Nitrate fertilisation does not enhance CO2 responses in two tropical seagrass species

Seagrasses are often considered “winners” of ocean acidification (OA); however, seagrass productivity responses to OA could be limited by nitrogen availability, since nitrogen-derived metabolites are required for carbon assimilation. We tested nitrogen uptake and assimilation, photosynthesis, growth, and carbon allocation responses of the tropical seagrasses Halodule uninervis and Thalassia hemprichii to OA scenarios (428, 734 and 1213 μatm pCO2) under two nutrients levels (0.3 and 1.9 μM NO3−). Net primary production (measured as oxygen production) and growth in H. uninervis increased with pCO2 enrichment, but were not affected by nitrate enrichment. However, nitrate enrichment reduced whole plant respiration in H. uninervis. Net primary production and growth did not show significant changes with pCO2 or nitrate by the end of the experiment (24 d) in T. hemprichii. However, nitrate incorporation in T. hemprichii was higher with nitrate enrichment. There was no evidence that nitrogen demand increased with pCO2 enrichment in either species. Contrary to our initial hypothesis, nutrient increases to levels approximating present day flood plumes only had small effects on metabolism. This study highlights that the paradigm of increased productivity of seagrasses under ocean acidification may not be valid for all species under all environmental conditions.

Values are given as mean ± S.D. Carbonate system parameters were calculated using measured values of total alkalinity (A T ), total dissolved inorganic carbon (DIC), temperature and salinity on CO 2 calc software 52 .
Net primary production in T. hemprichii did not increase with pCO 2 or nitrate enrichment ( Fig. 1; Table 2). In addition, no significant changes in leaf and rhizome respiration with pCO 2 and nitrate enrichment were detected.
Carbohydrates translocation and storage. For both H. uninervis and T. hemprichii, pCO 2 manipulation did not affect sucrose-phosphate synthase (SPS) and sucrose synthase (SS) activity indicative of carbohydrate  (Table 2). Nutrient enrichment reduced SPS activity in H. uninervis leaves (LME: P = 0.040) ( Table 2), but overall the effects were of limited consequence for our hypotheses (see Supplementary Fig. 1). Non-structural carbohydrates in H. uninervis and T. hemprichii rhizomes showed no change to pCO 2 and nitrate enrichment ( Table 2).

Discussion
This study aimed to test whether seagrass productivity is affected by pCO 2 and nitrate ( − NO 3 ) enrichment, and whether pCO 2 drives the demand for nitrogen in seagrasses. In H. uninervis, net primary production (NPP) and growth rates increased with higher pCO 2 but were not affected by nitrate enrichment. However, in T. hemprichii, NPP and growth were not affected by either pCO 2 or nitrate enrichment. In H. uninervis, pCO 2 enrichment did not increase nitrate uptake or assimilation while nitrate uptake was higher in CO 2 -enriched (simulating end of century RCP 8.5 emission scenario) 30 Table 2. Linear mixed effects models for measured productivity response variables. Variables were analysed with pCO 2 as a continuous predictor and nitrate as a categorical factor. Individual aquarium tanks were included as replicates (N = 3), with two sub-replicate pots nested within aquaria. For net primary production, shoot and rhizome-root respiration, linear models were used for analysis, with aquaria as replicates (N = 3) and without nested sub-replicate pots. P-values < 0.05 are in bold.
Scientific RepoRts | 6:23093 | DOI: 10.1038/srep23093  in ambient) raised leaf nitrate reductase (NR) activity in T. hemprichii. Therefore, productivity responses to pCO 2 and nitrate enrichment varied between species with different growth strategies.
H. uninervis and T. hemprichii differed in productivity responses to pCO 2 enrichment after 24 days exposure. In H. uninervis, NPP increased by 1.1 units for every 100 μatm rise in pCO 2 , an increase slightly higher than the 0.9 units measured in the same species by Ow et al. 4 . Other fast-growing seagrass species that have increased photosynthetic rates with pCO 2 enrichment include Z. marina (250% increase at pH 6.2, relative to 338 μatm pCO 2 ) 5 and Z. noltii (34% increase at pH 7.9, relative to 360 μatm pCO 2 ) 32 . Leaf growth rates in H. uninervis were also enhanced in pCO 2 enriched treatments, with the highest leaf growth rates [6.2 ± 0.40 (S.E.) mm shoot −1 day −1 ] being slightly lower than that measured in the field [7.0 ± 1.24 (S.E.) mm shoot −1 day −1 ]. Aquaria experiments may impose potential artefacts on leaf growth due to transplantation stress, which were minimised by allowing for acclimation prior to experiments. However, as described below, light levels within experimental tanks, which were lower than that of nearby shallow reef systems, most likely explained the lower growth rates in aquaria.
