Mesophotic corals in Hawai‘i maintain autotrophy to survive low-light conditions

Abstract

In mesophotic coral ecosystems, reef-building corals and their photosynthetic symbionts can survive with less than 1% of surface irradiance. How depth-specialist corals rely upon autotrophically and heterotrophically derived energy sources across the mesophotic zone remains unclear. We analysed the stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotope values of a Leptoseris community from the ‘Au‘au Channel, Maui, Hawai‘i (65–125 m) including four coral host species living symbiotically with three algal haplotypes. We characterized the isotope values of hosts and symbionts across species and depth to compare trophic strategies. Symbiont δ¹³C was consistently 0.5‰ higher than host δ¹³C at all depths. Mean colony host and symbiont δ¹⁵N differed by up to 3.7‰ at shallow depths and converged at deeper depths. These results suggest that both heterotrophy and autotrophy remained integral to colony survival across depth. The increasing similarity between host and symbiont δ¹⁵N at deeper depths suggests that nitrogen is more efficiently shared between mesophotic coral hosts and their algal symbionts to sustain autotrophy. Isotopic trends across depth did not generally vary by host species or algal haplotype, suggesting that photosynthesis remains essential to Leptoseris survival and growth despite low light availability in the mesophotic zone.

1. Introduction

For photosynthetic organisms, resource competition theory suggests that niches are differentiated by the use of limiting resources, such as nutrients and light, across a heterogeneous habitat [1,2]. Niche differentiation in reef corals can be driven by species-specific patterns in depth zonation [3], colony morphology [4,5] and nutrient acquisition [6–8]. Most shallow reef-forming corals depend on photosynthetic algal endosymbionts (family Symbiodiniaceae [9]) for energy from autotrophy [10]. To acquire nutrients, corals can also consume seston heterotrophically, including detritus and zooplankton, through oral openings in modular units called polyps [6,11,12]. The mutualisms between corals and their algal symbionts can develop distinct, mixotrophic niches in reef ecosystems defined by reliance on autotrophic versus heterotrophic energy sources [6]. These niches enable reef-building corals to construct the most biologically diverse marine ecosystems on the planet, supporting tens of millions of people worldwide with food, coastal protection and other ecosystem services [13].

Comparisons of stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopes between host tissue and algal symbionts have been used to assess coral trophic strategies [3,4,12,14,15]. For corals reliant on autotrophy, host δ¹³C and δ¹⁵N closely follow those of the symbionts because carbon and nitrogen are translocated to the host following photosynthesis [3,12,16]. During heterotrophy, corals may rely on seston more than symbiont photosynthates [17]. When seston isotopic values differ from Symbiodiniaceae values, increases in heterotrophy and decreases in autotrophy cause host δ¹³C and δ¹⁵N to deviate from symbiont values [3,6,12,14,15,18,19]. Muscatine et al. [12] reported increasing differences between host and symbiont δ¹³C with depth (1–50 m), hypothesizing that, as depth increases, symbiotic corals rely relatively less on autotrophy and more on heterotrophy for carbon. However, little is known about how trophic dynamics vary among communities of depth-specialist corals, particularly at depths greater than 60 m [3,7].

Mesophotic coral ecosystems are characterized by communities of corals and other organisms living from approximately 30 m to more than 150 m in depth [20]. Despite limited light availability and changing spectral composition with depth, photosynthetic corals are crucial members of the mesophotic community [15], surviving with less than 1% of surface irradiance [3,21]. Mesophotic corals have adapted to form vast reefs at low light levels through flattened colony morphologies [4,21,22], specialized skeletal micro-morphologies [5,23,24] and pigment concentrations [3,25]. While these photoadaptations may enhance autotrophy in mesophotic environments, numerous studies have hypothesized that mesophotic corals adjust trophic strategies between autotrophy and heterotrophy to inhabit deeper depths [3,7,8,15,21,26,27]. However, the extent that mesophotic reef corals rely on heterotrophy versus autotrophy with increasing depth is debated [16,27,28]. Recent evidence of anthropogenic impacts on mesophotic reefs, including elevated temperatures [29] and nutrient pollution [30], emphasize the need to conserve these vast ecosystems. Effective management strategies will require a better understanding of energy uptake, allocation and niche differentiation in mesophotic corals.

Clarifying the relative trophic contributions to corals at mesophotic depths could explain how they survive in light-limited environments. Most studies characterizing trophic dynamics in mesophotic corals have described their heterotrophic and autotrophic capacity relative to shallow reef counterparts [4,5,7,8,12,15]. While most isotopic studies support that heterotrophy supplants autotrophy in corals at increasing depths [27], variability in host–symbiont carbon exchange and symbiont type can obscure trophic trends across depth and light gradients [3,16,28]. Isotopic studies reveal conflicting trends across mesophotic depths, suggesting that some Caribbean and Red Sea corals may not increase rates of heterotrophy or reduce reliance on autotrophy from 5 m to at least 60 m [4,14,31]. δ¹⁵N values can decrease with increasing depth for both hosts and symbionts assessed separately [11], contrary to trophic increases in host δ¹⁵N expected under increased heterotrophy [6]. Some studies reported no change in host δ¹⁵N and symbiont δ¹⁵N across the mesophotic zone [4,15]. Prior studies have identified several hypotheses to explain changes in δ¹⁵N with depth. Variations in light quality/availability [32,33], nitrogenous waste exchange between host and symbiont cells [4,15,19], and nitrogen-specific rates of symbiont growth [11] impact the retention or excretion of ¹⁵N and ¹⁴N and resulting δ¹⁵N values. However, it remains unclear how the acquisition and/or exchange of nitrogen and carbon contribute to the trophic strategies of mesophotic corals.

