 2016). Mesophotic coral ecosystems, as currently defined, are deep fore-reef communities comprised largely of light-dependent zooxanthellate corals, azoxanthellate scleractinian corals, macroalgae as well as sponges from 30 m to over 150 m depths (Lesser et al. 2009;Hinderstein et al. 2010;Baker and Harris 2016). Upper mesophotic coral reefs inhabit depths from 30 to 60 m, whereas lower mesophotic reefs are present from 60 to 150 m depths. To date, mesophotic reef corals remain largely unexplored in terms of the physiological adaptations and mechanisms that allow resident species to live and successfully reproduce in these environments Shlesinger et al. 2018). However, it is important to study mesophotic reefs because of the fundamentally new scientific insights these environments can provide and their potential to serve as refuges and nursery habitats for many ecologically and commercially important species (Riegl and Piller 2003;Lesser et al. 2009;Bongaerts et al. 2010;Bryan et al. 2013;Holstein et al. 2016).
Light-dependent corals are important members of the mesophotic community (Lesser et al. 2009). Leptoseris is a common obligate zooxanthellate genus found in mesophotic environments around the world including the Pacific Ocean, Red Sea, and the Caribbean Sea (Bouchon 1981;Fricke and Knauer 1986;Maragos and Jokiel 1986;Kahng and Maragos 2006;Ziegler et al. 2015). The deepest reported zooxanthellate species is Leptoseris hawaiiensis, which has been found as deep as 165 m depth at Johnston Atoll (Maragos and Jokiel 1986).
One of the most studied corals from the deep mesophotic zone is Leptoseris fragilis from the Red Sea. This species lives exclusively between 40 and 145 m depths (154-1.2 μmol photons m −2 s −1 ; Fricke and Schuhmacher 1983;Fricke et al. 1987) and its success in the mesophotic zone has been attributed to its photophysiological flexibility, efficient use of low photon flux densities, and ability to switch between photoautotrophy and heterotrophy to meet its metabolic needs (Schlichter et al. 1985;Fricke et al. 1987;Schlichter 1991;Schlichter et al. 1997). However, the population genetics of L. fragilis host and Symbiodinum have not been studied and any physiological differences with depth are likely to reflect adaptations of different host species and/or Symbiodinium types. The Caribbean coral Montastrea cavernosa shows morphological and genetic differentiation across depths Brazeau et al. 2013), and displays physiological acclimatization and photoadaptation associated with different Symbiodinium composition between shallow (0-46 m) and greater depths (61-91 m). Furthermore, a shift from photoautotrophy to heterotrophy occurred between 45 and 61 m, and lower productivity was found at greater depths .
Previous studies have found that both photobiological flexibility and Symbiodinium type are essential to persist in lightlimited environments (Frade et al. 2008;Cooper et al. 2011;Ziegler et al. 2015). In the Caribbean, Madracis spp. showed photoacclimatization by varying Symbiodinium density and type, as well as efficiency of light harvesting across depths from 5 to 40 m (Frade et al. 2008). In Western Australia, Pachyseris speciosa and Seriatopora hystrix showed different patterns of host-Symbiodinium specialization across depths (0-60 m). P. speciosa hosted mainly Symbiodinium type C across its depth range (0-60 m), whereas S. hystrix hosted Symbiodinium type D1a at shallow depths (0-20 m) and Symbiodinium C type at deeper depths (10-45 m; Cooper et al. 2011). Interestingly, S. hystrix had an increase in metabolic costs when hosting Symbiodinium C compared to type D1a while exposed to higher irradiances, suggesting that metabolic demands may depend on Symbiodinium type Leal et al. 2015;Pernice et al. 2015). In the Red Sea, Symbiodinium densities and ratios of photoprotective and photosynthetic pigments decreased with depth (0-60 m) in Porites, Leptoseris, Pachyseris, and Podabacia. Porites harbored Symbiodinium type C15 while Pachyseris and Podabacia hosted mainly Symbiodnium type C1 and had a more limited depth range (Ziegler et al. 2015). These studies underlie the importance of genotyping Symbiodinium when investigating coral physiology from different light regimes. Both adaptation (change in the genetic makeup of a population over multiple generations) and acclimatization (when an organism adjusts to a change in its environment) strategies can play an important role in an organism's survival when light becomes limited.