In T. hemprichii, pCO 2 enrichment had no effect on NPP and growth rates after three weeks, in contrast to previous work on this species 4,7 . Jiang et al. 7 studied T. hemprichii from a nutrient-enriched meadow (0.8-4.6 μM NO 3 − + NO 2 − ) 33 and exposed to much higher CO 2 concentrations (25-1005 μM) compared to the present study (19-31 μM). T. hemprichii grown under high nitrogen might have utilised its pre-existing nutrients store 22 to supplement a rapid growth increase during strong CO 2 enrichment 7 . In the present study, T. hemprichii productivity did not appear to be nitrogen-limited (discussed below), indicating that light levels in experimental tanks, or phosphate availability in carbonate sediments 34 could have limited its growth response. Interestingly, leaf growth of T. hemprichii showed a transient rise with pCO 2 at day 10, but subsequently stabilised. This growth response to initial (short-term) pCO 2 exposure has been reported for T. hemprichii after 14 days of exposure 4 . However, NPP measured at the end of the experiment (22 days) suggest a downregulation in response to pCO 2 over time.
Nitrate addition did not increase NPP in H. uninervis. This was despite respiration rates of the rhizome-root complex in enriched pCO 2 being lowered with nitrate enrichment. Given the relatively large proportion of below-ground biomass for this species 35 , a reduction in rhizome-root respiration could be substantial for improving carbon use 36 . In the present study, lower sucrose phosphate synthase (SPS) activity 15,27 in H. uninervis exposed to nitrate enrichment suggested a decline in the export of fixed carbon from leaves, potentially due to reduced metabolic demand in the rhizome-root biomass 8 . Further quantification of nitrate uptake rates and of the activities of the key enzymes in the nitrogen assimilation pathway, nitrate reductase and glutamine synthetase 22,25 , revealed no effect of nitrate enrichment on nitrogen incorporation in H. uninervis.
Productivity in T. hemprichii did not increase with nitrate enrichment, even though nitrate enrichment increased nitrate uptake at high pCO 2 and assimilation in the leaves of T. hemprichii. Increased nitrate uptake and assimilation under water-column nitrate enrichment could be advantageous for seagrasses acclimatised to growing in a low-nitrogen environment 15 . This allows the plant to sequester and store nitrogen rapidly when it becomes available. Higher nitrogen content and a lowered C:N ratio were observed in nitrate-enriched T. hemprichii leaves. Therefore, nitrate enrichment appeared to have a greater influence on nitrogen incorporation in T. hemprichii than H. uninervis.
