Hypoxia disrupts metabolism in coral and sea anemone larvae Open Access

Anthropogenic pollution is driving a decrease in seawater dissolved oxygen (DO) content across a broad range of coastal marine ecosystems including tropical coral reefs and temperate estuaries (Altieri et al., 2017; Breitburg et al., 2018; Pezner et al., 2023; Rose et al., 2024). These declines in DO can lead to severe hypoxia (DO <2 mg l⁻¹; Pezner et al., 2023; Rosenberg, 1980), which drives decreases in performance for reef-building corals and estuarine sea anemones (phylum Cnidaria), and can result in the widespread mortality of these species (Altieri et al., 2017; Johnson et al., 2021; Nelson and Altieri, 2019). Although cellular mechanisms underpinning these responses are largely unexplored in these taxa, negative physiological outcomes may result in part from metabolic disruption under hypoxia (Lee et al., 2023), as molecular oxygen (O₂) is required for numerous metabolic processes including the production of cellular energy in the form of ATP through mitochondrial oxidative phosphorylation (OXPHOS; Senior, 1988). Indeed, growing physiological evidence from corals and sea anemones demonstrates decreases in aerobic respiration rates and increases in anaerobic metabolic processes (e.g. fermentation) under short-term hypoxia in multiple species (Dilernia et al., 2024; Ellington, 1980; Glass and Barott, 2025; Glass et al., 2025; Jain et al., 2023; Linsmayer et al., 2020; Mallon et al., 2025). This suggests possibly conserved effects of hypoxia on metabolism in these taxa, though this remains an important topic for further investigation.

Compared with adult corals and sea anemones, larvae are particularly sensitive to abiotic stressors (Byrne et al., 2020; Foo and Byrne, 2017; Przeslawski et al., 2015; Putnam et al., 2017), and the higher mass-specific O₂ consumption rates in larvae compared with adults suggest that hypoxia may result in more severe decreases in performance during this important early life stage (Bean and Edmunds, 2024). Further, larval hypoxia responses may differ between taxa owing to species-specific life history traits that influence larval metabolism, such as sexual system (e.g. gonochoric versus hermaphroditic), reproductive mode (e.g. brooding versus broadcast spawning) and algal endosymbiont (family Symbiodiniaceae) presence and transmission mode (vertical or horizontal; Baird et al., 2009; Figueiredo et al., 2012; Harrison, 2011; Kitchen et al., 2020). For example, eggs and larvae of gonochoric corals are generally larger than those of hermaphroditic corals (Szmant, 1986), suggesting increased maternal provisioning in gonochoric taxa that may support larval metabolic performance under hypoxia despite the potential for a size-driven increased requirement for O₂ (Bean and Edmunds, 2024). Additionally, as symbionts can provide cnidarian larvae with dietary resources in the form of photosynthate (e.g. carbohydrates and lipids; Harii et al., 2010; Hazraty-Kari et al., 2022; Huffmyer et al., 2024; Voss et al., 2023), non-symbiotic larvae may be more sensitive to hypoxia compared with symbiotic larvae owing to the reliance of the former on the β-oxidation of egg-derived lipid stores (Figueiredo et al., 2012; Harii et al., 2007, 2010; Hazraty-Kari et al., 2022). By contrast, symbiotic larvae may be able to process symbiont-derived glucose via anaerobic glycolysis and subsequent fermentation to maintain low levels of ATP production under hypoxia (Alderdice et al., 2022), as is observed under other stressors including heat and acidification (Huffmyer et al., 2024; Sun et al., 2024). Further, symbiont photosynthesis can result in tissue hyperoxia during the day that may protect larvae during daytime hypoxia (Gardella and Edmunds, 1999; Harland and Davies, 1995; Klein et al., 2017; Linsmayer et al., 2020; Rands et al., 1992). Alternatively, symbiont metabolism, including nighttime O₂ consumption, may compound negative effects of hypoxia, worsening larval outcomes (Baird et al., 2021; Hartmann et al., 2019). Clarifying the metabolic effects of hypoxia on larvae of cnidarian species with variation in life history traits would improve our ability to conserve these important taxa under ocean deoxygenation.