In Hawaiian waters, Leptoseris is the dominant coral genus in the mid to deep mesophotic zones (60-160 m) and colonies of Leptoseris species form the most spatially extensive mesophotic coral ecosystem documented to date (Costa et al. 2015;Veazey et al. 2016;Spalding et al. 2019). These reef-building corals form extensive reefs with up to 100% live coral cover predominantly at 90-100 m (Kahng and Kelley 2007;Rooney et al. 2010). Species of Leptoseris form large (> 1 m in diameter), thin-walled colonies with plating and foliaceous morphologies that can grow approximately 1 cm yr −1 (Kahng 2013;Pyle et al. 2016), comparable to some shallow water corals. In contrast, L. fragilis mesophotic colonies in the Red Sea grow~0.2-0.8 mm yr −1 and have a maximum diameter of 8-10 cm (Fricke et al. 1987). Because of the significant role of Leptoseris corals in creating mesophotic environments, insight into their physiology is critical to understand the ecosystem.
Leptoseris reefs in Hawai'i are hotspots of biodiversity and productivity with significantly higher rates of endemism in fishes compared to shallow reefs (Rooney et al. 2010;Kane et al. 2014;Pyle et al. 2016). These reefs facilitate recruitment of other species (Supporting Information Fig. S1), may function as refuges for fishery-targeted species already impacted on shallow reefs (Lindfield et al. 2016), and are distinctive habitats that harbor many undescribed and unique species (Pyle et al. 2016;Spalding et al. 2016).
Recent morphological and molecular studies have addressed the high cryptic diversity in Leptoseris from Hawai'i, revealing new insights about the diversity and adaptation of Leptoseris-Symbiodinium associations in this light-limited environment (Luck et al. 2013;Pochon et al. 2015). Using the cox-1-rRNA intron and skeletal micromorphology, Luck et al. (2013) reported a depth zonation pattern for different Leptoseris clades. Using a similar molecular approach (cox-1-rRNA intron), Pochon et al. (2015) examined host-Symbiodinium relationships in Leptoseris finding five cryptic species and variation in associated Symbiodinium across depths. However, the physiological flexibility of these different assemblages or their success in meeting host metabolic demands at different depths remains unexamined. Kahng et al. (2012) attributed the success of Leptoseris in the mesophotic zone to a low level of reflectivity associated with microscale optical geometry of the coral skeleton. However, this hypothesis remains largely untested.
In this study, we used an integrative approach to explore the status and flexibility of photobiology and trophic ecology of mesophotic species in the Leptoseris genus. Specifically, we address the following questions: (i) What mechanisms enable the reef-building coral Leptoseris spp. to live in the mesophotic zone? (ii) Is the photobiology of Leptoseris influenced by host-Symbiodinium specialization? (iii) What are the strategies for energy acquisition from 60 to 130 m depths? An understanding of the physiological state and flexibility of these species is essential for characterizing the status and resilience of mesophotic reefs in Hawai'i.

Study site
Samples of cryptically diverse Leptoseris spp. were collected across multiple depths ( Table S1). The 'Au'au Channel (Fig. 1B) separates the islands of Maui and L ana'i and has a bottom topography consisting of a gently sloping, continuous limestone bridge. Underwater irradiance was measured at six sites in areas with Leptoseris spp. reefs in the 'Au'au Channel and Maui Nui   Information Table S2) to characterize the light environment. Irradiance was measured by lowering a calibrated spherical (4π) quantum sensor (Underwater LI-193SA, LI-COR©) through the water via a profiling rig; data were stored with a LI-COR© LI-1400 data logger. Scalar irradiance (E 0 ) is the best all-round measure of light availability for photosynthesis at a given depth, because photons are equally useful in photosynthesis regardless of their source direction (Kirk 2011). Furthermore, we observed Symbiodinium within coral cells on both the top and undersides of coral colonies, suggesting that irradiance from multiple directions could be used for photosynthesis. The sensor was attached to a 1-m-long arm mounted on a polyvinyl chloride housing to reduce instrument shading. The housing was integrated with a calibrated pressure transducer for depth (m) and a temperature sensor. PAR (in μmol photons m −2 s −1 from 400 to 700 nm) was recorded at known depths in the water column to maximum depths ranging from 64 to 94 m on calm, clear days (< 10% cloud cover) during midday (12:00-13:00 h). Furthermore, irradiance profiles were conducted on the sunny side of the vessel to reduce shadows. The vertical attenuation coefficient for scalar irradiance (K 0 ) from the downward portion of each profile was calculated according to the relationship in Beer's Law: E z = E 0 × e −K0z where z represents depth, E z represents the intensity of irradiance at depth z, and E 0 represents the irradiance just beneath the surface (Kirk 2011). K 0 was then used to extrapolate irradiance at every 1 m depth for each profile, and an average irradiance profile (AE SE) was calculated from six profiles. Optical depths for 1% and 10% surface irradiances were calculated using K 0 (Kirk 2011). In deep, clear, homogenous waters with little scattering and a diffuse light environment, the relationship of the vertical attenuation coefficients K 0 (scalar irradiance) and K d (downwelling irradiance) is nearly one (Kirk 2011), thus allowing for a direct comparison between K 0 and K d . Daily mid-day irradiance vertical profiles showed little stratification in the water column, with r 2 values ranging from 0.97 to 0.99 (Supporting Information Table S3). The vertical attenuation coefficients, K 0 (m −1 ), ranged from −0.037 to −0.047 m −1 , with a mean K 0 (AE SE) of −0.041 m −1 (AE 0.001). The calculated mean optical depths, which correspond to the midpoint (10% surface irradiance) and the lower limit (1% surface irradiance) of the euphotic zone, were 56 and 112 m, respectively. This corresponded to a decrease in irradiance levels from 103 to 11 μmol photons m −2 s −1 , respectively (Supporting Information  Table S3).

