The short-term combined effects of temperature and organic matter enrichment on permeable coral reef carbonate sediment metabolism and dissolution

Rates of gross primary production (GPP), respiration (R), and net calcification (Gnet) in coral reef sediments are expected to change in response to global warming (and the consequent increase in sea surface temperature) and coastal eutrophication (and the subsequent increase in the concentration of organic matter, OM, being filtered by permeable coral reef carbonate sediments). To date, no studies have examined the combined effect of seawater warming and OM enrichment on coral reef carbonate sediment metabolism and dissolution. This study used 22 h in situ benthic chamber incubations to examine the combined effect of temperature (T ) and OM, in the form of coral mucus and phytodetritus, on GPP, R, and Gnet in the permeable coral reef carbonate sediments of Heron Island lagoon, Australia. Compared to control incubations, both warming (+2.4 C) and OM increased R and GPP. Under warmed conditions, R (Q10 = 10.7) was enhanced to a greater extent than GPP (Q10 = 7.3), resulting in a shift to net heterotrophy and net dissolution. Under both phytodetritus and coral mucus treatments, GPP was enhanced to a greater extent than R, resulting in a net increase in GPP / R and Gnet. The combined effect of warming and OM enhanced R and GPP, but the net effect on GPP / R and Gnet was not significantly different from control incubations. These findings show that a shift to net heterotrophy and dissolution due to short-term increases in seawater warming may be countered by a net increase GPP / R and Gnet due to short-term increases in nutrient release from OM.


Introduction
Despite occupying only 7.5 % of the seafloor, coastal marine sediments are responsible for a large fraction (55 %) of global sediment organic matter oxidation (Middelburg et al., 1997).Of the coastal marine sediment environments, coral reef sediments are one of the most severely threatened by global climate change (Halpern et al., 2007).Rates of sediment autotrophic production (gross primary productivity; GPP) on coral reefs are generally greater than rates of heterotrophic metabolism (respiration; R; GPP / R > 1), such that the sediments are a net source of oxygen (Atkinson, 2011).Similarly, rates of sediment calcification/precipitation are generally greater than rates of sediment dissolution (G net > 0) on most reefs under current ocean conditions, such that coral reef sediments on diel (24 h) timescales are net precipitating, resulting in the long-term burial of carbon in the form of calcium carbonate (Eyre et al., 2014;Andersson, 2015).This long-term production of calcium carbonate is an important component of reef formation and the creation of sandy cays (Atkinson, 2011).However, due to anthropogenically mediated processes such as sea surface temperature (SST) warming (Levitus et al., 2000) and coastal eutrophication (Fabricius, 2005), coral reef sediments may soon be subjected to elevated SSTs and excess loadings of OM (Rabouille et al., 2001).This could ultimately impact the balance in GPP / R and G net in the sediment and potentially alter the long-term accumulation of carbonate material on coral reefs (Orlando and Yee, 2016).
Given the recent projections of SST increases on coral reefs of between 1.2 and 3.2 • C by the end of this century Published by Copernicus Publications on behalf of the European Geosciences Union.
(IPCC, 2013), there are concerns that the net metabolic balance in coral reef sediments may shift away from net production and net calcification to a state of net heterotrophy and net dissolution (Pandolfi et al., 2011).While several coral reef studies have examined the response in individual calcifying organisms to increased seawater temperature (T ; e.g.Johnson and Carpenter, 2012;Shaw et al., 2016), only one study (Trnovsky et al., 2016) has examined the response in entire permeable coral reef carbonate sediments.Furthermore, the majority of warming studies on marine sediments have been performed ex situ in more poleward latitudes (temperate to arctic environments) over a wide range of temperatures (2-30 • C; e.g.Tait and Schiel, 2013;Hancke et al., 2014;Ashton et al., 2017).The bacterial communities residing in marine sediments generally display a hyperbolic temperature-production relationship where GPP increases with T (∼ +32 % per 1 • C increase) until an optimal rate is reached roughly +2-3 • C above naturally observed seasonal maxima.This T -GPP relationship then declines at higher temperatures (+4-6 • C) due to the deactivation of component reactions (Bernacchi et al., 2001).In Arctic and temperate marine sediment communities, the increase in T can alter the balance between GPP and R, with an observed shift towards net heterotrophy (GPP / R < 1; e.g.Hancke and Glud, 2004;Weston and Joye, 2005).Trnovsky et al. (2016) found that warming also decreased GPP / R in coral reef sediments and reduced G net due to enhanced sediment dissolution.
Ultimately, the magnitude of potential shifts in coral reef sediment GPP / R and G net under global warming scenarios will depend critically on the availability of organic matter (OM) substrate for remineralisation (Ferguson et al., 2003;Rabalais et al., 2009).Carbonate sediment dissolution is strongly controlled by the extent of OM decomposition in the sediments (Andersson, 2015).Coral reefs are classically characterised as oligotrophic, i.e. relatively deficient in major inorganic nutrients (Koop et al., 2001).Despite this classification, the relatively high rates of GPP (1 to 3 mol C m −2 d −1 ) for these ecosystems (Odum and Odum, 1955) are evidence of a tightly coupled nutrient cycling between autotrophs and heterotrophs.However, the balance in sediment metabolism on coral reefs may change in response to OM over-enrichment associated with eutrophication (Bell, 1992).Coral reefs affected by eutrophication (e.g.Hawaii: Grigg, 1995;Indonesia: Edinger et al., 1998;Jamaica: Mallela and Perry, 2007;Puerto Rico: Diaz-Ortega and Hernandez-Delgado, 2014) all exhibit elevated concentrations of OM in the water column (particulate OM: 10-50 µmol C L −1 ) and above-average rates of sedimentation (5-30 mg cm −2 d −1 ).Elevated concentrations of OM and increased rates of terrestrially derived sedimentation on coral reefs can cause a decline in hard coral cover and a relative increase in macroalgal cover, resulting in an overall degradation of coral reef habitat (Fabricius, 2005).
The amount of OM processed in coral reef sediments can be increased through several processes, two of which were simulated in this study: (1) through local phytoplankton blooms in the water column in response to the runoff of inorganic and organic nutrients and the eventual sediment deposition of dead phytoplankton, referred to herein as phytodetritus (Furnas et al., 2005), and (2) the release of coral mucus into the reef water column as a stress response of scleractinian corals to increased sedimentation and the subsequent sediment deposition of this bacteria-rich protein matrix (Ducklow and Mitchell, 1979).The sediment deposition of OM provides labile carbon substrate (and associated nitrogen and phosphorous) for immediate consumption by autotrophic and heterotrophic bacterial communities.
Studies which have examined the effect of increased concentrations of OM, such as coral mucus (e.g.Wild et al., 2004a;24 h) or coral spawn and phytodetritus (e.g.Eyre et al., 2008; 1 week), on coral reef sediment metabolism have shown a short-term increase in GPP / R, contrasting the results provided from short-term temperature studies on coral reef sediments, where GPP / R decreased (Trnovsky et al., 2016;24 h).Experimental additions of coral mucus from Acropora spp. on Heron Island, Australia (conducted only in the dark over 12 h), induced a ∼ 1.5-fold increase in R (Wild et al., 2004b) while additions of Fungia spp.mucus from a reef in Aqaba, Jordan (also conducted over 12 h in the dark; Wild et al., 2005), showed a ∼ 1.9-fold increase in R. OM associated with a mass coral spawning event (coral gametes and subsequent phytodetritus produced in the water column) on Heron Island, Australia, caused a 2.5-fold increase in sediment R and a 4-fold increase in sediment GPP over the course of 1 week (Glud et al., 2008).Unlike the short-term response in GPP / R to T , sediment metabolism remained net-autotrophic during the spawning event at Heron Island, with GPP / R ratios rising as high as 2.5-3.0 (Glud et al., 2008), implying that nutrients recycled from OM stimulated GPP in excess of R (Eyre et al., 2008) on relatively short timescales (hours to days).However, studies which have examined the effect of excess OM on coral reef sediment metabolism over longer timescales (months) have shown that, ultimately, GPP / R eventually shifts to net heterotrophy (e.g.Andersson, 2015;Yeakel et al., 2015;Muehllehner et al., 2016).This suggests that despite an initial OM-induced increase in GPP / R, the net long-term effect within reef sediments may be a preferentially heterotrophic recycling of nutrients released from organic matter degradation.Altogether, questions remain as to whether a predicted temperature-driven shift to net heterotrophy will be exacerbated or mitigated by the presence of excess organic matter filtered by coral reef sediments.There are, to date, no studies that have examined the effect of OM on coral reef sediment G net .The observed short-term (24 h to 1 week) increase in GPP / R in response to OM would imply that sediment G net may also increase given that coral reef sediments generally exhibit a positive GPP / R-G net relationship (Cyronak and Eyre, 2016), whereas the observed long-term (months) decrease in GPP / R may also reduce sediment G net .
Therefore, seawater warming and OM enrichment will likely increase GPP and R in coral reef sediments, but, altogether, there is a lack of research on how these perturbations, specifically in combination, will affect the balance in coral reef sediment organic (GPP / R) and inorganic (G net ) metabolism.To meet these needs, this study performed incubations using benthic chambers placed in situ in a shallow coral reef sediment environment for a period of 24 h.Phytodetritus and coral mucus were added to chamber seawater under ambient and increased SST (+2.4 • C) conditions and the corresponding changes in GPP, R, and G net were measured.We hypothesised that the short-term combined treatments of seawater warming and OM loading would enhance GPP and R in the sediment, but, given the previously shown short-term response in GPP / R and G net to seawater warming (decrease in GPP / R and G net ) and net response to OM enrichment (decrease in GPP / R, G net response unknown), there would be a net decrease in GPP / R and G net relative to control treatments.