To gain insight into the metabolic effects of hypoxia on early life stages of cnidarian species displaying diversity in life history traits, we exposed larvae of two symbiotic reef-building corals, Galaxea fascicularis (gonochoric broadcast spawner; non-symbiotic larvae) and Porites astreoides (hermaphroditic brooder; symbiotic larvae), and the non-symbiotic estuarine sea anemone Nematostella vectensis (gonochoric broadcast spawner; non-symbiotic larvae) to a simulated severe hypoxic event (6 h at <2 mg DO l⁻¹; Glass and Barott, 2025). The abundances of 81 metabolites across four major classes (organic acids, amino acids, nucleotides and acylcarnitines) were quantified in groups of larvae exposed to normoxia (i.e. controls) and hypoxia (i.e. treatment) via targeted liquid chromatography-mass spectrometry (LC-MS) metabolomics. Further, to interrogate whether changes in cellular metabolism were linked to organismal outcomes, metabolite concentrations were tested for correlations with physiological responses (Glass and Barott, 2025). These data provide important insights into baseline and hypoxia-modified metabolism among diverse cnidarian larvae, with implications for the prediction of population trajectories under ocean deoxygenation.

Full methods for the experiment are available in Glass and Barott (2025). A laboratory line of adult Nematostella vectensis Stephenson 1935 was cultured at 18°C in 12 ppt artificial seawater, and spawning was induced using a standard protocol for this species (Hand and Uhlinger, 1992; Stefanik et al., 2013). Adult Galaxea fascicularis (Linnaeus 1767) were cultured in an ex situ spawning system at Carnegie Science (Baltimore, MD, USA), and spawning was induced by simulating in situ diel temperature and irradiance cycles (Craggs et al., 2017; Puntin et al., 2022). Adult Porites astreoides Lamarck 1816 were collected from a patch reef in Bermuda prior to predicted planulation and cultured in a mesocosm system with flow-through seawater at the Bermuda Institute of Ocean Science, where larvae were captured during overnight release using a standard protocol for this species and facility (Goodbody-Gringley et al., 2018). Stage-matched, swimming planulae from all three species (N=1200–2400 larvae per species) were divided into six replicate groups (three normoxia/control and three hypoxia) and exposed to 6 h of normoxia (DO=6.8–8.69 mg l⁻¹) or severe hypoxia (DO=1.58–1.8 mg l⁻¹; seawater deoxygenated using N₂ gas) inside sealed glass jars (500 ml) overnight from 21:00 h to 03:00 h the following day. The jars were placed at temperatures matching those experienced by adults during spawning (N. vectensis: 18°C; G. fascicularis: 27°C; P. astreoides: 28°C). At the end of the treatment period, the jars were uncapped and groups of 20–30 larvae (N=5 groups per treatment for a total of 100–150 total larvae per treatment per species) were transferred to 1.5 ml tubes without seawater for storage at −80°C until processing for targeted metabolomics as described below. DO concentrations remained constant throughout the treatment period (Glass and Barott, 2025). The deoxygenation method was also repeated for jars without animals using 12 and 36 ppt water (N=3 jars per salinity), and the pH was measured before and after deoxygenation using a handheld probe (Pro2Go; precision=0.001 units, accuracy=±0.002 units; Mettler Toledo, Columbus, OH, USA).