Coral collections and abundance
At each site, coral cover was assessed visually by the scientific collector viewing an area~5 m 2 around the site of coral collection via a submersible window (Fig. 1). Representative corals approximately 20-30 cm in diameter were haphazardly selected from the middle of Leptoseris reefs, with each sample separated by at least 10 m from other samples. A small, triangular piece of coral from the middle to the outer edge of each coral head was removed using a Schilling Titan 4 manipulator arm ( Fig. 1C) and placed in an individual sample container in the sampling basket. After collection, samples were placed in a darkened container with ambient seawater at in situ seawater temperatures and processed in a darkened, air conditioned laboratory, using red light headlamps with an intensity of~1 μmol photons m −2 s −1 from a distance of 40-50 cm, onboard the R/V Ka'imikai-O-Kanaloa within 3-9 h of ascent. Two sets of coral samples were collected. One sample set was analyzed for photosynthetic potential, host genetics, Symbiodinium genetics, Symbiodinium density, photosynthetic pigments, and total lipids. A second set of samples from different coral colonies was collected for isotopic analyses. It is important to note that the second set of samples did not include genetic analyses for host and Symbiodinium identification (Supporting Information Table S1). Sample size for physiological and photobiological analyses differed due to sample and instrumentation availability and is reported for each parameter in the sections below. Supporting Information Table S1 includes the location and date of all collections with the physiological and photobiological analyses performed on each sample. After collecting and assessing photosynthetic potential (in the first set of coral samples), fragments were frozen using dry ice and maintained in a −80 C freezer until analyzed.

Laboratory analysis
Pulse amplitude-modulated fluorometry Photosynthetic potential was assessed with modulated chlorophyll fluorescence measurements taken with a pulse amplitude-modulated (PAM) fluorometer and a red (650 nm) excitation beam (Diving-PAM). A 2-cm-long piece of black tubing (1 cm diameter) was attached to the PAM fiber optic sensor to standardize the area measured and to ease the placement of the sensor onto the coral surface. Actinic PAR values from the Diving-PAM with the fiber optic sensor tubing were calibrated with a cosine underwater quantum sensor (LI-COR LI-192SA) and data logger (LI-COR LI-1400). Measurements of minimum (F o ) and maximum (F m ) fluorescence were used to calculate variable (F v = F m /F o ) fluorescence and subsequently the maximum quantum yield of photosystem II (PSII) fluorescence (F v /F m ) or the number of functional photosynthetic units (Ralph and Gademann 2005). To account for potential spatial variation in coral physiology,~10 measurements of F v /F m were taken from haphazardly selected, spatially separated points on the coral tissue surface and averaged for each sample (n = 57 coral colonies). Rapid light response curves (RLC) were then taken to measure photosynthetic potential under different light levels following Ralph and Gademann (2005). Each RLC exposed the coral to eight incremental steps of irradiance, similar to the local environment, from 0 to 75 μmol photons m −2 s −1 . This lower range of irradiances was used on corals from all depths to avoid photoinhibition in corals from deeper depths. Irradiances were set to provide sufficient measurement steps~4-5 measuring points below saturation, for accurate initial slope (α) calculations. The electron transport rate (ETR) at each irradiance was calculated using the formula ETR = F v /F m × PAR × 0.5, where the quantum yield is the parameter measured at each irradiance of PAR, and 0.5 is the theoretical distribution of absorbed photons between PSII and photosystem I (PSI), although this balance was not evaluated for these samples. Average ETRs by actinic irradiances for each algal sample were fit to a threeparameter nonlinear model as described by Ralph and Gademann (2005). RLC data were used to estimate the relative maximum ETR (rETR max ), α (initial slope of the RLC), and the light saturation coefficient (E k ). We used the relative measure of rETR because the exact absorbance of the coral is unknown and likely varies by species and depth-related skeletal phenotypic differences. Curves were fit using the Regression Wizard in Sigmaplot (v. 12.0, SPSS), and estimates of rETR max , α, and E k were used for analyses. In all samples, the model fits the data well with an r 2 of 0.97 AE 0.01 (mean AE SEM).