Assessing trophic status across mesophotic depths is complicated by various genotypic combinations of host species and algal symbionts. Depth zonation within mesophotic coral communities occurs based on host species [23,34,35] and algal symbiont diversity [8,25,35]. While some coral species maintain one haplotype of algal symbionts across mesophotic depths [25], others associate with multiple haplotypes across depths [3,35]. Variable combinations of host species and symbiont haplotypes may create divergent photophysiological and trophic strategies across the mesophotic zone.

The Hawaiian Archipelago's Leptoseris communities create the most spatially extensive mesophotic coral ecosystems on record, dominating from 60 to 160 m [36,37] with up to 100% live coral cover at some depths [34,38]. Phylogenetic analyses of the Leptoseris community (65–150 m) have identified six host species (L. hawaiiensis, L. yabei, L. papyracea, L. scabra, L. tubulifera and Leptoseris sp. 1) [23,35] and three algal haplotypes of the genus Cladocopium [35]. Depth partitioning of Leptoseris species has been linked to different photoacclimatization strategies including increases in chlorophyll a, symbiont density and dark-adapted yield in species persisting through deeper depths, which could sustain autotrophy at depth [3].

We investigated how trophic strategies influence niche partitioning across depth in a Leptoseris mesophotic reef community in the ‘Au‘au Channel off Maui, Hawai‘i (65–125 m). We analysed mesophotic Leptoseris colonies representing four host species living symbiotically with three algal haplotypes ([35]; electronic supplementary material, figure S1A). The six combinations of host species and algal haplotypes varied in distribution across depth ([35]; electronic supplementary material, figure S1B). While Padilla-Gamiño et al. [3] reported stable isotope values from a Leptoseris sample set collected from the ‘Au‘au Channel in 2009–2010, the coral and algal genotypes were unknown. Here, we measured the δ¹³C and δ¹⁵N of hosts and symbionts across genotype and depth collected from a separate cruise in February–March 2011 to assess differences in trophic strategies among Leptoseris species.

2. Methods

(a) . Colony collection and genetic identification

Fragments of cryptic Leptoseris spp. colonies (n = 47) were collected from depths of 65–125 m from the ‘Au‘au Channel offshore of Olowalu, West Maui from 27 February to 3 March 2011 by submersible (Pisces IV and Pisces V). Irradiance was approximately 5–75 µmol photons m⁻² s⁻¹ [3]. Collections are described in Pochon et al. [35] and Padilla-Gamiño et al. [3]. Briefly, coral fragments (approx. 5–8 cm plates) were collected haphazardly from the centre of Leptoseris reefs across depth; colonies at the same depth were interspaced by at least 10 m. Collected samples were processed and frozen at −80°C within 3–9 h of ascent aboard the R/V Ka‘imikai-O-Kanaloa. Leptoseris colonies analysed in this study were identified by host species and symbiont haplotype in Pochon et al. [35]. Pochon et al. identified each colony by host species using COX1–1-rRNA introns and by symbiont haplotype using algal symbiont COI mitochondrial markers, distinguished by ITS2 community sequence profiles [3,35]. We referred to COI sequence types because we lacked the samples required to reanalyse colony fragments for individual ITS profiles. Since COI sequence types represented distinct genetic groups of symbionts (i.e. haplotypes, with some overlap in ITS2 assemblages) within Cladocopium [35], we labelled colonies by COI haplotype 1–3. We studied the subset of the colonies in Pochon et al. [35] with remaining tissue available for stable isotope analyses.

As presented in Pochon et al. [35], our samples represent four Leptoseris host species (L. hawaiiensis, L. scabra, L. tubulifera and Leptoseris sp. 1) and all three algal haplotypes (electronic supplementary material, figure S1A). All colonies hosted only one algal haplotype; several Leptoseris species can host the same algal haplotype and different colonies of the same species can host different algal haplotypes [35]. Eleven colonies were L. scabra (65–99 m); ten hosted haplotype COI-2 and one hosted COI-3 (electronic supplementary material, figure S1B). Eleven colonies were L. tubulifera (73–99 m), hosting haplotype COI-2. Fourteen Leptoseris sp. 1 colonies were found across the widest depth range (67–123 m); five hosted haplotype COI-2 (67–99 m) and nine hosted COI-3 (67–123 m). Eleven L. hawaiiensis were found only at the deepest depths of 123–125 m, all hosting haplotype COI-1. Notably, all L. hawaiiensis and one Leptoseris sp. 1 colony were collected past the depth of 1% surface irradiance [3] (112 m; electronic supplementary material, figure S1B), traditionally considered the depth limit of photosynthetic corals [21].