Groups of frozen larvae were thawed on ice and homogenized in 160 µl of 50:50 0.3% formic acid:acetonitrile in tough microorganism tubes (Revvity, Waltham, MA, USA) at 4°C in a Precellys homogenizer (Bertin Technologies, France). Aliquots of homogenates (20 µl) were extracted with organic solvents for individual targeted LC-MS metabolomics assays. Specifically, samples were processed at the University of Pennsylvania Metabolomics Core (RRID:SCR_022381), and quantification was performed for compounds belonging to four major classes central to cellular metabolism: acylcarnitines, amino acids, organic acids and nucleotides. These classes were chosen based on their importance to multiple metabolic pathways (e.g. glycolysis, citric acid cycle, fatty acid β-oxidation, anaplerotic reactions) that we hypothesized were affected by hypoxia based on physiological outcomes (Glass and Barott, 2025). A 10 µl aliquot of each homogenate was also used for protein concentration determination (to normalize metabolite concentrations) via the Bradford method using a bovine albumin serum standard curve. Quantitation of metabolites in each assay module was achieved using multiple reaction monitoring of calibration solutions and study samples on an Agilent 1290 Infinity UHPLC/6495 triple quadrupole mass spectrometer. Metabolites were separated on a Waters Acquity bridged ethylene hybrid (BEH) 2×100 mm, 1.7 µm column using a linear gradient from 95% A/5% B (A: 0.1% formic acid in water; B: acetonitrile with 0.1% formic acid) to 5% A/95% B at 0.4 ml min⁻¹ and 45°C over 10 min. Raw data were processed using Mass Hunter quantitative analysis software. Calibration curves (precision CVs≤15%, accuracies >90%, R²=0.9900 or greater linear fit) were either fitted with a linear or a quadratic curve with a 1/x or 1/x² weighting.

All data were analyzed in RStudio running R version 4.2.1 (https://www.r-project.org/). Principal components analysis (PCA) was performed using the ‘prcomp’ function, and the statistical significance (P<0.05) of species, treatment and their interaction as predictors of metabolite profiles was tested via a permutational analysis of variance (PERMANOVA) using the ‘adonis2’ function from the vegan package (Dixon, 2003). Differentially abundant metabolites between species, groups of species or treatments within each species were identified using pairwise tests with a Benjamini–Hochberg adjusted P-value cutoff ≤0.1 and fold change cutoff ≥1.5 (for species comparisons) or 2 (for treatment comparisons). To gain insight into the functions of metabolites differentially abundant between species or treatments, groups were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis using the web interface of MetaboAnalyst 6.0 (Ewald et al., 2024). Correlation analysis was performed to relate metabolite concentrations to physiological metrics (Table 1) using the psych package (https://CRAN.R-project.org/package=psych). Physiological metrics were quantified in larvae under normoxic conditions immediately following hypoxia exposure (alongside sampling for metabolomics), and metabolite concentrations for a given group of larvae were correlated with metrics measured in larvae from the same experimental treatment jar (N=5 groups per treatment per species).

The deoxygenation method resulted in pH increases of <0.01 units for both 12 and 36 ppt seawater (Fig. S1). Larval protein content (µg animal⁻¹) was not significantly affected by treatment (type III ANOVA, P=0.77; Fig. S2). Larval metabolite profiles were significantly influenced by species, DO treatment and their interaction (PERMANOVA, P<0.05), with clustering by both variables along PC1 (38% variance explained) and PC2 (22% variance explained) in a PCA analysis of all larval groups (Fig. 1A). Nucleotides and amino acids were generally responsible for separation along PC1 and PC2, respectively, with organic acids and acylcarnitines driving separation along both components (Fig. 1A). For each metabolite class, mean metabolite concentrations (nmol mg⁻¹ protein) by species and treatment displayed both conserved and divergent effects of hypoxia among species (Table S1; Fig. 1B). For example, organic acids were highest in N. vectensis, and increased in all three species following hypoxia exposure (Table S1; Fig. 1B). By contrast, amino acids were higher in the corals compared with N. vectensis, and significantly increased in the non-symbiotic larvae (N. vectensis and G. fascicularis) but not the symbiotic larvae of P. astreoides under hypoxia (Table S1; Fig. 1B).