Study site
This study was conducted at Heron Island, Australia (23 • 27 S, 151 • 55 E), in November 2016.The island is situated near the Tropic of Capricorn, at the southern end of the Great Barrier Reef (GBR) and contains a ∼ 9 ha island surrounded by a ∼ 24 ha coral reef with an average hard coral cover of ∼ 39 % (Salmond et al., 2015).The study site was located on the leeward side of the reef flat, roughly 100 m from the island shore, in a sandy patch where water depth varies between ∼ 0.1 and 2.7 m due to semi-diurnal tidal changes.The site was predominately covered in permeable CaCO 3 sediments (∼ 63 %) with interspersed patches of hard coral dominated by Acropora spp.(Roelfsema and Roelfsema, 2002).The CaCO 3 sediment at this site has a ∼ 2 : 1 ratio of aragonite : high-magnesium calcite (Cyronak et al., 2013a).Sediment grain size at this site showed the following relative abundances at each listed size class (Cyronak et al., 2013b): 12.1 % > 2 mm, 30.5 % between 1 and 2 mm, 27.3 % between 500 µm and 1 mm, 14.1 % between 250 and 500 µm, 11.2 % between 125 and 250 µm, 4.2 % between 63 and 125 µm, and 0.6 % < 63 µm.For a more detailed overview of the sediment grain characteristics at this site, we direct the reader to Glud et al. (2008) and Cyronak et al. (2013a, b).