Algal pigments
Algal pigments were extracted as previously described in Padilla- Gamiño et al. (2013). Briefly, glass fiber filters (0.7 μm pore size, GF/F; Whatman) containing the coral and algal tissue homogenate (n = 75) were extracted in 3 mL of highperformance liquid chromatography-grade acetone in glass culture tubes along with 50 μL of an internal standard (canthaxanthin) at 4 C in the dark for 24 h. The extracts were processed following Bidigare et al. (2005); pigments were detected with a ThermoSeparation Products UV2000 detector (λ 1 = 436, λ 2 = 450). Concentrations of photosynthetic (chlorophyll a [Chl a], Chl c 2 , and peridinin) and photoprotective (β-carotene, dinoxanthin, diadinoxanthin, and diatoxanthin) pigments and ratios at the colony and algal cellular level were computed by normalizing to coral surface area (μg pigment cm −2 ) and Symbiodinium density (pg pigment cell −1 ). Our study did not attempt to differentiate the sources of Chl a among Symbiodinium and endolithic algae. Approximately 85% of the coral skeletons had visible endoliths; 71% of the skeletons had green endoliths; and 23% had both green and orange endoliths. To minimize endolithic contribution in the pigment extractions, we only used homogenate derived from coral tissue (not skeleton). Identities of pigments produced exclusively by dinoflagellate symbionts (i.e., peridinin and Chl c; Kleppel et al. 1989) show proportions or similar patterns to our Chl a data suggesting that our pigment quantification mostly represented extracts from Symbiodinium (Supporting Information Fig. S2).

Symbiodinium densities
The densities and size of Symbiodinium cells were examined according to Apprill et al. (2007). In brief, coral tissue (n = 75) was removed from the frozen fragment using a Waterpik ® and filtered seawater (FSW, 0.2 μm). The homogenate tissue was then blended for 30 s and centrifuged for 5 min. Multiple washing steps were performed with FSW to ensure the complete separation of Symbiodinium and coral tissue. The Symbiodinium suspensions were analyzed using a Beckman-Coulter XL flow cytometer with a 15-mW argon ion laser set to excite at 488 nm. The flow cytometer was interfaced with an Orion syringe pump for quantitative sample analysis using a 3 mL syringe delivering 100 μL of suspended cells at a flow rate of 50 μL min −1 for measurement of fluorescence emission of Chl a (630 nm dichroic filter, 680 nm bandpass filter), as well as forward and side scatter signals. Duplicate abundance estimates were averaged for each sample. Symbiodinium abundances were standardized to coral surface area measured with the aluminum foil method (Marsh 1970). Symbiont subpopulations were identified based on side scatter (cell size) and fluorescence characteristics. Listmode files generated by the Flow cytometer were analyzed using FlowJo software (Treestar) by species. The number of symbionts from each symbiont subpopulation for each individual was quantified using FlowJo software and converted to the percent distribution of the three subpopulations and graphically represented using a ternary diagram (SigmaPlot 13.0 ©2014, Systat Software; Supporting Information Fig. S3).

Total lipids and biomass
Total lipids and biomass (n = 75) were obtained following Rodrigues and Grottoli (2007). In brief, ground coral samples were extracted in a 2 : 1 chloroform : methanol solution, the organic phase was then washed using 0.88% KCl, and the lipid extract dried to a constant weight. Tissue biomass was estimated using the difference between dry weight and ash free dry weight. Lipids and biomass were normalized to surface area.

Stable isotopes
Coral tissue and Symbiodinium were removed from skeletons using a WaterPik with FSW (n = 21-25; Johannes and Wiebe 1970). Host and Symbiodinium fractions were separated using a centrifuge that isolated zooxanthellae as a pellet. Pellets were washed and resuspended using FSW, then centrifuged, repeating the procedure twice. To ensure separation, pellets and supernatant were examined periodically under a microscope. Both resuspended pellets and supernatant were separately collected on a precombusted GF/F filter using vacuum filtration. Filtered fractions were analyzed for elemental and isotopic composition of carbon and nitrogen (δ 13 C = ratio of 13 C : 12 C relative to Vienna Peedee Belemnite Limestone Standard and δ 15 N = ratio of 15 N : 14 N relative to air, reported in permil units) using an elemental analyzer coupled to a Thermo Delta V isotope ratio mass spectrometer. When ample material was present, sample analyses were duplicated. A subset of samples was acidified on the filter to test for residual skeletal material. Acidification showed no indication of skeletal contamination with replicate standard deviations averaging 0.2‰ for Symbiodinium (C and N) and 0.3‰ and 0.8‰ for C and N, respectively.