There were 33 and 10 metabolites significantly more abundant in N. vectensis and G. fascicularis, respectively, in the comparison between these species (Fig. 1C). For example, N. vectensis showed significantly higher abundances of nucleotides including ATP, whereas G. fascicularis larvae displayed significantly higher abundances of amino acids including 1-methylhistidine (Fig. 1C). In the comparison of G. fascicularis and P. astreoides, 19 and 36 metabolites were significantly more abundant in the former and latter, respectively (Fig. 1D). Interestingly, G. fascicularis displayed higher abundances of multiple low-energy nucleotides including GMP and NAD⁺, whereas P. astreoides displayed higher abundances of multiple hydroxylated acylcarnitines including hydroxyhexanoyl (C6-OH) and hydroxyoctanoyl (C8-OH) carnitines (Fig. 1D). The species comparison with the greatest number of differentially abundant metabolites was N. vectensis versus P. astreoides, with 31 and 26 metabolites significantly more abundant in the former compared with the latter and vice versa (Fig. 1E). Nematostella vectensis displayed significantly higher abundances of ATP in this comparison (also higher in N. vectensis compared with G. fascicularis). By contrast, P. astreoides displayed higher abundances of hydroxylated acylcarnitines (also higher in P. astreoides compared with G. fascicularis; Fig. 1C–E).

A family-level differential abundance analysis (i.e. N. vectensis versus both corals) revealed 37 metabolites that displayed significantly higher abundance in N. vectensis compared with the corals, whereas 19 metabolites were significantly more abundant in the reciprocal comparison (Fig. 2A). These metabolites included cAMP and ATP with significantly higher abundance in N. vectensis, whereas glutamate and 3-methylhistidine were significantly more abundant in the corals (Fig. 2A). KEGG analyses demonstrated the significant enrichment of the terms ‘purine/pyrimidine metabolism’ (adjusted P=0.032) and ‘pentose phosphate pathway’ (adjusted P=0.103) for metabolites more abundant in N. vectensis, whereas those significantly more abundant in the corals contributed to the significant enrichment of numerous terms including ‘ammonia recycling’ (adjusted P<0.001) and ‘alanine, aspartate, and glutamate metabolism’ (adjusted P=0.005; Fig. 2B).

A differential abundance analysis comparing the non-symbiotic species (i.e. N. vectensis and G. fascicularis) with the symbiotic P. astreoides revealed 21 metabolites significantly more abundant for both sides of this comparison (Fig. 2C). Among these metabolites was NADP⁺, which was significantly more abundant in the non-symbiotic species, whereas several hydroxylated acylcarnitines (e.g. C6-, C10- and C14:1-OH) and amino acids (e.g. glutamate and glutamine) were more abundant in P. astreoides (Fig. 2C). These metabolites contributed to the significant enrichment of numerous KEGG terms, including ‘glycerolipid metabolism’ (adjusted P=0.003) and ‘mitochondrial β-oxidation of short-, medium-, and long-chain (saturated) fatty acids’ (adjusted P=0.045, 0.013 and 0.045, respectively) in the non-symbiotic species and ‘nitrogen metabolism’ (adjusted P=0.047) in P. astreoides (Fig. 2D).