(b) . Stable isotopic analyses

For each fragment (approx. 5–8 cm), coral tissue was removed from the skeleton using a pressurized airbrush and deionized water spray. We separated each colony into host and symbiont fractions prior to isotopic analyses [3,4,12,14,15]. The slurry was ground in a tissue homogenizer and centrifuged at 2880 relative centrifugal force (RCF) for 10 min to separate host and symbiont cells into distinct layers [39]. Residual skeletal material pelleted into its own layer below the symbionts during centrifugation. Routine microscope checks of symbiont samples confirmed a lack of skeletal contamination in the symbiont cells; to avoid altering the isotopic signatures, we did not acidify samples [40]. Host and symbiont fractions were separated by microspatula and pipetted into pre-weighed tin capsules, dried in an oven at 60°C for at least 24 h [41], and reweighed. All capsules were folded into compact cubes and combusted using an Elementar Pyrocube elemental analyser; resulting CO₂ and N₂ gases were analysed using an Elementar Isoprime100 isotope ratio mass spectrometer. Host and symbiont δ¹³C values were reported relative to Vienna Peedee belemnite (vPDB) limestone standard (δ¹³C = per mil deviation of the ratio of stable carbon ¹³C : ¹²C relative to vPDB) and δ¹⁵N values were reported relative to air (δ¹⁵N = per mil deviation of the ratio of stable nitrogen ¹⁵N : ¹⁴N relative to air). Standards were analysed, including B2150 (EA Consumables, LLC, Marlton, NJ, USA), internal elk tissue, dogfish muscle reference material (DORM), and bird feathers, with a precision of ± 0.1‰ for δ¹³C and ± 0.3‰ for δ¹⁵N. Approximately 21% of host and symbiont samples (n = 10 each) were analysed in duplicate.

(c) . Statistical analysis

For statistical analyses of the Leptoseris community, similar to Padilla-Gamiño et al. [3], we included all Leptoseris colonies across the collection depth range (65–125 m). Our sampling location enabled us to collect L. hawaiiensis only from 123 to 125 m; L. hawaiiensis was the only species collected past 100 m except for one Leptoseris sp. 1 colony. Although L. hawaiiensis is regarded as a depth specialist in the lower mesophotic zone [3,35], its depth range (80–130 m; [23]) overlaps with the other species. In addition, while our L. hawaiiensis were found exclusively with haplotype COI-1, other Leptoseris colonies containing haplotype COI-1 have been identified at depths as shallow as 95 m [35]. Since the known habitat range of L. hawaiiensis and its algal haplotype overlap with the other study species, we included it in all community assessments and statistically tested its influence on community trends.

To identify outliers, histogram distributions of all δ¹³C and δ¹⁵N values were plotted by fraction type (host or symbiont) in JMP Pro 14; residuals and Cook's distances were used to confirm outliers (electronic supplementary material, table S1). We excluded outlier values that exceeded the interquartile range by more than ±1.5 times (n = one δ¹³C from a host sample; one δ¹³C from a symbiont sample; two δ¹⁵N from host samples and two δ¹⁵N from symbiont samples were excluded).

To initially assess the trophic relationship between hosts and symbionts across the Leptoseris community, we compared δ¹³C and δ¹⁵N values using the Stable Isotope Bayesian Ellipses in R (SIBER) package (R version 4.0.2) [42]. SIBER visualizes host and symbiont isotopic niches to determine the trophic niche overlap (i.e. extent of symbiotic resource sharing) between coral hosts and their algal symbionts [6,27]. Since SIBER can only effectively compare groups with the same or different sample sizes of at least 20 samples each [43], we pooled samples (n = 47 colonies) at the genus level [6] and used SIBER to assess broad trophic trends across the community. We generated 4000 standard ellipse areas corrected for sample size (SEA_C) with the SEA_C encompassing 40% of the dataset variation [6]. We conducted a residual permutation procedure of 100 000 permutations of the δ¹³C and δ¹⁵N values to verify isotopic differences between the centroids of host and symbiont datasets [6,27,44].

We compared ellipse overlap for all Leptoseris colonies against a proposed spectrum of coral trophic strategies using percentage cut-offs for heterotrophy (less than or equal to 10% overlap), mixotrophy (10–70% overlap) and autotrophy (greater than or equal to 70% overlap) [6]. We calculated mean per cent overlap between host and symbiont ellipses, with per cent overlap defined as the overlapping isotopic area divided by the host ellipse area [6,27]. To evaluate the potential effect of collection depth on Leptoseris host and symbiont isotope values, we replotted values pooled into 5 m depth bins. Sample sizes among host species and symbiont haplotypes (electronic supplementary material, figure S2) were insufficient to analyse independently by SIBER.

To investigate trophic relationships among Leptoseris host species and symbiont haplotypes across depth, we conducted simple linear regressions in JMP Pro 14. To investigate niche partitioning in Leptoseris colonies across depth, we tested for significant linear relationships between isotope values and depth for each host species or symbiont haplotype. We tested for linear relationships of host δ¹³C, symbiont δ¹³C and δ¹³C_host-δ¹³C_symbiont versus depth by host species and symbiont haplotype. We repeated the same analyses for host δ¹⁵N, symbiont δ¹⁵N and δ¹⁵N_host-δ¹⁵N_symbiont versus depth (electronic supplementary material, table S2). Sample distributions were checked for normality and homoscedasticity when tested for statistical significance at the two-sided α = 0.05 level. To test the influence of L. hawaiiensis on community trends, we repeated community-wide regressions excluding L. hawaiiensis (electronic supplementary material, table S3).