Experimental design
A total of four 22 h diel incubations were conducted during 5-12 November 2016 in advective benthic chambers.Benthic net primary production (NPP), gross primary productivity (GPP), respiration (R), and net calcification (G net ) were compared under ambient (∼ 0.63 µmol C L −1 ) and elevated concentrations of particulate organic matter (OM; additions coral mucus) at ∼ 28.2 and ∼ 30.6 • C in an orthogonal design.Eight chambers were used per incubation day, with each of the four OM-temperature combinations replicated in two randomly assigned chambers (Fig. 1).The first two incubations included two replicate chambers using phytodetritus crossed with temperature (6 and 7 November 2016), while the next two incubations included two replicate chambers using coral mucus crossed with temperature (9 and 11 November 2016).Incubations were started at sunset (18:00) and ended the following day at dusk (16:00).This allowed for a 2 h period (16:00-18:00) where chambers could be moved to a new area of sediment, closed, and heated to the desired temperature offset before beginning the next set of incubations.

Benthic chambers
Advective benthic chambers were constructed out of clear acrylic with a height of 33 cm and a diameter of 19 cm (Huettel and Gust, 1992).A motorised clear disc in the top of the chamber was programmed to spin at a rate of 40 revolutions min −1 , which had previously been determined to induce an advection rate of ∼ 43 L m −2 d −1 at the study site (Glud et al., 2008).About 10-12 cm of the base of the cham- ber was inserted into the sediment such that a ∼ 4 L water column of seawater was enclosed within the chamber (height ∼ 15 cm) upon closing of the lid.The exact water volume varied within each chamber and was calculated for each incubation by multiplying known areal coverage by measured chamber height (at three positions above the sediment).Prior to closing the chambers, the tops were left open for ∼ 1 h to allow settlement of disturbed sediment.Chambers were then sealed ∼ 1 h prior to the beginning of each incubation to allow each temperature treatment chamber to reach the desired temperature offset.Following this, at the beginning of each incubation, selected chambers (four of the eight) were injected with OM (either coral mucus or phytodetritus).

Temperature manipulation
The international panel on climate change (IPCC) representative concentration pathway (RCP) 8.5 projects an average 2.2-2.7 • C increase in SST (IPCC, 2013).A similar increase in temperature within the benthic chambers was achieved with 5 W silicone heating pads (RS Australia) inserted inside of each of the four temperature treatment chambers (e.g.Trnovsky et al., 2016).These pads resided in the middle of the chamber water column and were powered by a 12 V battery on a surface support station tethered roughly 3 m away.
Temperature and light was measured in all eight chambers and in the water column using HOBO temperature loggers, which recorded temperature ( • C) and light (lux) at an interval of 15 min.Light intensity (lux) was converted to µmol quanta of photosynthetic active radiation (PAR) m −2 s −1 using a conversion factor of 0.0185, derived from correlations with PAR measurements of a calibrated ECO-PAR (Wetlabs) sensor over a period of 5 days (R 2 = 0.89).Heating pads increased temperature (T ) within the chambers by 2.4 ± 0.5 • C and maintained this offset on top of the natural diel temperature fluctuations measured in the control chambers (Table 1).As HOBO temperature loggers may record potentially seawater temperatures higher than the surroundings due to internal heating of the transparent plastic casing (Bahr et al., 2016;Trnovsky et al., 2016), HOBO temperature data was corrected for precision (48 h side-by-side logging of all nine loggers in an aquarium) and accuracy (deployment next to an in situ SeapHOx (Sea-Bird Electronics) for 48 h).The conductivity sensor of the SeapHOx was used to record water column salinity for the duration of the experiment (7 days) at a sampling frequency of 30 min.

Organic matter manipulations
Phytodetritus (PD) was injected into treatment chambers to achieve a concentration increase by ∼ 20 µmol C L −1 , a value analogous to mean conditions observed on degraded eutrophic coral reefs, where water column concentrations can range from 10 to 50 µmol C L −1 (Fabricius, 2005;Diaz-Ortega and Hernandez-Delgado, 2014).Phytodetritus was produced from unfiltered seawater (6 L) collected from the coastal ocean adjacent to the SCU laboratories (Lennox Head, NSW, Australia) and containing naturally occurring assemblages of phytoplankton species common to the East Australian current.Phytoplankton growth in the collected seawater was stimulated by additions of 128 µmol L −1 NO − 3 , 8 µmol L −1 PO 3− 4 , and 128 µmol L −1 H 4 SiO 4 (buffered by additions of 256 µmol L −1 of HCl), and a solution of trace metals and vitamins (F 1/8 ; Guillard, 1975).Total amounts of nutrients were chosen to allow for a community production of up to 850 µmol C L −1 assuming a classical C : N : P Redfield ratio of 116 : 16 : 1 and a N : Si requirement of diatoms of 1.After 1 week of incubation at 150 µmol quanta of PAR m −2 s −1 at 20 • C, the phytoplankton community was concentrated to 1/50th the original volume (0.12 L) via gentle (> −0.2 bar) vacuum filtration over GF/F filters and rinsed with artificial seawater to remove residual concentrations of dissolved organic and inorganic nutrients.The resulting phytoplankton concentrate (measured at 8.5 mmol C mL −1 and 0.9 mmol N mL −1 of particulate organic carbon (POC) and nitrogen (PON) respectively per 1 mL of PD concentrate; see Sect.2.6 for details) was stored in the dark at 4.0 • C until experimental use (6 days).At the beginning of an incubation, 10 mL of the dead phytoplankton concentrate, referred to as PD hereafter, was injected into each treatment chamber (∼ 4 L volume), raising the concentration of carbon and nitrogen by ∼ 21.3 ± 1.0 µmol C L −1 and ∼ 2.2 ± 0.8 µmol N L −1 respectively (Table 1).
The amount of coral mucus (CM) added to the chambers was chosen to represent a reef-wide discharge based on reported average mucus secretion rates for Acropora spp.(4.8 L mucus m −2 d −1 ; Wild et al., 2004a), the dominant genus on the Heron Island reef flat.Mucus was collected from scattered branching coral fragments (Acropora spp.) using a non-destructive method whereby loose individual colonies naturally exposed to air during low tide were inverted so that gravity facilitated the pooling of secreted mucus through a cone filter into a 5 L beaker.This mucus was returned to the lab, particle filtered (5.0 µM) to remove the bulk of seawater, re-filtered to separate out particle carbonates, and stored in the dark at 4.0 • C until experimental use (2 days).vNinety-four millilitres of mucus was injected into each treatment chamber to simulate the equivalent reported Acropora spp.mucus secretion rate (4.8 L mucus m −2 d −1 ) for Heron Island given the average percent of this secreted mucus filtered by the sand (∼ 70 %; Wild et al., 2004a) and the benthic area enclosed by each chamber (0.028 m 2 ).Based on measured POC and PON concentrations of the mucus (1.2.mmol C mL −1 and 0.08 mmol N mL −1 respectively per 12 mL of CM concentrate; see Sect.2.6) this represented an addition of ∼ 23.6 ± 1.1 µmol C L −1 and 1.4 ± 0.4 µmol N L −1 (Table 1).