Differential abundance analyses of metabolite abundances under normoxia versus hypoxia for each species revealed numerous shared and species-specific effects (Fig. 3). Lactate was the only metabolite that was significantly more abundant in all three species following hypoxia exposure (Fig. 3D). There were seven additional metabolites significantly more abundant under hypoxia in at least two of the three species: pyruvate, succinate, NAD, 3-hydroxybutyric acid, glutamate and hydroxyhexadecanoyl (C16-OH) carnitine, all of which were significantly more abundant under hypoxia in both of the non-symbiotic species; and valine, which was more abundant under hypoxia in both G. fascicularis and P. astreoides (Fig. 3A–D). These eight metabolites contributed to the significant enrichment of KEGG terms including ‘ketone body metabolism’, ‘gluconeogenesis’, ‘citric acid cycle’ and ‘glycolysis’ (adjusted P=0.001, 0.004, 0.01 and 0.047, respectively; Fig. 3E). Species-specific effects observed in N. vectensis included a significant decrease in the abundance of multiple triphosphate nucleotides, including ATP, CTP, TTP and GTP, whereas metabolites including NAD⁺, malate and pyruvate were significantly more abundant under hypoxia (Fig. 3A). Further, both G. fascicularis and P. astreoides displayed significantly lower abundances of long-chain acylcarnitines (e.g. C14 and C14:1 in the former and C10 in the latter) under hypoxia, but higher abundances of short-chain acylcarnitines [e.g. C4(:1)] and branched-chain amino acids (BCAAs; e.g. valine) under hypoxia, whereas these changes were not observed in N. vectensis (Fig. 3B,C).

Correlations between metabolite concentrations and physiological metrics performed for each species revealed numerous significant positive and negative correlations (P<0.05; Figs 4 and 5). For example, in N. vectensis, several metabolites, including multiple triphosphate nucleotides (e.g. ATP and GTP), displayed significant positive correlations with heat tolerance, aerobic respiration rates, swimming behavior and settlement rates (Fig. 4A). By contrast, other metabolites in N. vectensis, including several amino acids (e.g. valine and tryptophan) and organic acids (e.g. pyruvate and lactate), displayed significant positive correlations with larval protein content and/or organic biomass (ash-free dry mass), but negative correlations with most other metrics including heat tolerance, swimming behavior and settlement rates (Fig. 4A). Notably, a similar pattern was observed in G. fascicularis, with several metabolites showing positive correlations with metrics including heat tolerance and aerobic respiration rates, and another group showing positive correlations with size (Fig. 4B). Specifically, malonyl (C3-DC) and tetradecenoyl (C14:1) carnitines were positively correlated with heat tolerance in G. fascicularis (Fig. 4B). By contrast, metabolites including hydroxyvaleryl (C5-OH) and hydroxyisobutyryl (C4-OH-ISO) carnitines and GMP showed strong negative correlations with most metrics but positive correlations with size in this species (Fig. 4B). Interestingly, a set of largely different compounds compared with the previous two species were correlated with physiological metrics in P. astreoides (Fig. 4C). For example, several medium-chain acylcarnitines [e.g. octanoyl (C8) and decanoyl (C10) carnitines] were positively correlated with symbiont photosynthetic efficiency (F_v/F_m), whereas a diverse group of metabolites that included many (branched-chain) amino acids (e.g. phenylalanine, leucine, valine, isoleucine) were negatively correlated with metrics including settlement and aerobic respiration rates, symbiont densities and photosynthetic rates (Fig. 4C). Interestingly, one of the strongest positive correlations observed in P. astreoides was between ATP and the maintenance of symbiont photosynthetic efficiency under heat stress (Fig. 4C). Another notable metabolite in P. astreoides was histidine, which was the only metabolite in this species with significant positive correlations with multiple metrics including metabolic rates (i.e. aerobic respiration and photosynthesis; Fig. 4C).

Marine hypoxic events are increasing in frequency and severity as a result of anthropogenic activities (Breitburg et al., 2018; Pezner et al., 2023; Rose et al., 2024), making it essential to understand how the sensitive early life stages of cnidarians are affected by low seawater DO levels. Here, we quantified compounds foundational for cellular metabolism in larvae exposed to normoxic (i.e. control) and hypoxic (i.e. treatment) conditions. Our results demonstrate that hypoxia disrupts larval metabolism through conserved upregulation of glycolysis, fermentation and fatty acid β-oxidation, alongside species-specific effects on metabolic pathways that corresponded with distinct life history traits (e.g. symbiotic state), which appear to have a profound influence on metabolic hypoxia responses. These results build our mechanistic understanding of cnidarian hypoxia responses to better explain the pervasive negative effects of hypoxia on cnidarian larval physiology (Glass and Barott, 2025).