Statistical analyses
Independent data analyses for each species were grouped based on a depth-dependent . Physiological values of Leptoseris papyracea were not included in any of the statistical analyses associated with depth because the species was found at only one depth range (90 m). Comparisons of physiological traits between species were performed using a one-way analysis of variance (ANOVA). Prior to analyses, data were normalized as necessary using logarithmic or square root transformations to achieve homogeneity of variances and normality. Homogeneity of variance and normality was assessed using Shapiro-Wilk W and Levene tests, respectively. When significant effects were identified, Tukey's post hoc tests were performed to determine differences between depths. Statistical differences were significant at the α < 0.05 level. Wilcoxon-Signed Rank Tests were performed when normality was not achieved after transformation. Data are represented as mean AE SEM. We did not perform ANOVA analyses for isotopic signatures because genetic information was not available for this subset of samples, instead the relationship between isotopic signature and depth in Leptoseris spp. was assessed using linear regressions. Statistical analyses were performed using JMP version 12.2.0 (SAS Institute).
Photophysiological specialization of Leptoseris spp. across depth L. scabra Our data suggest that L. scabra is a broadly shade-tolerant species, adjusting subcellular photosynthetic components to maintain optimal performance with decreasing PAR and depth. Photosynthetic and accessory pigments in L. scabra increased with depth (p < 0.05; Fig. 3; Supporting Information Table S4) allowing colonies to maintain ETR max . Chl a cm −2 concentrations at 70 m were only 28% of the pigment concentrations at 100 m depth (0.43 and 1.53 μg cm −2 , respectively; Supporting Information Table S4). Therefore, surface area for light harvesting is the most important factor limiting the depth distribution of this species. Despite being found across a wide depth range (70-100 m), L. scabra did not exhibit significant changes in other photophysiological variables across depths, such as pigments per cell, pigment ratios, Symbiodinium cell size ( Fig. 3; Supporting Information Table S4), dark-adapted quantum yield, maximum rETR, or E k (p > 0.05; Fig. 4). Similarly, lipid content did not change with depth in L. scabra or any of the other Leptoseris spp. (Supporting Information  Table S4). L. scabra showed very similar physiological characteristics to L. tubulifera including close association with Symbiodinium haplotype CO1-2, high tissue biomass, small-size Symbiodinium, and higher β-carotene : Chl a ratios compared to the other Leptoseris spp. (Table 1).
L. tubulifera Similar to L. scabra, L. tubulifera showed an increase in pigment concentrations with depth (p < 0.05; Fig. 3; Supporting Information Table S4) to maintain optimal performance with decreasing PAR and depth; with Chl a cm −2 increasing~55% from 80 to 90 m (0.77 to 1.40 μg cm −2 , respectively; Fig. 3). Although L. scabra was present from 70 to 100 m, we excluded samples from 70 and 100 m in the statistical analyses due to low sample size. Not surprisingly, between 80 and 90 m, there was no significant change in photosynthetic pigments normalized by cell, lipids, dark-adapted quantum yield, maximum rETR max , or E k (p > 0.05; Figs. 3-4; Supporting Information Table S4). However, β-carotene : Chl a decreased~8% from 80 to 90 m depth (Fig. 3). Physiological trends in L. tubulifera were very similar to L. scabra (see above; Table 1) most likely due to their shared association with Symbiodinium haplotype CO1-2.

Leptoseris spp. complex isotopic signatures
Carbon isotopic composition of Symbiodinium ranged from −21. respectively). Host δ 13 C values decreased linearly with increasing depth (p < 0.0001, F = 49.5; Fig. 5A). However, Symbiodinium δ 13 C values did not show a significant trend with depth. Bulk stable isotopic values for nitrogen in Symbiodinium ranged from 1.1‰ to 3.4‰, whereas that of the host tissue ranged from 2.7‰ to 6.3‰ (mean 2.4‰ and 4.1‰, respectively). Symbiodinium δ 15 N values decreased with increasing depth (p = 0.008, F = 8.69; Fig. 5B). However, host nitrogen isotopic composition did not show any overall trend with increasing depth (p = 0.243). Differences between isotopic composition of coral host and symbiotic algae exhibited dissimilar trends with depth. With increasing depth, Symbiodinium became significantly more distinct via changed δ 15 N compared to host tissue, whereas host tissue became significantly more depleted in δ 13 C compared to Symbiodinium (p = 0.0397, F = 4.83 and p = 0.007, F = 9.04, respectively; Fig. 5C). C : N ratios for Symbiodinium increased with depth (p = 0.025) and ranged from 5.7 to 9.5 (mean = 6.9), whereas C : N ratios for the host decreased with depth (p = 0.0116) and ranged from 5.5 to 16.75 (mean = 7.79).