We compared the slopes of significant regressions across depth with one another by host species and algal haplotype using a multi-variable regression analysis. To determine whether there was a significant difference in the slope of isotope values based on host species or symbiont haplotype, each model included an interaction term between collection depth and either host species or symbiont haplotype. The error rate for the two-sided p-value for the interaction term was α = 0.10 accounting for multiple analyses on the same dataset. When interaction term p < 0.10, the model was used to determine whether there were significant differences among regression slopes (two-sided α = 0.05) based on colony host species or algal haplotype (electronic supplementary material, table S2). The same multi-variable regression analysis was applied to the community dataset excluding L. hawaiiensis to determine whether this deeper species had a notable effect on community trends (electronic supplementary material, table S3).

3. Results and discussion

(a) . Heterotrophy plays a dominant role in mesophotic coral nutrition

For the Leptoseris community, host and symbiont isotopic niche spaces were significantly different from one another (residual permutation procedure p < 0.001), with 3.63% of the SIBER host ellipses overlapping the symbiont ellipses (figure 1a). A host–symbiont ellipse overlap less than or equal to 10% of the host SEA_C suggests a heterotrophy-dominant strategy for this community [6], supporting that heterotrophy is essential to mesophotic reef-building corals including Leptoseris [3,4,7,8,31]. While heterotrophy is an important energy source for mesophotic corals, deep-water Leptoseris have been found with only vestigial tentacles that cannot extend [24,45]. These corals might obtain heterotrophic nutrients from the environment by moving their vestigial tentacles to circulate organic matter into their polyps. A similar mechanism occurs in Leptoseris fragilis, which lacks tentacles [21,26] and instead pumps water through the mouth and out of epidermal pores. This mechanism generates a flow-through filtration system for consuming particulate organic matter (POM) and dissolved organic matter (DOM) like a sponge [26]. Coral species such as Mycetophyllia reesi, which lacks tentacles, capture food particles with mucus secretions [46]. Mucus channels can regulate shared feeding across colony polyps [47] and may facilitate feeding in Leptoseris corals. While the plating morphology of Leptoseris is probably adapted to optimize light capture at depth [11,14], plating may enhance passive feeding on sinking particles of detritus that collect on colony surfaces [7].

Figure 1.

Heterotrophy plays a dominant role in Leptoseris nutrition, but host and symbiont stable isotope values converge across mesophotic depths. (a) SIBER model outputs for δ¹³C, δ¹⁵N values of all samples with dotted lines denoting convex hulls of hosts (orange) and symbionts (teal). Solid rings of the same colours denote standard ellipse areas corrected for sample size (SEA_C). (b) Isotopic values for hosts and symbionts pooled by 5 m collection depth bins (means ± s.e.; electronic supplementary material, table S4).

While SIBER suggests that heterotrophy is a major energy source for Leptoseris corals, various factors could have obscured the contribution of autotrophy to colony survival and growth. Our colonies were collected during the reduced seasonal photoperiods of winter–early spring. As observed in temperate symbiotic corals with seasonal variation in irradiance and photosynthetic rate, corals in low irradiance can still rely on autotrophic carbon for metabolic energy even when irradiance is insufficient to sustain host growth [17]. Additionally, SIBER trophic strategy classifications were based on shallow reef corals living at a maximum depth of 5 m [6] where up to 90% of carbon from photosynthates may be translocated to the coral host [48]. At 5 m, trophic niche overlaps of up to 94% indicate a high degree of resource sharing through photosynthesis [6]. Photosynthetic rates that facilitate coral growth at shallow depths are unlikely in low-irradiance mesophotic environments. Red Sea Leptoseris colonies (82–90 m) collected in autumn exhibited photosynthetic rates up to 10 times lower than most shallow (10 m) symbiotic corals from the same reefs [7]. Photosynthetic rates were too low to sustain daily respiration rates, which were supplemented by heterotrophy [7]. Trophic classifications established by SIBER may not account for photosynthetic contributions to metabolism required for growth, or how autotrophic contributions to growth and metabolism vary in the mesophotic zone [7,16,24,28].

Grouping by 5 m collection depth bins indicates that δ¹³C values for both fractions decreased with depth (65–125 m), with the mean difference between host and symbiont fractions varying by 0.3–0.9‰ (figure 1b; electronic supplementary material, table S4). Host and symbiont δ¹⁵N converged with deeper depths, as the mean difference between host and symbiont decreased by 2.5‰ from 65 to 125 m (figure 1b; electronic supplementary material, table S4). SIBER analyses of shallow (5 m) corals indicated that δ¹⁵N was a stronger determinant of trophic niche overlap than δ¹³C, which can have high overlap despite distinct host and symbiont ellipse areas [6]. SIBER could not be used to assess the effect of depth or host/symbiont genotype on isotopic data due to insufficient sample size (electronic supplementary material, figure S2). To compare the trophic profile of each genotypic division in the Leptoseris community across its depth range and to contextualize changes in metabolism [16,24], we characterized the independent relationships of δ¹³C and δ¹⁵N values across depth.