Overall there was no evidence in the present study that nitrate enrichment enhanced productivity responses to pCO 2 for either species. This was surprising as nitrogen had been suggested 32 and shown to limit the productivity  of marine macrophytes to pCO 2 enrichment 21 in subtidal rocky habitats. The experiment duration might not have been long enough for pCO 2 enrichment to induce a significant change in nitrogen demand (24 days vs 5 months 32 ), which may still be covered by pre-existing nitrogen-resources. Previous work reported increases in leaf tissue carbon-to-nitrogen (C:N) ratios in CO 2 enriched seagrasses 7,37 , which suggested nitrogen limitation in these plants. However, C:N ratios in both H. uninervis and T. hemprichii here revealed no evidence that pCO 2 enrichment led to the seagrasses requiring more nitrogen. In the Great Barrier Reef (GBR) region, seagrass growth was limited by nitrogen at some sites 13,38 . In the present study, leaf nitrogen content and C:N ratios of H. uninervis (N = 2.53%; C:N = 16.3) and T. hemprichii (N = 2.75%; C:N = 14.4) were similar to previous values measured in GBR seagrasses 39 . These were well above the values assumed to indicate nitrogen limitation (N = 1.8%; C:N = 20) 39,40 and suggest that the two species were not nitrogen limited. DIN levels in sediment pore-water and that adsorbed to sediments were not quantified here, but typical concentrations can be 200 times higher than in the water column 24 . Thus sediment pore-water may have supplied sufficient DIN to maintain productivity rates measured here. Another possible explanation for apparent nutrient sufficiency (C:N < 20) 41 is that light levels during the experiment, averaging 9 mol m −2 d −1 , were low compared to longer-term monitoring from shallow seagrass meadows in far north Queensland which typically reach 15-20 mol m −2 d −1 . Furthermore light levels dropped in the region of the study site (Cape York) in early 2014 39 . Lowered levels of natural light, relative to the typical levels available 39 , may also explain the limited productivity responses to pCO 2 . Carbon dioxide enrichment did not drive nitrogen demand in H. uninervis and T. hemprichii. In other marine macrophytes, CO 2 enrichment was shown to increase nitrate reductase activity 32,42 . Here, increased CO 2 availability did not affect nitrate uptake and assimilation (measured as nitrate reductase and glutamine synthetase activity) in H. uninervis, whereas the effect was dependent on nitrate enrichment in T. hemprichii. This is interesting as water column DIN concentrations at northern mid-shelf GBR (e.g. Lizard Island) are typically lower than that at inshore reefs 43 , where the majority of seagrass grows 44 . Perhaps experiments on longer time-scales are needed to evaluate the effects of nitrogen availability on productivity, as seagrasses possess mechanisms to improve nitrogen-use efficiency, likely through recycling or re-allocation of nitrogen within the plant 24 . At natural CO 2 seeps with elevated pCO 2 , no difference in tissue nutrients were found between seagrasses growing around, and away from the CO 2 seeps, suggesting CO 2 -induced nitrogen limitation was not present 45 . Continual flux in nutrients in coastal habitats, supplemented by nitrogen fixation in the sediments 46 , may enable seagrasses to be more productive without facing nitrogen limitation with future OA.
In conclusion, the tropical seagrasses, H. uninervis and T. hemprichii, did not appear to be strongly nitrogen limited despite being collected from a mid-shelf reef where ambient water column nitrogen concentrations were low (0.13 μmol DIN). Consequently, nitrate fertilization of the water column did have some effect on nitrate uptake rates, but did not enhance seagrass productivity or leaf growth rates. Furthermore, in contrast to our initial hypothesis, responses to pCO 2 enrichment, simulating future ocean acidification scenarios, were also unaffected by nitrate fertilisation. To better reconcile the effects of nutrient enrichment on seagrass CO 2 responses with previous studies, there is the need to account for differences in background light, nutrient levels and durations between experiments. This helps to circumvent the current experimental limitations in expanding our findings to a wider environment. Ocean acidification can also promote the growth of epiphytic filamentous algae, outweighing the influence of nutrient addition on seagrass epiphytes 47 . Nutrient enrichment could encourage a shift in the dominance of submerged vegetation, from seagrasses to fast-growing macroalgae and phytoplankton, such as that observed in habitats exposed to eutrophication 48 . Hence, while seagrass meadows may potentially flourish in a future where the oceans are enriched in CO 2 , ecological effects of ocean acidification and nutrient fertilisation, such as competition from macroalgae and epiphytes, may outweigh gains to seagrass productivity. Potted seagrasses were stored in outdoor flow-through aquaria (50 L) for three to six days prior to the initiation of the experiment. Experimental treatments consisted of three pCO 2 levels (ambient ~428 μatm, moderate ~734 μatm and high ~1213 μatm pCO 2 ) and two nitrate treatments (ambient ~0.3 μM and enriched ~1.9 μM) crossed in a fully factorial design. Each treatment comprised of three replicate 25 L aquaria leading to a total of eighteen aquaria, supplied with seawater at 24 L h −1 directly from the adjacent lagoon. Two sub-replicate pots of each species were placed in each aquarium. The aquaria were situated outdoors under a solid translucent roof, which attenuated 50% of down-welling light. 