Sample collection and analysis
Seawater samples (120 mL total) were extracted from the top of each chamber via two two-port valves using two 60 mL syringes without headspace at ∼ 12 h intervals (sunset, dawn, and dusk) and returned to the lab for immediate analysis and/or preservation.10 mL of unfiltered seawater from each chamber was analysed for dissolved oxygen (DO; mg L −1 ) with a Hach HQ30D meter and Luminescent DO (LDO) probe.Samples for seawater total alkalinity (A T ; µmol kg −1 ) were filtered (0.45 µm; Chanson and Millero, 2007) and stored in 100 mL plastic, airtight bottles for immediate analysis (< 24 h).Samples for dissolved inorganic carbon (DIC; µmol kg −1 ) were also filtered (0.45 µM) into the bottom of 6 mL vials with 5 mL overflow, poisoned (6 µL of saturated HgCl 2 ; Dickson et al., 2007) and crimped (rubber butyl septum).
Seawater A T was analysed using a potentiometric titration method (Dickson et al., 2007) on a Metrohm 888 Titrando automatic titrator using ∼ 10 mL of weighed-in seawater per sample.DIC was analysed in triplicates on a Marianda AIR-ICA coupled to a LI-COR LI 7000 CO 2 /H 2 O analyser on 0.4 mL of seawater per sample.A T and DIC sample precision was estimated with replicate analyses conducted on every fifth sample (A T SD = ±1.7 µmol kg −1 ; DIC SD = ±1.8µmol kg −1 ).Measurements were corrected against certified reference material (CRM; Batch 155) from the Scripps Institute of Oceanography (A T SD = ±2.2µmol kg −1 ; DIC SD = ± 1.3 µmol kg −1 ).Parameters for the seawater carbonate system ( ar ; pH T , total scale) were calculated from measured A T , DIC, temperature, and salinity using the R package seacarb (Lavigne and Gattuso, 2013) with K 1 and K 2 constants applied from Mehrbach et al. (1973) and refit by Dickson and Millero (1987) and the total borate-to-salinity relationship adapted from Lee and Millero (1995).Because changes in A T could be due to processes other than the precipitation and dissolution of carbonates (e.g.sulfate reduction associated with organic matter additions), fluxes in DIC were corrected for the assumed A T fluxes due to calcium carbonate precipitation/dissolution (0.5 mol CO 2 : 1 mol A T ) and compared against fluxes in O 2 , with an expected 1 : 1 molar flux ratio (DIC org : O 2 ).
Prior to chamber additions, subsamples (1 mL, n = 3) were taken from the concentrated PD culture, CM, and the water column and analysed for particulate organic carbon (POC) and nitrogen (PON).These subsamples were filtered on pre-combusted 25 mm GF/F filters, dried at 60 • C, fumed with 12 M HCl to dissolve any particulate carbonates on the filter, and wrapped in pre-combusted tin capsules.These capsules were analysed for carbon (C) and nitrogen (N) using an elemental analyser (Thermo Flash ES) coupled to an isotope ratio mass spectrometer (Thermo Delta V PLUS) via a Thermo Conflo V (see Eyre et al., 2016, for details).

Calculating sediment metabolism
Benthic metabolism (NPP, GPP, R, G net ) in each chamber was estimated based on the fluxes of measured solutes (DO, and A T respectively).For flux calculations, DO was converted from mg L −1 to mmol L −1 .A T and DIC were converted from µmol kg −1 to mmol L −1 using calculated temperature-and salinity-dependent seawater density.The solute flux equation (Glud et al., 2008) was as follows: where ) is the area of sediment enclosed by the chamber, and t (hours) is the time elapsed between seawater samplings.Rates of sediment net primary production (NPP), gross primary production (GPP), and respiration (R) were calculated from O 2 fluxes (mmol O 2 m −2 h −1 ), and rates of net sediment calcification (G net ) were calculated from A T fluxes (mmol CaCO 3 m −2 h −1 ; Table 2).Both NPP and GPP are reported as positive values to represent flux of O 2 from the sediment into the chamber water column, whereas R is reported as a negative value to represent the flux of O 2 from chamber water column into the sediment.To calculate the GPP / R ratio, positive values of R were used.To determine the sensitivity of GPP and R to changes in temperature, the absolute difference in diel GPP and R (mmol O 2 m −2 d −1 ) between the control and warming treatments was divided by the increase in temperature (2.4 ± 0.5 Additionally, to provide comparability with the literature and determine the numerical relationship between a 10 • C change in temperature and GPP and R, Q 10 values were estimated for temperature treatments according to the following equation: where M1 is the metabolic rate (GPP or R) at temperature T 1 (control) and M2 is the metabolic rate (GPP or R respectively) at temperature T 2 (warming treatment), with T 1 < T 2 .