Discussion
In this study, we explore the depth-dependent variability in trophic strategies, photophysiological, and genetic traits of four Leptoseris spp. Our robust, cross-cutting approach revealed different physiological strategies used by the algae and coral species to obtain energy and persist in a deep-shade environments. This work uncovered three fundamental insights into coral-algal symbiosis, photophysiology, and trophic plasticity    in mesophotic environments. To begin, plasticity of Leptoserisalgal photophysiology across depths was associated with the type of Symbiodinium spp. Colonies harboring Symbiodinium spp. COI-2 (L. tubulifera and L. scabra) showed significant increases in photosynthetic pigment content with increasing depth but no change in chlorophyll fluorescence with depth ( Fig. 6), suggesting a complex that is broadly tolerant to its depth conditions. In contrast, colonies harboring Symbiodinium spp. COI-1 or COI-3 (L. hawaiiensis and Leptoseris sp. 1) showed variability in pigment ratios, chlorophyll fluorescence, and Symbiodinium density and/or size (Fig. 6), suggesting that flexibility in photoacclimatization as well as photoadaptation among these Leptoseris-algal complexes depends primarily on Symbiodinium genotypes. Furthermore, despite remarkable differences in photosynthetic adaptive strategies with depth, we found no differences in total lipid content of Leptoseris spp. species over the same depth range (Supporting Information Table S4). This finding suggests that photosynthetic acclimatization with depth resulted in similar acquisition and translocation of lipids to the hosts and/or hosts may be supplementing their total lipid reserves from nonphotosynthetic sources at deeper depths. Finally, isotopic signatures of both host and Symbiodinium changed with depth supporting the lipid findings and indicating that coral colonies may acquire energy from different sources along their depth gradient. Overall, our results show marked complexity in physiological adaptations of species in the Leptoseris spp. complex and highlight the diversity of strategies used to acquire energy and succeed in environments where irradiance is extremely limited.

Photophysiology
Because mesophotic reef-building corals live close to the limits of the depth distribution for Scleractinian corals, their photophysiology is likely to reveal novel acclimatization and adaptation strategies in photosynthetic symbioses. Our results show that Leptoseris species have different host-Symbiodinium specializations and physiological plasticity along a depth gradient. L. scabra and L. tubulifera were almost exclusively associated with Symbiodinium spp. COI-2 haplotype but showed the ability to increase Symbiodinium pigment concentrations (per surface area) with depth. This pattern is similar to M. cavernosa in the mesophotic Caribbean ), but the opposite trend was observed for L. fragilis in the mesophotic Red Sea (Fricke et al. 1987;Schlichter et al. 1997). L. tubulifera, however, showed greater phenotypic variation in symbiont populations (Supporting Information Fig. S3) and lower β-carotene : Chl a with increasing depths, suggesting that this species has more diverse Symbiodinium (type, size, quantum yield, or cell division rates) and that β-carotene has primarily a photoprotective function at shallower depths.
Leptoseris sp. 1 (70-100 m) was the only species that increased Symbiodinium density with depth. Colonies of this species at 100 m had twice the density of Symbiodinium than colonies at 70 m. Increased Symbiodinium density under low light conditions has been observed in the field experimentally in Stylophora pistillata (Dubinsky and Jokiel 1994;Titlyanov et al. 2001). Under light-limited conditions (~8% surface irradiance), light harvesting in S. pistillata was maximized by an increase in Symbiodinium density, which was primarily regulated by division and degradation of Symbiodinium cells (Titlyanov et al. 2001). Thus, it is likely that increased Symbiodinium density may facilitate photosynthetic energy acquisition of Leptoseris sp. 1 at deeper depths. Nutrients have also been associated with increased densities of Symbiodinium (Sawall et al. 2014); however, the nutrient dynamics in the mesophotic region and the role of nutrient limitation or enrichment in the ecophysiology of Leptoseris spp. in the mesophotic zone remains unknown. Conversely, Symbiodinium densities of L. fragilis in the mesophotic zone of the Red Sea decreased~50% between 100 and 130 m (Fricke et al. 1987) and Symbiodinium densities decreased in colonies of the genera Leptoseris, Pachyseris, Seriatopora, Porites, and Podabacia from 1 to 60 m depths in Australia and the Red Sea (Cooper et al. 2011;Ziegler et al. 2015).
The increase in Symbiodinium densities with depth in Leptoseris sp. 1 was not associated with changes in Chl a cm −2 or Symbiodinium size. However, Chl c : Chl a showed an increase with depth (70 to 80 m), indicating photoacclimatization by increasing light harvesting antennae to gather light and augment light energy capture to the photosynthetic reaction center (Roth et al. 2010). Additionally, at shallow depths (70-80 m), Leptoseris sp. 1 was strictly associated with Symbiodinium haplotype COI-3, whereas at deeper depths colonies were associated with Symbiodinium haplotypes COI-2 and COI-3 (Pochon et al. 2015). Thus, it is likely that the differences in photosynthetic pigment content, composition, and Symbiodinium density are not only the result of photobiological adaptation but also the result of strong selective pressures on the interactions between the hosts and their different Symbiodinium assemblages (Baker 2003) to optimize light harvesting at these remarkable depths.