Larvae of N. vectensis, G. fascicularis and P. astreoides displayed evidence of increases in glycolysis and lactic acid fermentation under hypoxia, suggesting that all three species were able to supplement energy demands with anaerobic metabolism when aerobic pathways were limited. Lactic acid fermentation occurs when lactate is produced via the reduction of pyruvate using NADH, which can provide NAD⁺ for the continued production of ATP through glycolysis when O₂ limitation inhibits the mitochondrial respiratory chain (Kierans and Taylor, 2021). An increase in coupled glycolysis and fermentation is a known metabolic strategy for sustained ATP production by vertebrate cells experiencing hypoxia (Kierans and Taylor, 2021), and our data suggest that this also occurs in coral and sea anemone larvae. Interestingly, fermentation under low O₂ has also been observed in adults of the corals Acropora yongei and Montipora capitata (Linsmayer et al., 2020; Murphy and Richmond, 2016). However, in contrast to our observations in larvae, adult A. yongei and M. capitata do not produce appreciable quantities of lactate under hypoxia, instead producing opine compounds (not quantified here) as the main fermentative end products (Linsmayer et al., 2020; Murphy and Richmond, 2016). These differences in fermentation pathways may reflect species-specific differences and/or a life-stage specific differences between adults and larvae. In support of the latter hypothesis, an increase in the expression of the lactic acid fermentation gene lactate dehydrogenase B was observed in larvae of the coral A. selago under hypoxia (Alderdice et al., 2022), whereas adult N. vectensis exposed to hypoxia displayed no evidence of changes in the expression of genes involved in this pathway (Glass et al., 2025). Together, these findings suggest that the upregulation of glycolysis and lactic acid fermentation to support ATP production is a conserved metabolic hypoxia response among diverse cnidarian larvae. Importantly, although this metabolic adjustment likely promoted larval survival in the short term, glycolysis yields only a fraction of the ATP produced via aerobic mitochondrial respiration, and this may in part explain why larvae still displayed numerous negative physiological outcomes following hypoxia exposure. For example, larval swimming behavior of all three species was significantly impaired immediately following hypoxia exposure (Glass and Barott, 2025). This and other outcomes resulting from a decrease in cellular ATP levels under hypoxia could have severe implications for cnidarian life cycle completion in a deoxygenating ocean. However, the ability to switch to fermentation to generate ATP indicates some resilience against transient exposures to hypoxia.

Metabolomic evidence suggested that hypoxia led to increased fatty acid β-oxidation in all three species, likely indicating increased consumption of maternally provided fatty acid stores. However, this was less pronounced in P. astreoides larvae, possibly reflecting metabolic supplementation from algal endosymbionts. The end product of fatty acid β-oxidation is acetyl-CoA, which can be used in the citric acid (TCA) cycle to produce reducing equivalents (e.g. NADH) that ultimately power the mitochondrial electron transport chain (ETC; van Vlies et al., 2005; Vissing et al., 2019). The metabolic benefit of fatty acid catabolism under hypoxia may have been minor given that the ETC should have been O₂-limited. However, the shared increase in succinate also indicated that larvae may have initiated AMP-dependent activation of the purine nucleotide cycle to produce fumarate, which can act as an alternative electron acceptor in the ETC when O₂ is low and leads to succinate accumulation (Chouchani et al., 2014; Flores et al., 2018; Sridharan et al., 2008). Nonetheless, it is unlikely that this alternative route to the ETC would by itself support the regeneration of sufficient concentrations of oxidized electron carriers (namely NAD⁺ and FAD) to promote the continuation of β-oxidation. Although lactic acid fermentation was increased and can provide NAD⁺, there would have also been augmented demand for this molecule to support glycolysis (see above), and this cannot explain the regeneration of FAD. Therefore, the metabolic source of these oxidized electron carriers remains unclear and should be investigated further. Additionally, even if fatty acid catabolism supported short-term performance under hypoxic conditions, the depletion of fatty acid stores under extended and/or repeated hypoxia may worsen long-term outcomes (e.g. metamorphosis, settlement and survival).