Statistical analyses
Results are displayed as the mean ± standard deviation (SD).Data were organised as the hourly average for both day and night and were pooled together within each T , OM, and T + OM treatment where results did not significantly differ between incubations.All statistical analyses were performed with the SPSS statistics software (SPSS Inc.Version 22.0) running in a Windows PC environment, and the assumptions of normality and equality of variance were evaluated with graphical analyses of the residuals.To test for the effect of each treatment (T , PD, and CM) on respiration, photosynthesis, and calcification, measured R, NPP, GPP, and G net were analysed using a repeated-measures three-way analysis of variance (ANOVA).In this model, temperature and OM (PD and CM) were fixed effects, the within-subject factor was time (days), and replicate chambers were a nested effect.
To compare the significance of temperature and OM between and within treatment chambers, a one-way ANOVA model was used in which average seawater temperatures ( • C) and POC and PON concentrations respectively were treated as the response variable.In these analyses, Bonferroni post hoc tests were used to conduct pair-wise comparisons between treatments.

Measured seawater chemistry and sediment metabolism in control chambers
Temperatures measured in both the water column and chambers exhibited typical diel changes, and were slightly warmer in the controls (28.2 ± 1.3 • C) in comparison to the water column (−0.8 ± 0.5 • C; Fig. 2).Mean water column salinity throughout the experiment was 35.8 ± 0.1.Over the course of each diel incubation period, changes in water chemistry (Fig. 3) were driven by benthic metabolism.
Control (C) chambers, over the diel cycle, were net autotrophic and net calcifying.C chambers were net dissolving at night and net calcifying during the day.Mean particulate organic carbon (POC) and nitrogen (PON) concentrations in the four C chambers were 0.63 ± 0.1 µmol C L −1 and 0.12 ± 0.1 µmol N L −1 respectively.The DIC org : O 2 quotient for all treatments was 0.94 ± 0.09 on average and did not significantly differ from 1 (p < 0.05; Fig. 4), suggesting that sulfate reduction did not significantly contribute to the A T fluxes.

The effects of temperature on sediment metabolism
Mean seawater temperature in the C and temperature (T ) treatments during the four incubation periods was 28.2 ± 1.1 and 30.6 ± 1.2 • C respectively (Table 1).Temperature differed between C and T treatments (F 1,31 = 384.38,p < 0.05), but there was no significant difference between replicate chambers within each treatment (F 1,31 = 0.76, p = 0.768).Temperature in all eight chambers exhibited typical diel changes throughout all four incubation periods, driven by sunlight and tidal changes in water depth (Fig. 2).
Treatment chambers followed the same natural diel change measured in control chambers and maintained an average +2.4 ± 0.5 • C offset over the course of the study (Table 1).
Values correspond to the mean ± SD.