2π light loggers (Odyssey, New Zealand) were randomly allocated to aquaria to record photosynthetically active radiation (PAR). Over the course of the experiment, the net daily PAR in aquaria ranged from 1.2-5.2 mol m −2 d −1 , averaging 3.8 mol m −2 d −1 . Mid-day maximum PAR averaged to 480 μmol m −2 s −1 . Treatments were randomised between the aquaria to eliminate any potential environmental effects within the set-up area. The experiment ran for 24 days before it had to be terminated due to an approaching cyclone. pCO 2 concentrations were manipulated by injecting different amounts of CO 2 gas into sump tanks. pH levels in the sump tanks were monitored with six potentiometric sensors (± 0.01 pH unit) calibrated on the NIST (National Institute of Standards and Technology) scale as a proxy to control for CO 2 input. The sensors provide feedback to a control system that regulates pH levels via CO 2 gas injection (AquaMedic, Germany) 4 . We recognise that over natural seagrass meadows, seawater pH fluctuates and does not have a set point. However, such fluctuations are hard to emulate while controlling for pCO 2 concentrations with our current set-up. Hence, pCO 2 concentrations were controlled using fixed pH levels instead. Seawater pCO 2 concentrations in mid-shelf reefs, such as Lizard Island in the GBR averaged about 380 μatm (1 S.D. = 15 μatm) 49 during the dry season from 2011-2013. During the wet season, when the present experiment was conducted, pCO 2 concentrations tend to be higher (460 μatm; 1 S.D. = 33 μatm) than during the dry season 49 .

Methods
Across the Great Barrier Reef (GBR), DIN (nitrate, ammonium and nitrite) levels in the water column over seagrass meadows are relatively low, averaging 0.13 μM 11 . However, terrestrial run-off into coastal areas can deliver DIN loads that are an order of magnitude or more higher (1.54 to 7.02 μM, or 2.20 μM averaged across the GBR) 50 . Nitrate enrichment was achieved by dripping sodium nitrate solution (Sigma-Aldrich, Australia) into individual aquaria. Peristaltic pumps (Cole Palmer, USA) delivered 2 mM of NaNO 3 solution into the individual aquaria at a rate of 0.5 ml min −1 . Small aquaria pumps (Hailea, China) in each aquarium provided mixing.
Seawater chemistry. pH total in treatment tanks were monitored by spectrometric determination of m-cresol absorbance 51 , and additionally checked against TRIS seawater standard (A. G. Dickson, Scripps Institute of Oceanography, Batch 106). Weekly water samples were analysed for total alkalinity (A T ) by gran titration with 0.5 M HCl on a Metrohm 855 titrosampler (Metrohm, Switzerland), and for total dissolved inorganic carbon (DIC) by acid titration on a VINDTA 3C. Carbonate system parameters were calculated using measured values of A T , DIC, temperature and salinity on CO2calc software 52 . Duplicate water samples for dissolved inorganic nutrient analysis were filtered through 0.45 μm cellulose acetate filters and stored at − 20 °C before determination of seawater ammonium, nitrate, and phosphate concentrations according to standard procedures outlined in Ryle et al. 53 . Temperature in the treatment tanks was logged by HOBO tidbit loggers (Onset, USA) every 5 min. Salinity readings were taken from an IMOS weather buoy (Integrated Marine Observing System; www.aims.gov. au) situated in the lagoon.
Productivity. After 22 days, photosynthetic and respiration rates were measured using the second youngest leaf of a shoot from each sub-replicate pot using optical oxygen sensors ("optodes", PreSens, Germany) and a fiber-optic oxygen meter (PreSens Oxy 4, Germany). Respiration rates of below-ground rhizome with associated roots (~2.5 cm) from each pot were quantified similarly. Measurements were conducted in 70 mL chambers at constant 28 °C water temperature following procedures described in Ow et al. 4 . Respiration of the leaves and below-ground rhizome-roots were measured separately over a 20-min period in the dark while photosynthetic rates were measured on the same leaf at 400 μmol m −2 s −1 PAR over 30 min. Plant material was dried (60 °C for 48 h) and weighed after incubation. Photosynthetic and respiration rates were normalised to the dry weight of the leaf and rhizome. Optodes were calibrated according to protocol described in Collier et al. 54 .