(b) . Autotrophy remains integral to colony survival across mesophotic depths

Host and symbiont δ¹³C decreased with increasing depth for the Leptoseris community (figure 2a,b; electronic supplementary material, table S2), as observed in other mesophotic corals [4,12,14,15]. However, host and symbiont δ¹³C did not diverge [27]; instead, host δ¹³C was consistently approximately 0.5‰ lower than symbiont δ¹³C across depth (figure 2c). Decreases in δ¹³C of suspended POM across depth associated with bacterial remineralization are expected to decrease host δ¹³C via heterotrophy at deeper depths [12,49]. As symbionts primarily obtain carbon from the host (e.g. via metabolites including CO₂), the decrease in symbiont δ¹³C across depth is probably linked to the decrease in host δ¹³C [12,14]. Across depth, heterotrophy provides additional carbon to the host that is not shared with the symbionts, explaining why host δ¹³C remains more negative than symbiont δ¹³C [3,12,27]. However, the consistent relationship between host and symbiont δ¹³C suggests no relative change in photosynthates translocated from symbionts to hosts across depth [12,16,27]. Grouping by host species (figure 2d–f) and symbiont haplotypes (figure 2g–i) displayed a consistent trend of decreasing host and symbiont δ¹³C values with depth (electronic supplementary material, table S2). Leptoseris hawaiiensis and COI-1 were only found from 123 to 125 m (figure 2d–f,g–i); omitting L. hawaiiensis from our regressions of δ¹³C versus depth across coral hosts and algal haplotypes increased the steepness of host and symbiont slopes by 33% and 34%, respectively (electronic supplementary material, table S3). However, the slopes exhibited across coral hosts and algal haplotypes remained statistically similar to one another without L. hawaiiensis (model F = 16.7, model p < 0.0001, slope interaction p = 0.71).

Figure 2.

Carbon exchange between hosts and symbionts remains constant for most Leptoseris genotypes across depth, supporting sustained reliance on autotrophy. δ¹³C (‰) of (a) host, (b) symbionts and (c) differences between host and symbiont isotopic values for the Leptoseris community across depth (m). Points indicate means ± s.e., binned by collection depth and jittered for clarity. Solid lines show significant linear regressions (p < 0.05; electronic supplementary material, table S2). Dashed lines in a and b represent the overlayed regressions for symbionts and hosts, respectively. The red shaded area indicates the difference between host and symbiont δ¹³C values plotted in c. Trend lines in a and b show host δ¹³C was consistently about 0.5‰ lower than symbiont δ¹³C across depths. The same data by (d–f) host species or (g–i) symbiont haplotype show host and symbiont δ¹³C and the difference between them. Re-categorizing points changed some of their values and total counts across plots. Dashed lines in d–e and g–h indicate the community trends shown in a and b, respectively. Solid lines show significant linear regressions for host species (d–f) and algal haplotype (g–i) (p < 0.05; electronic supplementary material, table S2). Zero-lines (grey) in c, f and i indicate no isotopic difference between host and symbiont fractions.

The only exception to the trends in δ¹³C across depth was that the negative δ¹³C_host-δ¹³C_symbiont values decreased with increasing depth in COI-2 colonies (figure 2i; electronic supplementary material, table S2). The increasing difference in δ¹³C between COI-2 fractions across depth suggest a haplotype-specific trophic shift for COI-2 colonies towards increased reliance on heterotrophy [27]. Lower dark-adapted photosynthetic yield in some species harbouring COI-2 symbionts (70–100 m) relative to COI-1 at the deepest depths (110–120 m) may indicate a reduced capacity for autotrophy [3], possibly explaining why COI-2 colonies were not found below 100 m. The negative isotopic shift of approximately 1‰ difference for this haplotype between 65 and 100 m is notably less than the negative shift in δ¹³C_host-δ¹³C_symbiont reported (approx. 2–8‰) for other mesophotic corals [12,14,15,27]. For the rest of the community, the similarity in slope for hosts and symbionts (electronic supplementary material, table S2), despite drastic decreases in available light, suggests that autotrophy remains crucial for Leptoseris across depth [12,15]. Differences between host and symbiont δ¹³C for this Leptoseris community (figure 2c,f,i; mean (δ¹³C_host-δ¹³C_symbiont) ± s.e. = −0.5 ± 0.1‰ across depths) are similar to those reported for some shallow reef corals at just 5 m (mean (δ¹³C_host-δ¹³C_symbiont) approximately 0.4–0.6‰ for three species, [6]). Differences in trophic strategy among those shallow-reef species were determined by differences in δ¹⁵N between host and symbiont fractions [6]. Therefore, nitrogen must be considered with δ¹³C to substantiate and contextualize relative trophic rates for mesophotic Leptoseris.

δ¹⁵N values across depth suggest nitrogenous resources are shared between Leptoseris hosts and their algal symbionts through the mesophotic zone. Unlike previous studies, where host and symbiont δ¹⁵N diverged [3] or remained constant [4,15] with increasing depth, we found that host and symbiont δ¹⁵N converged through the mesophotic zone (figure 3a,b; electronic supplementary material, table S2). For the Leptoseris community, host δ¹⁵N decreased with depth (figure 3a), while symbiont δ¹⁵N increased with depth (figure 3b). Consequently, the difference between host and symbiont δ¹⁵N decreased with increasing depth, approaching zero (figure 3c) and causing host and symbiont δ¹⁵N to converge at an average of approximately 4‰ for both fractions at 125 m (figure 3a,b). Like δ¹³C, host and symbiont δ¹⁵N trends across depth were well-conserved across species and haplotypes. While not all colony host species/algal haplotypes produced significant regressions for δ¹⁵N across depth, all significant regressions matched the community trends, and all non-significant relationships conformed to the overall pattern (figure 3d–i; electronic supplementary material, table S2). Omitting L. hawaiiensis did not change the statistical significance of the trends for δ¹⁵N regressions across coral hosts and algal haplotypes across depth, and host and symbiont δ¹⁵N regressions containing all other species/haplotypes combined maintained negative and positive slopes across depth, respectively (electronic supplementary material, table S3).

Figure 3.