The corals, but not N. vectensis, displayed modified metabolism of nitrogen-containing compounds including numerous amino acids under hypoxia, which likely contributed to the disruption of the coral–dinoflagellate symbiosis observed in both species. For example, both coral species displayed significant increases in BCAAs, including valine and isoleucine, and their catabolic intermediates, including isobutyryl (C4-ISO), succinyl (C4-DC) and isovaleryl (C5-ISO) carnitines, under hypoxia (Neinast et al., 2019). This may serve multiple purposes under hypoxic conditions. First, converting excess pyruvate to BCAAs can permit the continuation of glycolysis, which was increased under hypoxia (see above). Alternatively, or in addition, increases in BCAAs and their catabolic intermediates may reflect increased protein breakdown, which is likely to occur under hypoxia given that one cellular effect of hypoxia is the accumulation of misfolded proteins that must be broken down to avoid toxicity (Koumenis et al., 2007; Tu and Weissman, 2002). However, it is interesting that this response was not observed here in N. vectensis larvae, as hypoxia exposure leads to activation of the unfolded protein response in adults of this species (Glass et al., 2025). However, this difference may be due to distinct exposure regimes and merits further investigation. Notably, BCAA and protein catabolism result in the production of ammonium, which can cause coral symbionts to retain more photosynthate, thereby disrupting symbiosis (Cui et al., 2023; Rädecker et al., 2023). This may explain why hypoxia exposure induced symbiont loss (i.e. bleaching) in P. astreoides larvae and diminished symbiosis establishment in G. fascicularis juveniles (Glass and Barott, 2025). These results build upon recent advances demonstrating a key role of coupled carbon and nitrogen metabolism, largely through amino acids, in the regulation of symbiosis in coral adults (Cui et al., 2023; Rädecker et al., 2023). However, it was notable that metabolites central for nitrogen cycling were present at high abundances even in the symbiont-free larvae of G. fascicularis, as this suggests the importance of nitrogen metabolism for coral larvae beyond the regulation of symbiosis. One possibility is that the oligotrophic seawater conditions on coral reefs make it important to recycle nitrogen rather than lose nitrogenous compounds to secretion. These hypotheses should be explored further to better understand how ocean deoxygenation influences coral–dinoflagellate symbiosis and organismal outcomes across life stages.

In N. vectensis larvae, energy-rich triphosphate nucleotides (NTPs) were high at baseline and rapidly consumed under hypoxia, likely supporting this species' resilience. This suggests that although the mitochondrial respiratory chain was likely limited by the lack of O₂ to act as a terminal electron acceptor, a cellular ‘stockpile’ of NTPs may have promoted short-term metabolic acclimation. Particularly important for N. vectensis was GTP, as this compound was almost entirely consumed under hypoxia and displayed significant correlations with every physiological metric quantified. For example, there was a strong positive correlation between GTP and aerobic respiration rates, indicating that the consumption of GTP under hypoxia may in part explain why aerobic respiration rates were not significantly affected by hypoxia in this species (Glass and Barott, 2025). A major role of GTP is to provide energy for the de novo formation of glucose via gluconeogenesis, and GTP can also be converted directly to ATP (Gupta and Gupta, 2021). Production of glucose and/or ATP via these pathways may have supported the overall resilience of N. vectensis under hypoxia. However, it is notable that these NTPs were nearly undetectable after only 6 h at <2 mg DO l⁻¹, suggesting that even N. vectensis may lack resilience to extended and/or recurring hypoxic events, which are becoming more common in its native estuaries under global change (Breitburg et al., 2018; Glass et al., 2025; Schmidt et al., 2019; Sharp, 2010). Indeed, adults of this species exposed transiently to severe hypoxia displayed reduced metabolic and reproductive performance (Glass et al., 2025). A more detailed investigation of the role of nucleotides in N. vectensis metabolism and hypoxia responses across life stages is thus warranted.