The response in coral reef sediment metabolism to seawater warming
Under control conditions, rates of GPP, R, and G net were similar to those measured in advective benthic chambers simulating equivalent percolation rates (Table 4) over 24 h diel timescales.Furthermore, carbonate sediments were net au- ) and calculated gross primary productivity (GPP: mmol O 2 m −2 h −1 ), respiration (R: mmol O 2 m −2 h −1 ), the ratio of GPP / R, and net calcification (G net : mmol CaCO 3 m −2 h −1 ) under ambient conditions.In this study, data from Cyronak et al. (2013a, b), Cyronak andEyre (2016), andTrnovsky et al. (2016) were collected in situ at Heron Island, Australia, while data from Lantz et al. (2017a) were collected ex situ in Moorea, French Polynesia.
This study −1.totrophic (GPP / R = 1.31 ± 0.1), similar to previous studies (Eyre et al., 2014).The sediments were net calcifying during the day under all treatment conditions, which was likely due to a combination of light-stimulated biogenic calcification by infaunal organisms (e.g.symbiont-bearing foraminifera: Yamano et al., 2000;or dinoflagellates: Frommlet et al., 2015) and by a photosynthetically mediated increase in porewater aragonite saturation state to a value that would allow for abiotic precipitation ( > 8; Cohen and Holcomb, 2009).However, the exact organisms and geochemical conditions responsible for the measured net diurnal calcification signal was beyond the scope of this study and should be examined in future work.It should also be noted that the daytime incubations in this study were terminated at 16:00, 2 h before sunset (18:00), to allow time to move each chamber and establish new treatment conditions for the next set of incubations.It is therefore possible that the calculated daytime GPP was slightly overestimated given that the sediments in these final 2 h before sunset generally exhibit a lower rate of oxygen production relative to the 06:00 to 16:00 time period due to a reduction in light intensity (Cyronak et al., 2013b).However, a comparison of the mean GPP in control chambers to prior chamber work at the same study site, where incubations lasted until sunset (Cyronak and Eyre, 2016;Table 4), shows that GPP in this study was lower.This suggests that temporal variability in light intensity, temperature, and other abiotic factors likely exerts a greater influence on GPP than a 2 h difference in incubation period.
In our experiments, seawater warming (+2.4 ± 0.5 • C) was within the projection of the IPCC RCP 8.5 (+2.2-2.7 • C).Under this elevated seawater temperature, R increased to a greater extent than GPP, shifting the sediments to net heterotrophy (GPP / R = 0.93) over the diel incubation period (Fig. 8).The decrease of GPP / R due to warming can be explained by the relatively lower temperature sensitivity value for GPP (16. . This is further supported by the relatively lower measured Q 10 value for GPP (7.3 ± 1.2) compared to R (10.7 ± 3.1), similar to those measured by Trnovsky et al. (2016) for GPP (3.1-4.1) and R (7.4 to 13.0).It is important to note that the established Arrhenius relationships in the literature suggest that development and growth rates should increase at a rate of 7-12 % • C −1 of warming (Clarke, 2003), much lower than the observed 74 and 42 % increase in R and GPP respectively per 1 • C of warming in this study.However, recent work in the Antarctic by Ashton et al. (2017) on marine benthic assemblages showed that, in some species, the growth rate exhibited a 100 % increase per 1 • C of warming, yielding Q 10 values around 1000.Therefore, while the temperature sensitivity estimates reported in this paper and in Trnovsky et al. (2016) exceed the expected rate for biological reactions and enzyme activity, evidence exists in other benthic marine environments to support the notion that the impact of temperature on biochemical processes may be more complex than previously thought at the organism level (Ashton et al., 2017).
Overall, the response in GPP / R to temperature agrees with other studies showing that seawater warming preferentially enhances R to a greater degree than GPP in marine sediments (Hancke and Glud, 2004;Weston and Joye, 2005;Tait and Schiel, 2013).The decline in GPP / R in response to warmer seawater temperature may be a product of the differential ranges in activation energies for GPP and R (Yvon-Durocher et al., 2010), where R exhibits a stronger and more rapid physiological acclimation to warming compared to GPP during short-term temperature variations (Wiencke et al., 1993;Robinson, 2000).The observed 29 % decrease in GPP / R in response to warming leads to a net 109 % decrease in G net (relative to control chambers), resulting in a transition to net sediment dissolution over the diel incubation period (Fig. 8).This decrease in G net was most likely due to a respiration-driven increase in porewater pCO 2 (e.g.Cyronak et al., 2013a), thereby decreasing pH and the mean porewater aragonite saturation state, as evidenced by decreasing water column levels (mean arg = −0.7 relative to control chambers).While rising T increases arg geochemically, with less than 0.03 units per degree of temperature increase, this effect is negligible and by far outweighed by biologically driven changes in arg , leading to an overall decrease.In summary, a warming of seawater by 2.4 • C decreased GPP / R by 0.38 units and G net by 0.2 mmol CaCO 3 m −2 h −1 in the permeable calcium carbonate sediments at this study site on Heron Island.The decline in the GPP / R in response to warming implies that a greater fraction of the carbon fixed by autotrophs was remineralised by heterotrophic bacteria and released as CO 2 , thus compromising the capacity of coral-reefpermeable carbonate sediments to remain net autotrophic at an elevated seawater T .
While a decline in marine sediment GPP / in response to seawater warming has been previously reported in several studies (e.g.Woodwell et al., 1998;Hancke and Glud, 2004;Weston and Joye, 2005;Lopez-Urrutia and Moran, 2007), the response in G net has only been examined by Trnovsky et al. (2016).It is important to note that these results should not be extrapolated beyond 2100, where SST rises above +2.4• C. The T increase simulated in this study (+2.4 • C) was within the optimal temperature range (30.6 • C) of previously reported temperature-metabolism hyperbolic relationships in marine sediments (Yvon-Durocher et al., 2010).Given the nature of hyperbolic relationships a further increase in temperature will eventually have an opposite effect on sediment metabolism (net decrease in GPP and R; Weston and Joye, 2005).Thus, the temperature sensitivity reported here should not be extrapolated beyond 2.4 • C.