Strikingly, L. hawaiiensis was the dominant species below the 1% optical depth where irradiance levels are lower than 1% of the surface irradiance. This species exhibited a highlyspecialized association with Symbiodinium haplotype COI-1, which had the highest variability in photobiology with depth compared to other species within the Leptoseris complex. In this species, dark-adapted yield increased with depth, whereas values for rETR max , α, and E k declined with depth. At the deepest depth range (115-125 m), we found lower rates of photosynthetic electron transport and unexpectedly lower efficiency at subsaturating irradiances (α). Similar results were found in M. cavernosa , where rETR max and gross primary productivity decreased with depth (3-90 m depth range). For L. hawaiiensis, Symbiodinum size and Chl c : Chl a decreased with depth, suggesting genetic or energetic limitations on cell parameters and/or the amounts of pigments that can be produced. This could enforce less reliance on light capture for carbon gain by the host.
Photoprotective pigments had different patterns with depth in L. hawaiiensis. As expected, the relative xanthophyll pool : Chl a pool decreased with depth. However, β-carotene in L. hawaiiensis increased with depth, indicating that it is more likely to play a role as a structural component of the lightharvesting complex rather than as a photoprotective pigment. Further research including collections in different seasons is necessary to quantify the physiological flexibility of Leptoseris spp. over larger temporal scales and examine whether pigment composition and relative abundance change as light becomes more available with higher sun declination and longer day length. It is important to note that L. hawaiiensis photosynthesizes in a habitat with approximately half of the light available (~6 μmol photons m −2 s −1 ) at the 1% surface irradiance. This is a remarkable system that warrants further photosynthetic research and a re-evaluation of our understanding of the mechanisms that set lower limits of the euphotic zone and the strategies evolved by these endemic species to persist in these environments.
This study reveals remarkable species differentiation by Symbiodinium and Leptoseris under extreme light limitation in the mesophotic zone including changing pigment quantities and composition, photochemistry parameters, and/or Symbiodinium type, density and size. Chlorophyll fluorescence RLC measurements should be interpreted with caution given the caveats and assumptions involved with data collection and interpretation (Warner et al. 2010). The RLC data should not be confused with traditional oxygen-based photosynthesis to irradiance (P-E) curves, and cannot be used to infer total photosynthetic productivity. Absorption was not measured in this study but was assumed to be about 1, given the lower light environment and likely maximized light harvesting found at 65 to 125 m depths. Measurements of M. cavernosa absorbance had similar values ranging from 0.951 AE 0.010 to 0.963 AE 0.012 from 45 to 91 m depths, respectively (Lesser et al. 2009). Future studies should involve simultaneous measurements of photosynthesis and ETR, detailed absorbance and reflectance measurements, and the role of endolithic green algae in influencing the spectral signal of mesophotic corals. The use of a modified ETR equation that uses the spectral reflectance of the coral surface where fluorescence measurements are recorded and then converted to the absorbance band at 675 nm (Enríquez et al. 2005) would also be useful for comparison for future studies.
Another aspect to consider is that morphological, physiological, and behavioral adaptations may also occur in the host to control and optimize light acquisition (Maxwell and Johnson 2000). In response to light, corals can modify skeletal morphology (Muko et al. 2000;Enríquez et al. 2005;Todd 2008), produce antioxidants and fluorescent proteins (Muko et al. 2000;Enríquez et al. 2005;Todd 2008), and change tissue thickness, polyp size, density, and behavior (Porter 1976;Fitt et al. 2000;Levy et al. 2003;Wangpraseurt et al. 2014). In Hawai'i, Leptoseris in the mesophotic zone exhibits flatter morphologies with increasing depth (Kahng et al. 2012; J.L.P.-G. personal observation), nonphotosynthetic fluorescent pigments are found in the host throughout its depth range (Porter 1976;Fitt et al. 2000;Levy et al. 2003;Wangpraseurt et al. 2014), and skeletal design and structures maximize light scatter through the coral tissue (Kahng et al. 2012;Kahng 2014). The upper side of Leptoseris skeletons has ordered rows of concave cavities that increase the probability of light scattering to maximize light capture (Kahng et al. 2012;Kahng 2014).