Growth rates were measured according to the method described in Short and Duarte 55 . At day 0 and day 14 of the experiment, all shoots were marked at the top of the bundle sheath with a needle. Length of new tissue growth was measured with vernier callipers regularly throughout the experiment, totalled and normalised to the number of shoots and days since marking. Growth rates of plants, from three separate plots in each source meadow, were also obtained using the same method from day 13 to day 17.
Nitrogen uptake. Leaf nitrate uptake rates were estimated at the end of the experiment. Seagrass shoots were incubated in seawater enriched with 15 N labelled potassium nitrate (atom% = 98; Novachem, Australia), and the final 15 N in the leaf tissue was used to calculate the uptake of 15   . Incubations were carried out on individual shoots in their pots, in their respective treatment tanks, via a method similar to that described in Prado et al. 56 . Individual shoots were enclosed within a plastic bag (~250 mL volume) fitted with a filter cassette and a plug that could be sealed. No leakage was detected when tested using a food dye. Potassium nitrate solution was injected into the chambers to achieve around 20% 15 NO 3 − enrichment of the initial ambient DIN concentration 57 . The shoots were incubated for one hour at ambient mid-day temperature (28 °C) and light (450 μmol m −2 s −1 ). After one hour, the shoots were excised from the rhizomes and rinsed with deionized water to remove excess adherent label. Non-incubated leaf samples were collected from each tank to provide background leaf 15 N levels for each species. Leaf material was processed and measured for total nitrogen content and atom% 15 N according to method described in Takahashi et al. 45 . Uptake rates (μmol N g −1 dry weight h −1 ) of 15 NO 3 − were calculated following equations outlined in Nayar et al. 23  Nitrogen assimilation and carbon translocation. Plant material used for measuring nitrogen assimilation and carbon translocation (i.e. enzyme analyses), except for nitrate reductase (NR), were collected at the end of the experiment and stored in liquid nitrogen until analysis.
NR activity in fresh shoot tissue was determined using the in vivo assay described for Zostera marina 58 . The in vivo technique was shown to yield consistently higher activity than the in vitro assay, which often gave negligible readings 15 . Extraction and assay for glutamine synthetase (GS) activity in new and fully extended leaf tissue was carried out following the method developed for Z. marina 25 , except that the incubation was carried out nearer to the aquaria temperature (30 °C).
To study carbon translocation, sucrose-phosphate synthase (SPS) from young but fully extended shoot tissue and sucrose synthase (SS) from the root-rhizome complex were extracted using a technique described in Brun et al. 27 and assayed according to the protocol outlined in Zimmerman et al. 28 . The sucrose produced was quantified colorimetrically using anthrone assay 59 . Shoot and rhizome-root biochemistry. Shoot tissue nutrients (carbon and nitrogen) of ashed samples were analysed using an elemental analyser (Elementar Vario EL, Germany) interfaced to an isotope-ratio-mass-spectrometer (PDZ Europa 20-20, Sercon Ltd; Cheshire UK), as described in Takahashi Scientific RepoRts | 6:23093 | DOI: 10.1038/srep23093 et al. 45 . To study carbon storage, ground rhizome-roots samples were analysed for non-structural carbohydrates content according to procedure described in Collier et al. 35 . The summed amount of soluble carbohydrates and starch gave total non-structural carbohydrates (TNSC) content, expressed as milligrams dry weight −1 of tissue. Statistical analysis. Parameters were analysed using linear mixed effects models with pCO 2 as a continuous predictor, and nitrate (ambient and enriched) as a categorical factor. Individual tanks were included as replicates, with sub-replicate pots nested within tanks. The nested factor was omitted for parameters without sub-replicate measurements (T. hemprichii: net primary production, respiration; both species: 15 NO 3 − uptake). For these parameters, measurements were terminated prematurely due to an unforeseen evacuation of the research station caused by a cyclone, and therefore the second sub-replicate could not be measured. Assumptions of normality and homogeneity of variances were tested with Shapiro-Wilks' and Bartlett's tests, respectively. Percentage data (%C and %N) were arcsine square-root transformed to meet the assumptions 60 . All statistical tests were assessed at α = 0.05 and analysed using R statistical software (R Development Core Team).