Host δ¹⁵N and symbiont δ¹⁵N converge across depth, providing further evidence for sustained autotrophic reliance through the mesophotic zone. δ¹⁵N (‰) of (a) host, (b) symbionts and (c) differences between host and symbiont isotopic values for the Leptoseris community across depth (m). Points indicate means ± s.e., binned by collection depth and jittered for clarity. Solid lines show significant linear regressions (p < 0.05; electronic supplementary material, table S2). Dashed lines in a and b represent the overlayed regressions for symbionts and hosts, respectively. The red shaded area indicates the difference between host and symbiont δ¹⁵N values plotted in c. Trend lines in a and b show that host δ¹⁵N decreased and symbiont δ¹⁵N increased across depths, converging at approximately 4‰ at 125 m. The same data by (d–f) host species or (g–i) symbiont haplotype show host and symbiont δ¹⁵N and the difference between them. Re-categorizing points changed some of their values and total counts across plots. Dashed lines in d,e and g,h indicate the community trends shown in a and b, respectively. Solid lines show significant linear regressions for host species (d–f) and algal haplotype (g–i) (p < 0.05; electronic supplementary material, table S2). Zero-lines (grey) in c, f, and i indicate no isotopic difference between host and symbiont fractions.

Our δ¹⁵N results contradict any expected shift to heterotrophic dependence at depth. SIBER analyses suggested that heterotrophy is the primary nutritional source for our Leptoseris community; however, if Leptoseris colonies relied more on heterotrophy relative to autotrophy with increasing depth, δ¹⁵N values of hosts and symbionts should become more distinct. Nitrogenous resources acquired heterotrophically by the host are expected to decouple host and symbiont δ¹⁵N values as autotrophic contributions from the symbionts to the host decrease [6,18,19]. The δ¹⁵N of both suspended and sinking POM are expected to increase across depth as a function of processes including ¹⁴N-preferred bacterial degradation [49,50]. This degradation probably contributed to δ¹⁵N increases of 1‰ in coral skeletons observed from 5 to 105 m in Palau, mirroring increases in the δ¹⁵N of suspended POM [49]. Around Maui, increases in mean macroalgal δ¹⁵N from shallow (0–30 m; 2.4‰) to mesophotic depths (30–117 m; 2.9‰) were attributed to submarine groundwater discharge potentially linked to anthropogenic nutrient pollution [30]. All such environmental increases in source δ¹⁵N and hypothesized increases in heterotrophy through the mesophotic zone [27] should increase host δ¹⁵N alone or increase both host and symbiont δ¹⁵N across depth. Yet, in our Leptoseris community, host δ¹⁵N did not increase for any genotype across mesophotic depths (figure 3a,d,g). Along with δ¹³C, these δ¹⁵N findings suggest that relative rates of heterotrophy are not increasing or replacing autotrophy as depth increases. If heterotrophy decreased relative to autotrophy, δ¹³C values between fractions should have become more similar across depth. Instead, both δ¹³C and δ¹⁵N regressions reveal a consistent reliance on autotrophy across depth.

These δ¹³C and δ¹⁵N trends across depth differed from Padilla-Gamiño et al. [3], which found that Leptoseris host and symbiont regressions for both stable isotopes diverged with increasing depth, supporting increased heterotrophy at depth. In that study, Leptoseris stable isotope samples were collected earlier, in January–April 2009–2010, and may have included any of the six Leptoseris species from the mesophotic zone of the ‘Au‘au Channel, including L. yabei and L. papyracea. Both were absent from our samples collected in February–March 2011. Seasonal variability in the environment at collection time may add to isotopic differences between studies. Over half (approx. 54%) of isotopic samples from the prior study [3] were collected in April when subsurface current speed can increase in Hawaiian mesophotic habitats approximately 50 m deep [38]. This current can increase mixing of zooplankton and detritus across mesophotic depths [27,51], potentially increasing heterotrophy relative to our winter collections. Differences among submersible collection sites and microhabitats in the ‘Au‘au Channel between years may have compounded trophic differences between these two distinct collections due to spatial variations in submarine groundwater discharge [30] and thermal variability [16].

Our results are supported by studies of Carribean Montastraea cavernosa corals across a shallow-to-mesophotic gradient. Mesophotic M. cavernosa grow flat, plating morphologies [15,31] like Leptoseris optimizing incident irradiance at depth [31]. Decreased host δ¹³C across depth (3–91 m) was originally attributed to increased heterotrophy and reliance on zooplankton [15,27]. Subsequent analyses found that relative compositions of heterotrophic sterol markers and phytosterols from symbionts remained the same between 18–20 m and 55–60 m, indicating primary reliance on autotrophy to at least 60 m [31]. Similarly, while host and symbiont δ¹³C decreased with depth in our Leptoseris community, relative differences between host and symbiont δ¹³C did not change from 65 to 125 m (figure 2c).

Both skeletal microstructures that amplify scattered light and low polyp density that lessens self-shading support photoacclimatization of Leptoseris at extreme mesophotic depths [5,21–24]. These adaptations may benefit mesophotic corals relative to other coral morphotypes. While mounding corals maximize theoretical gross primary productivity across depth, integrated light scattering demonstrates that plating species have higher photosynthetic potential across bathymetric slopes into the mesophotic zone than mounding or branching species [52]. Plating could optimize symbiont photosystems within coral tissues at depth [53] and maximize carbon translocation under low irradiance [17], enabling autotrophic advantages over other morphotypes in these systems. Our results support that autotrophy persists in these corals across depth, possibly facilitated by plating and other morphological adaptations.