Energy reserves and acquisition
Despite dynamic changes in photophysiology, total host lipids were not significantly different across depths, suggesting that lipid acquisition is maintained by nonphotosynthetic sources at some depths. As shown by Muscatine et al. (1989) and observed here for Leptoseris spp. (Fig. 5), the more rapid depletion of 13 C in host tissue of Leptoseris spp. compared to Symbiodinium indicates that animal tissue at depth incorporates carbon from other sources, including heterotrophy and/or dissolved organic carbon, in addition to photosynthesis. Moreover, the relative similarity of carbon isotopic composition between host and Symbiodinium at the shallowest depths suggests that heterotrophic inputs do not become significant until well into the mesophotic realm (120-130 m). Similar patterns were observed in M. cavernosa in the Caribbean where δ 13 C values of host and Symbiodinium only differed at the deepest collection site (91 m), suggesting less translocation of photosynthates to the host and larger dependency by the host cells on heterotrophy and/or other sources of carbon at depth .
The Leptoseris spp. complex consists of several cryptic species (Pochon et al. 2015); however, species data were not available for the isotopic samples presented (Fig. 5; Supporting Information Table S1). Therefore, it is possible that the stable carbon isotopic patterns that we observed in host and symbiont across depth may be the result of differences in heterotrophic plasticity as previously described (Muscatine et al. 1989) or differences in lipid production and storage among species, as lipids typically lower the δ 13 C values of host tissue (Deniro and Epstein 1977;Alamaru et al. 2009). Although we cannot confirm the species analyzed in the isotopic samples, our study suggest that greater differences in δ 13 C across depth between host and symbiont tissues are more likely influenced by preferential reliance on heterotrophic sources at depth than differences in storage or production of lipid content between species (Supporting Information Table S4; Fig. 6). This is further supported by C : N depth-patterns in host and Symbiodinium. Lower C : N ratios in the host at deeper depths indicate higher nitrogen content from heterotrophic activity and/or less carbon translocated to the host by Symbiodinium due to lower light levels at depth and increased nitrogen limitation (Dubinsky and Jokiel 1994).
The δ 15 N values were higher in the host than in Symbiodinium at all depths. This trend is consistent with previous observations in M. cavernosa (Muscatine and Kaplan 1994;Lesser et al. 2010), S. pistillata and Favia favus (Alamaru et al. 2009). Enriched stable nitrogen isotopic signatures in the host may be attributable to protein catabolism and excretion of isotopically light ammonium (Deniro and Epstein 1977;Alamaru et al. 2009) and/or metabolic fractionation related to changes in trophic level (Muscatine and Kaplan 1994). Because corals use the products of photosynthesis by Symbiodinium, an isotopic enrichment in the host is expected.
In Symbiodinium, δ 15 N values decreased with depth, whereas host δ 15 N values remained constant across depths (Fig. 5). Muscatine and Kaplan (1994) found a similar trend of depth-related depletion of δ 15 N in Symbiodinium of several species of Jamaican scleractinian corals that corresponded to decreasing nitrogenspecific growth rates in deeper waters. Likewise, the Symbiodinium of S. pistillata and F. favus decreased with depth, whereas the host tissue remained the same across depth (Alamaru et al. 2009). The mesophotic depths in the 'Au'au Channel correspond to the seasonal thermocline in that region (Pyle et al. 2016), which may serve as a location for turbulent mixing and upwelling of deep nutrients. During the warmer months (September-November), however, the thermocline is strongest (Pyle et al. 2016) and during this period vertical mixing of nutrients from below may be limited. This pattern of 15 N depletion in the symbiont tissue with depth, while there is no change in host tissue, is expected if corals are primarily feeding on isotopically heavier allochthonous sources of particulate organic matter (POM) at mesophotic depths (> 100 m).
In L. fragilis, POM feeding is possible because their gastrovascular system works like a filtration system. Water flow enters through the mouth of L. fragilis and leaves the body through microscopic pores in the oral epithelia. Moreover, nematocysts are abundant and present in this species (Schlichter 1991), suggesting that feeding by predation of microzooplankton could be another important strategy for nutrient acquisition in Leptoseris from Hawai'i. Further research is necessary to fully characterize the gastrovascular system in this species and examine the isotopic signature of zooplankton and POM in the mesophotic zone; this will help to better understand the heterotrophic capacity of this important genus across depths.

Conclusions
Species of Leptoseris in the mesophotic zone in Hawai'i have different associations with dinoflagellate symbionts that can contribute to the colony's capacity to acquire energy photoautotrophically. Our results show that the distribution of these species is a consequence of host-Symbiodinium specialization, physiological plasticity as well as photoadaptation across species. However, species living at deeper depths also showed a capacity to acquire energy heterotrophically, possibly by filter feeding (Schlichter 1991) and/or feeding on detritus and/or dissolved organic matter as reflected by no change in total lipid concentrations across depth and supported by differences between host and Symbiodinum isotopic values. These findings serve as a foundation to study physiological flexibility in the mesophotic zone and help us to better understand the ecology and resilience of these understudied but highly important native ecosystems.