The response in coral reef sediment metabolism to organic matter enrichment
Increased concentrations of organic matter (OM), analogous to eutrophic conditions on degraded coral reefs, enhanced both GPP and R in the sediment, likely by releasing nitrogen and phosphorus via organic matter degradation.These results agree with prior work, where increased concentrations of OM were quickly aerobically degraded by bacteria within minutes (Maher et al., 2013) to hours (Ferrier-Pages et al., 2000) and enhanced GPP more than R (Glud et al., 2008;Eyre et al., 2008).While some of this OM was likely degraded in the water column, previous experiments (e.g.Wild et al., 2004b) have shown that the high permeability of carbonate sediments permits the transport of OM into the upper centimetres (1-4 cm) of the sand, where bacterial degradation rates can exceed those of the water column by a factor of 10-12 (Moriarty, 1985;Wilkinson, 1987).Phytodetritus (PD) and coral mucus (CM) enhanced respiration rates 1.1-and 0.6-fold respectively which was a less pronounced increase in R than the 1.5-fold increase observed by Wild et al. (2004b) using the same Acropora spp.mucus at Heron Island.This difference may be due to the fact their study used almost 3 times more CM (∼ 280 mL) per treatment than this study (94 mL).An increase in GPP / R to 1.7 one day following the deposition of coral spawning material at the same study site (Glud et al., 2008), was similar to the average increase in GPP / R to 1.6 observed under increased OM concentrations in this study.PD enhanced GPP and R to a greater degree than CM, which may be explained by the higher nitrogen content, or more precisely, the lower C / N ratio in the former.Particulate organic carbon additions differed by less than 10 % between PD and CM treatments, whereas particulate organic nitrogen addition (N) was almost twice as high by PD compared CM.In general, bacterial communities responsible for the cycling of nutrients in sediments are thought to be nitrogen limited (Eyre et al., 2013).Given the relatively short timescale (24 h) in which the response in sediment metabolism to OM was measured, we reason that the PD was more rapidly mineralised than CM due to a higher N content in the added PD (Eyre et al., 2016).
To our knowledge, this is the first experiment to examine the short-term relationship between OM degradation and G net in coral reef sediments.Our results show that increased concentrations of PD and CM both enhanced G net .Most likely the increase in G net was a product of the same biogeochemical mechanism influencing G net under seawater warming, whereby changes in GPP / R modify porewater pCO 2 and thus arg .In the case of OM, a preferential enhancement of GPP over R resulted in an increase in arg (mean arg = +0.6 relative to control chambers) and subsequent increase in G net (+1.4 mmol CaCO 3 m −2 h −1 relative to control chambers).While the results presented here are the first to report a positive OM-G net relationship specifically in permeable calcium carbonate sediments, a similar response has also been observed at ecosystem level in coral reefs (Yeakel et al., 2015), where increased offshore productivity in the Sargasso Sea over the course of several months lead to an increase in community G net on the adjacent Bermuda coral reef flat.Interestingly, this increase in G net in Bermuda coincided with a period of net heterotrophy on the reef.The difference in the G net -GPP / R relationship between the data in this study (OM increased GPP / R and increased G net ) and those in Yeakel et al. (2015; OM decreased GPP / R and increased G net ) may be a result of the timescale of observation.This implies that, should elevated concentrations of OM persist for an extended period of time (weeks to months), the immediate preferentially phototrophically mediated recycling of nutrients, and associated increased GPP / R and G net in coral reef sediments, may eventually shift to net heterotrophy despite the ability to maintain a positive G net .
www.biogeosciences.net/14/5377/2017/Biogeosciences, 14, 5377-5391, 2017 4.3 The response in coral reef sediment metabolism to a combination of seawater warming and organic matter enrichment The combination of seawater warming and increased concentrations of OM, for both PD and CM, enhanced GPP (+17 % relative to the temperature alone) and R (+11 % relative to temperature alone) but countered the effect on GPP / R and G net (no significant difference from the control).Given the effect of each of these treatments (T and OM) independently on sediment GPP / R and G net , this result is not surprising.
A decrease in GPP / R and G net due to warming was countered by an increase in GPP / R and G net due to an increased concentration of OM.This finding raises questions within the context of each treatment, as mean SST on coral reefs will continuously rise from now until beyond 2100, consistently affecting sediment metabolism.However, organic matter enrichment of permeable coral reef carbonate sediments is also likely to gradually increase due to enhanced algal production from elevated nutrients (Furnas et al., 2005), elevated terrestrial input of OM (Diaz-Ortega and Hernandez-Delgado, 2014) and enhanced mucus production due to enhanced terrestrial sedimentation (Alongi and McKinnon, 2005).As discussed above this longterm enrichment with OM will most likely make coral reef sediments more heterotrophic (and not more autotrophic as in this short-term study).However the subsequent response in G net over longer timescales is less clear, as some work has shown that the degradation of organic matter can enhance sediment dissolution (Andersson, 2015), whereas other work (e.g.Yeakel et al., 2015) has shown that community calcification may actually increase.Therefore, combined with an increase in T , the effect of long-term enrichment of OM on GPP / R is likely to be additive (decrease GPP / R), but the long-term response in G net still needs further examination.
Similarly, the effect of other, more persistent products of eutrophication, namely dissolved inorganic nutrients (DIN: NH + 4 , NO − 3 , PO 3− 4 ), on coral reef sediment GPP / R and G net have yet to be studied and may become more frequent and persistent as coastal land use changes continue to facilitate the increased runoff of fertilisers (Koop et al., 2001).Consequently, the results presented here provide an estimation of the future short-term response in coral reef sediment GPP / R and G net to warming (+2.4 • C) and eutrophication (PD and CM), but by no means have explored other potential warming-and eutrophication-mediated perturbations that continue to threaten coral reef ecosystems.Future work should consider varying durations (e.g.> 24 h) and forms of eutrophication (e.g.DIN) as well as a range of T , both within and beyond reported optimal ranges (> 2.4 • C), to better constrain our understanding of the potential feedback responses in coral reef sediment GPP / R and G net .

Conclusions
This study suggests that seawater warming will shift GPP / R and G net in permeable calcium carbonate coral reef sediments to a state of net heterotrophy and net dissolution respectively by the year 2100.In contrast, short-term eutrophication, and the subsequent production of OM in the form of phytodetritus and coral mucus, could enhance sediment GPP / R and G net .The combined effect of seawater warming and increased concentrations of OM may additively enhance sediment GPP and R, but the net effects on GPP / R and G net will likely counter one another on relatively short timescales of days.The future response in the net flux behaviour of CO 2 and O 2 in the coral reef sediment environment, and the consequent rate of carbon sequestration into the sediments, will likely depend on the relative frequency and duration of each perturbation.The effects of OM (e.g.phytoplankton growth, reef-wide mucus secretion) on sediment metabolism generally persist temporarily (days to weeks) relative to global warming, a constant process which will continue to occur throughout this century and beyond.Provided this ecological context and the findings from this study, we propose that increased concentrations of OM, in the form of phytodetritus and coral mucus, will increase G net and GPP / R in the sediment on relatively short timescales.However, once seawater temperature on coral reefs rises 2.4 • C above the present-day mean, the immediate effect of OM on sediment metabolism will be compromised by a warming-mediated net decrease in G net and GPP / R, thereby limiting the ability of permeable calcium carbonate sediments on coral reefs to accumulate calcium carbonate.
Competing interests.The authors declare that they have no conflict of interest.