(c) . Nitrogen exchange could sustain autotrophic capacity at extreme mesophotic depths

The convergence between Leptoseris host and symbiont δ¹⁵N values with increasing depth suggests that nitrogenous resources are increasingly shared between partners at depth (figure 4a–d), similar to Stylophora pistillata [4,19] and Favia favus [4]. In coral symbiosis, hosts typically have a higher δ¹⁵N relative to their symbionts as ¹⁴N-ammonium waste is preferentially excreted or shared with symbionts (figure 4a) [19,54]. At depths less than 90 m in the ‘Au‘au Channel, current speeds of at least 10–15 cm s⁻¹ occur during the winter and may have coincided with our collections, exposing Leptoseris to abundant heterotrophic nitrogen (e.g. zooplankton/detritus) [27,38,51]. Mean δ¹⁵N ± s.e. of Leptoseris host and symbiont fractions across depth (4.5 ± 0.4‰ at 65 m, decreasing to 3.4 ± 0.4‰ at 100 m; electronic supplementary material, table S4) resemble the upper bound of 3–3.5‰ for sinking POM at 100–150 m depth off the Hawaiian Islands [50]. Leptoseris may also feed on suspended particulate nitrogen (PN) [49]; however, δ¹⁵N_PN measured at the nearby ALOHA station was much lower than host δ¹⁵N values (mean δ¹⁵N_PN = −0.3–2.2‰ from 43 to 175 m, respectively; [50]). Mean host δ¹⁵N at 65 m (approx. 6.3‰) is in the trophic range of 2.5–3.5‰ higher than sinking POM expected if these corals acquire nitrogen heterotrophically [6,50,55,56]. Therefore, nitrogen may be obtained at shallower depths by coral hosts through heterotrophic feeding while carbon is still obtained through photosynthesis, reflected in consistent δ¹³C_host-δ¹³C_symbiont across depth (figure 2). High heterotrophy on POM supplying abundant nitrogen at shallower depths could occur with relatively high rates of autotrophy requiring relatively low nitrogen exchange and producing distinct host and symbiont δ¹⁵N values.

Figure 4.

Together, δ¹³C and δ¹⁵N explain how nitrogen exchange supports autotrophy at depth. Schematic hypothesis for our results across collection depth. Leptoseris hosts can obtain carbon and nitrogen heterotrophically, most likely from POM. Symbionts can supply carbon and nitrogen to hosts autotrophically by taking dissolved inorganic carbon and nitrogen from the water column and translocating photosynthates (PS) to the host. (a) At 65 m, with abundant nitrogen (N), ${\mathrm{NH}}_4^{+}$ waste can be excreted from the coral and/or symbionts. Symbionts do not readily share/exchange N with the host due to high abundance of N. Preferential excretion of ${\hspace{0.17em}}^{14}{\mathrm{NH}}_4^{+}$ and limited N sharing increase host δ¹⁵N; as ¹⁴N is excreted, host ¹⁵N : ¹⁴N increases. Symbiont δ¹⁵N remains low as symbionts transfer less N to hosts, producing lower symbiont ¹⁵N : ¹⁴N. Heterotrophy contributes more to Leptoseris energy inputs than autotrophy (figure 1). However, relative carbon exchange between hosts and symbionts is consistent across depth (i.e. symbiont δ¹³C approximately 0.5‰ higher than host δ¹³C across depth; figure 2). Therefore, high heterotrophy on POM that supplies high nitrogen inputs at 65 m corresponds with proportionally high rates of autotrophy. (b) At 100 m, weak currents limit POM, so Leptoseris increase N exchange between host and symbionts. Excretion is limited and hosts share higher amounts of ${\hspace{0.17em}}^{14}{\mathrm{NH}}_4^{+}$ waste to the symbionts (larger ${\hspace{0.17em}}^{14}{\mathrm{NH}}_4^{+}$ arrow towards Sym). Symbionts transfer more N to hosts. N exchange homogenizes host and symbiont ¹⁵N : ¹⁴N ratios, decreasing host δ¹⁵N and increasing symbiont δ¹⁵N compared with 65 m (figures 3 and 4a). Since symbiont δ¹³C remains approximately 0.5‰ higher than host δ¹³C (figure 2), reduced heterotrophy at 100 m coincides with a proportional reduction in autotrophy, consistent with light limitation at depth. Leptoseris reefs from the ‘Au‘au Channel, Hawai‘i at (c) 86 m and (d) 100 m (submersible photographs courtesy of the Hawai‘i Undersea Research Laboratory).