Symbiotic P. astreoides larvae displayed metabolic outcomes not observed in symbiont-free N. vectensis and G. fascicularis larvae, despite sharing a close evolutionary relationship and most ecological traits (e.g. tropical habitat) with G. fascicularis. Most notably, P. astreoides displayed the lowest number of metabolites with significantly differential abundance under hypoxia among the three species, possibly reflecting a diminished holobiont response and/or conflicting outcomes between host and symbiont. Additionally, there were a number of metabolites present in P. astreoides that were undetectable in the other two species, including multiple hydroxylated acylcarnitines (e.g. C6-OH, C8-OH, C10-OH, etc.). These compounds are markers of mitochondrial dysfunction (Vissing et al., 2019), and have been detected in coral holobionts (Deutsch et al., 2024b), yet their origin and role remain unclear. As metabolites were quantified simultaneously for both P. astreoides larvae and their endosymbionts, these outcomes represent the holobiont (i.e. host and symbionts together) response to hypoxia. Interestingly, multiple of these metabolites were positively correlated with symbiont photosynthetic efficiency, possibly indicating that they originate from and support the performance of symbionts, yet further research is needed to clarify this. These results suggest that symbiotic state may be an important factor in determining the metabolic effects of hypoxia, making it essential to consider life history traits such as symbiotic relationships in future research and cnidarian conservation.

Exposure to hypoxia had numerous effects on the metabolism of coral and sea anemone larvae that were correlated with both declines in and the maintenance of physiological performance. Further research is needed to more fully develop our understanding of the effects of hypoxia on cnidarian early life stages. For example, glycolysis and lactic acid fermentation consistently increased under hypoxia, yet it remains unknown how these pathways may change when hypoxia co-occurs with other environmental stressors. For example, heat stress results in decreases in these same pathways in larvae of the coral Montipora capitata (Huffmyer et al., 2024), and may thus lead to synergistic effects in combination with hypoxia. Furthermore, hypoxic events are often accompanied by decreases in seawater pH (Deutsch et al., 2024a; Pezner et al., 2023; Rose et al., 2024), making it essential to further investigate how multiple stressors may interact to affect the metabolism of cnidarian larvae. In particular, the quantification of metabolites belonging to other classes not explored here (e.g. carbohydrates and lipids) in larvae following exposure to hypoxia with and without additional stressors (e.g. heat, seawater acidification) would offer additional insights into how these stressors influence larval metabolism. We also suggest further research surrounding the role of BCAAs in cnidarian metabolism and stress responses, as these compounds are largely unexplored in these taxa yet display metabolic importance under multiple stressors (e.g. present study; Huffmyer et al., 2024). Relatedly, the finding that the three species investigated here displayed some convergent metabolic hypoxia responses suggests that other cnidarian taxa may also display these responses. Indeed, an increase in lactate following hypoxia exposure has also been observed in other cnidarians, including the jellyfish Aurelia sp. (Wang et al., 2017). Further application of targeted metabolomics to other cnidarian species exposed to hypoxia will better inform our understanding of cnidarian physiology and the evolution of metazoan metabolic hypoxia responses. Overall, limiting drivers of ocean deoxygenation will likely be central to the persistence of cnidarian populations.

The authors thank Kristen Brown for valuable feedback on data analysis and presentation, and two anonymous reviewers for constructive comments. Metabolomics was performed by the University of Pennsylvania Metabolomics Core (RRID:SCR_022381) supported by the Penn Cardiovascular Institute and National Institutes of Health awards P30CA016520 and P30DK050306.