Figure 1 .
Figure 1.Layout of the experimental design using benthic chambers.Eight chambers were used in total, which provided two replicates per treatment.Chambers are organised by the presence (+) and absence (−) of the warming (+2.4 • C) and organic matter (OM; phytodetritus or coral mucus) treatments.

Figure 2 .
Figure 2. Water column parameters measured during the four incubations, each starting at sunset (18:00) and ending at the following day's dusk (16:00).Data are presented from the first phase (Incubation 1 and 2) where phytodetritus was used as an organic matter (OM) treatment, and from the second phase (Incubation 3 and 4), where coral mucus was used as an OM treatment.Shaded grey bars represent night-time.(a) Mean temperature ( • C) measured by Hobo temperature recorders that logged temperature at 15 min intervals during each incubation period.Data are pooled together as the mean from control (grey dots) and warming (black dotted line) treatments (n = 4 per incubation).Mean water column temperature (n = 1 per incubation) shown as a black dashed line.(b) Measured light intensity (µmol quanta m −2 s −1 ) in the water column (black line) and water height (m) during each incubation period (grey dashed line).

Figure 3 .
Figure 3. Water chemistry (mean ± SD) measured and calculated during the four incubations.Control (C), warming (T ), phytodetritus (PD), coral mucus (CM), and combination (T + PD, T + CM) treatments are averaged over the two incubations (and replicate chambers therein) in which each respective OM treatment was used (n = 4).Shaded grey bars represent the dark, and time of sampling is labelled on the x axis.(a) Measured fluxes in dissolved oxygen (DO: µmol L −1 ).(b) Measured fluxes in total alkalinity (A T : µmol kg −1 ).(c) Measured fluxes in dissolved inorganic carbon (DIC: µmol kg −1 ).(d) Calculated changes in pH (total scale: pH T ).(e) Calculated fluxes in aragonite saturation state ( ar ).

Figure 4 .
Figure4.A linear correlation between calculated changes in dissolved inorganic carbon ( DIC org : µmol kg −1 ) as a function of measured changes in dissolved oxygen ( DO: µmol L −1 ) over each 12 h sampling period from all chambers and incubations.To examine the variation in DIC due solely to photosynthesis and respiration (DIC org ), changes in DIC were corrected for calcium carbonate precipitation/dissolution using the measured changes in total alkalinity (A T ; 0.5 mol CO 2 : 1 mol A T ).

Figure 5 .
Figure 5. Mean sediment gross primary production (GPP: mmol O 2 m −2 h −1 ) and respiration (R: mmol O 2 m −2 h −1 ) in response to warming (+2.4 • C) and each OM treatment (phytodetritus and coral mucus).Control (C; n = 9) and warming (T ; n = 7) treatments are averaged over all four incubations and the replicate chambers therein.Phytodetritus (PD), coral mucus (CM), and combination (T + PD, T + CM) treatments are averaged over the two incubations (and replicate chambers therein) in which each respective OM treatment was used (n = 4).Average measured rates ± SD are represented in white for GPP (positive) and grey for R (negative).

Figure 6 .
Figure 6.Ratios of sediment gross primary production (12 h) to respiration (24 h; GPP / R) in response to warming (+2.4 • C) and each OM treatment (phytodetritus and coral mucus).Control (C; n = 9) and warming (T ; n = 7) treatments are averaged over all four incubations and the replicate chambers therein, while phytodetritus (PD), coral mucus (CM), and combination (T + PD, T + CM) treatments are averaged over the two incubations (and replicate chambers therein) in which each respective OM treatment was used (n = 4).Dashed grey line represents the divide between net heterotrophy and net autotrophy (GPP / R = 1), while the * indicates if the presented value is significantly different the control.

Figure 7 .
Figure 7. Mean sediment net calcification (G net : mmol CaCO 3 m −2 h −1 ) in response to warming (+2.4 • C) and each OM treatment (phytodetritus and coral mucus).Control (C; n = 9) and warming (T ; n = 7) treatments are averaged over all four incubations and the replicate chambers therein, while phytodetritus (PD), coral mucus (CM), and combination (T + PD, T + CM) treatments are averaged over the two incubations (and replicate chambers therein) in which each respective OM treatment was used (n = 4).Average measured rates ± SD are represented in white for light G net (positive) and grey for dark G net (negative).Black bars represent the 24 h diel G net averaged from light and dark measurements, and the * next to these bars indicates if the value is significantly different from the control.

Figure 8 .
Figure 8. Measured metabolic rates from the control (C; n = 9) and warming (T ; n = 7) treatments are displayed from all four incubations and the replicate chambers therein.Phytodetritus (PD), coral mucus (CM), and combination (T + PD, T + CM) treatments are displayed from the two incubations (and replicate chambers therein) where each respective OM treatment was used (n = 4).(a) Respiration (R: mmol O 2 m −2 d −1 ) plotted as a function of gross primary production (GPP: mmol O 2 m −2 d −1 ).Dashed line represents the divide between net heterotrophy and net autotrophy (GPP / R = 1).(b) Dark dissolution (dark G: mmol CaCO 3 m −2 d −1 ) plotted as a function of daytime calcification (diurnal G: mmol CaCO 3 m −2 d −1 ).Dashed line represents the divide between net calcification and net dissolution (G net = 0).

Table 1 .
Concentrations of carbon (µmol C L −1 ) and nitrogen (µmol N L −1 ) and measured temperature ( • C) in the control and treatment chambers.Values correspond to the mean ± SD.

Table 2 .
Eyre et al., 2011) in this study to calculate rates of sediment metabolism based on measured fluxes in dissolved oxygen (DO) and total alkalinity (A T ;Eyre et al., 2011).

Table 4 .
A comparison of studies which employed the same methodology (advective chamber incubations) under a similar advection rate (∼ 43 L m −2 d −1