By contrast, increased exchange of limited nitrogen could explain the similarity in host and symbiont δ¹⁵N values in the lower mesophotic zone (figure 4b). At depths greater than 90 m, ‘Au‘au Channel currents decrease to near-stagnant levels throughout the year [38], potentially limiting nitrogenous resources including POM [27,51]. Nitrogen limitation in both hosts and symbionts has been observed in mesophotic Stylophora pistillata (50 m) compared with shallower conspecifics (5 m) [8]. Increased flow of nitrogen exchanged from symbionts to the host should decrease host δ¹⁵N [18]. Therefore, decreased excretion of nitrogen from the holobiont and increased homogenization between symbionts and hosts could explain the convergence of host and symbiont δ¹⁵N values at depth (figure 3). Previous experimentation indicated that symbionts of Red Sea Leptoseris spp. corals (82–90 m) transferred at least 50% of N acquired from dissolved inorganic nitrogen (DIN) to their hosts after an 8-hour incubation period [7]. The capacity for Hawaiian Leptoseris to increasingly exchange nitrogen [7,31] between host and symbiont compartments may enable autotrophy at deeper depths despite reduced light availability. Lower rates of photosynthetic electron transport at the deepest collection depths [3] probably decrease photosynthate production compared with shallower colonies, as observed for Red Sea Leptoseris spp. [7]. Nitrogen limitation linked to reduced heterotrophy, matched with a light-limited decrease in photosynthate production, could explain how relative carbon exchange between hosts and symbionts seems to remain consistent while nitrogen exchange increases across depth (figures 2–4).

(d) . Implications for mesophotic Hawaiian reefs

Reliance on photosynthesis at depth may contribute fundamentally to the skeletogenesis of mesophotic reefs. Mesophotic colonies of Leptoseris fragilis from the Red Sea have maximum growth rates of 0.8 mm yr⁻¹, growing up to 8–10 cm in diameter [21]. Leptoseris colonies in Hawai‘i can grow diameters greater than 1 m with growth rates of 1–2.5 cm yr⁻¹, comparable to some shallow water corals [3,38,57]. Hawaiian Leptoseris colonies probably invest heavily in skeletal extension to increase lateral growth relative to thickness due to the stable environment of their deep habitat that is minimally impacted by wave action [38,45]. Low metabolic demand for carbon resources [16,28], coupled with increased nitrogen exchange, may enable Leptoseris to survive and grow in light-limited conditions. High skeletal growth rates observed in Hawaiian Leptoseris communities could be maintained by conserving photosynthates required for light-enhanced calcification [21,22,58,59]. Investment in skeletal microstructures that enhance scattering and amplify light for symbionts [5,22] could enable skeletal growth to synergistically fuel photosynthesis in a positive feedback loop at mesophotic depths.

Based on our stable isotope analyses, trophic strategy cannot explain any competitive advantage of one Leptoseris species relative to others at least to 125 m in Hawaiian mesophotic reefs. We did not find that depth partitioning of Leptoseris corals in the ‘Au‘au Channel corresponded to niche partitioning of trophic strategies. While heterotrophy may serve as the main trophic input of carbon and nitrogen for Leptoseris across depths, the ability to photosynthesize under low-light conditions may enable coral survival and growth throughout the mesophotic zone. Although Leptoseris species in this community share host macromorphologies [5,23,24], niche diversity of their microbial communities, including dark-adapted yields of algal symbionts [3,35] and nitrogen fixation and exchange by bacteria [60], may drive the depth partitioning of different Leptoseris genotypes [3]. Characterizing microbial diversity and function in corals across mesophotic depths could help explain the sustained autotrophy extending across host species and algal haplotypes.

This study provides a novel example of isotopic data over a wide mesophotic gradient [16,27,28] where examining carbon and nitrogen, both independently and combined, contributed to the understanding of trophic strategies that sustain coral survival in light-limited environments. Conservation strategies prioritizing water clarity by minimizing shoreline run-off could bolster the autotrophic capacity of Leptoseris colonies. An enhanced perspective of mesophotic reefs as both exponentially light-limited and persistently light-dependent coral communities could improve our ability to understand and conserve the rich biodiversity and vast extent of these largely unexplored ecosystems.

Ethics

Coral samples were collected under SAP permit 2009-72 from the Department of Land and Natural Resources, State of Hawai‘i. However, the majority of corals were collected in US Federal Waters and did not require a permit for collection.

Data accessibility

Copies of all data files have been submitted to Dryad: https://doi.org/10.5061/dryad.f1vhhmh0z [61].

Supplementary material is available online [62].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

C.H.B.: conceptualization, formal analysis, investigation, visualization, writing—original draft, writing—review and editing; J.L.P.: conceptualization, funding acquisition, project administration, supervision, writing—review and editing; H.L.S.: funding acquisition, project administration, writing—review and editing; M.S.R.: writing—review and editing; C.M.S.: funding acquisition, writing—review and editing; R.D.G.: funding acquisition; L.J.R.: conceptualization, funding acquisition, investigation, resources, supervision, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

The authors declare no competing interests.

Funding

This article includes research funded by the National Oceanic and Atmospheric Administration (NOAA) Center for Sponsored Coastal Ocean Research (Coastal Ocean Program) under award to the Bishop Museum, NA07NOS4780187 and NA07NOS478190 to the University of Hawai‘i and NA07NOS4780189 to the State of Hawai‘i; submersible support provided by the Hawai‘i NOAA Undersea Research Program (HURL); and funding from the NOAA Coral Reef Conservation Program research grants program administered by HURL under award NA09OAR4300219, project number HC08-06. Analyses were funded by grants from the National Science Foundation (NSF) Experimental Program to Stimulate Competitive Research (EPS-0903833) and Bio Oce (OCE-0752604), both to R.D.G. Work by L.J.R. and C.H.B. was supported by NSF (IOS-1655888); work by J.L.P. was supported by the Sloan Foundation and NSF (IOS-1655682, OCE CAREER-2044840); and work by M.S.R. was supported by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy, FWP number 449B, and the Agriculture and Food Research Initiative Competitive grant no. 2013-67012-21272 from the USDA National Institute of Food and Agriculture.