Ammonia oxidation is a key microbial process within the marine N-cycle. Sediment and water column samples from two contrasting sites in the English Channel (mud and sand) were incubated (up to 14 weeks) in CO2-acidified seawater ranging from pH 8.0 to pH 6.1. Additional observations were made off the island of Ischia (Mediterranean Sea), a natural analogue site, where long-term thermogenic CO2 ebullition occurs (from pH 8.2 to pH 7.6). Water column ammonia oxidation rates in English Channel samples decreased under low pH with near-complete inhibition at pH 6.5. Water column Ischia samples showed a similar though not statistically significant trend. However, sediment ammonia oxidation rates at all three locations were not affected by reduced pH. These observations may be explained by buffering within sediments or low-pH adaptation of the microbial ammonia oxidizing communities. Our observations have implications for modeling the future impact of ocean acidification on marine ecosystems.
 Sustained observations and modeling studies have shown that oceanic pH is currently decreasing by 0.02 units per decade and is expected to decrease by up to 0.7 units by the year 2100 as a result of atmospheric CO2 dissolution in the oceans [Bates, 2007; Caldeira and Wickett, 2003; Olafsson et al., 2009; Santana-Casiano et al., 2007]. This ocean acidification (OA) is predicted to affect marine ecosystems through the alteration of community structure [Fabry et al., 2008]; nutrient cycles [Hutchins et al., 2009]; productivity [Riebesell et al., 2007] and carbon export [Mari, 2008; Schulz et al., 2008]. Of significant concern is the potential impact of OA on the processes involved in the biogeochemical cycling of nitrogen (N) and, in particular, nitrification; a key microbial process through which remineralized N in the form of ammonia (NH3) is oxidized to inorganic nitrite (NO2−) and subsequently to nitrate (NO3−). A number of recent and past studies have demonstrated an inhibitory effect of decreasing pH on water column nitrification activity [Beman et al., 2011; Huesemann et al., 2002; Jones, 1992; Stein et al., 1997]. The implications of such an effect on ecosystem function could be significant with modeling studies suggesting that substantial reductions in water column nitrification over the next century would affect nutrient stoichiometry, denitrification and by extension marine productivity and the biological carbon pump [Beman et al., 2011; Blackford and Gilbert, 2007]. However, the impact of OA on nitrification within sediments has not been examined. It is conceivable that alkalinity generation in sediments [Thomas et al., 2009] and/or low-pH-adaptation by microorganisms may moderate the effects of decreasing pH on nitrification in sediments. Here we test the hypothesis that CO2-induced acidification causes a reduction in water-column and surface-sediment NH3 oxidation using data generated from two controlled manipulative incubations and a set of field-measurements in the vicinity of natural CO2 vents.
2. Sampling and Methods
2.1. Experimental Manipulations
 Sediment was collected from Plymouth Sound (hereafter PS; 50.35°N, 4.13°W) and the Eddystone shell gravel site (hereafter ED; 50.19°N, 4.28°W) in the western English Channel during January 2010 and September 2010 respectively. The ED station is characteristic of open continental shelf regions while the PS site is characteristic of coastal waters that are influenced by freshwater inputs. Sediments at these sites are dominated by sand-shell fragments (ED) and fine mud (PS). ED and PS sediment cores (0.38 m height, 0.18 m diameter and 0.65 m height, 0.30 m diameter respectively) were kept under seawater at ambient temperature (11.5 ± 1.3°C) and light in the Plymouth Marine Laboratory mesocosm facility for 4–6 days (ED) and 8 weeks (PS) prior to the start of experiments. PS sediments stored for 8 weeks had similar NH3 oxidation rates to ongoing measurements at this site (no storage). All pH measurements were carried out in the overlying water with a glass electrode pH-meter (Metrohm, pH-826) against the NBS scale. For ED sediment incubations, overlying water in the cores was flushed with seawater from five header tanks (19 ± 9 mL min−1), with each tank acidified by purging with CO2(g) to set pH values of 6.1, 6.7, 7.1 and 7.5. A control sample of pH 8.0 was purged with air. Seawater for the header tanks was topped up continuously from a master tank filled weekly with seawater from the English Channel (salinity 33.8 ± 0.5). Header tank pH, monitored weekly, fluctuated by <0.2 units in each treatment over the course of the incubations. After two and ten weeks replicate cores (3 and 4 cores respectively per pH treatment) were removed for sampling. For the PS sediments, replicate cores (n = 4 per pH treatment) were incubated for 14 weeks in 1 m3 tanks which were filled with seawater and individually acidified with CO2(g) to pH values of 6.8, 7.3, 7.7, 7.9 and 8.1 (control). At the end of these incubations, 24-hour slurry-type incubations were set up in 14 mL glass vials with approximately 2/3 surface sediment (top 1 cm) and 1/3 seawater from the corresponding header tank in each vial. A further set of vials was filled with header tank seawater only, to determine NH3 oxidation rates in the overlying water. Average NH3 concentration (0.15 ± 0.16 μmol L−1) did not differ between pH treatments for our experiments. Aliquots (0.1 mL) of 0.1 M allylthiourea (ATU) and sodium chlorate (NaClO3) were added to separate vials (n = 3 per ATU/NaClO3-treatment, per core). ATU and NaClO3 are inhibitors of NH3- and NO2− -oxidation respectively [Hynes and Knowles, 1983]. Incomplete inhibition of nitrification by ATU has previously been observed and attributed to the resilience of some nitrifying archaea [Santoro et al., 2010]. However, there was no evidence of this in our study as evidenced by the absence of NO2− production in ATU treated samples compared to the start of incubations. The vials were sealed with rubber septa and incubated in the dark for 24 hours at 12°C. At the end of each incubation, the supernatant in each vial was filtered (0.7 μm, Whatman GF/F) and the filtrate NO2− concentration was determined by manual colorimetric assay against six NaNO2 standards (Sigma; >99.9% purity; standard range: 0.0–2.0 μmol L−1; R2>0.999) [Grasshoff, 1983]. NH3 oxidation rates were calculated as accumulation of NO2− in the NaClO3 treatment compared to the ATU treatment. The precision (0.3 μM NO2− standard; n = 3) and detection limit of NO2− analysis were 5 nmol L−1 and 9 nmol L−1 respectively (NO2− concentration in all samples was >10 nmol L−1). The corresponding limits of detection for rate calculations were 0.3 nmol L−1 h−1 in water and 0.2 nmol L−1 h−1 in sediments. The organic matter content of the sediments (% LOI), determined by weight loss after baking the vials at 350°C overnight, was 2.5 ± 0.2% (PS) and 1.7 ± 0.3% (ED) and did not differ between pH treatments.
2.2. Field Observations
 Sediment samples (muddy-sand) for nitrification were collected by diver in May 2008 from three sites (N1-3) in the vicinity of natural CO2 vents off the island of Ischia in the Mediterranean Sea (40.73°N, 13.95°E). CO2 ebulition has likely occurred at these sites for centuries as fluid seepage (thermal springs) has been known on Ischia since Roman times [Buchner, 1965]. Seawater pH at each of these sites was 8.2 (N1), 7.9 (N2) and 7.6 (N3), although advection of high-CO2 water causes substantial pH variability (± 0.2) at each site [Hall-Spencer et al., 2008]. NH3 oxidation rates were determined as described above, except water from each of the three Ischia sites was used to fill the glass vials.
 Ammonia oxidation activity for Ischia samples, under long-term exposure to low pH, are shown alongside water column and sediment data from the Plymouth Sound (PS) and English Channel (ED) acidification experiments (Figure 1). NH3 oxidation rates measured here were comparable to previously published data for coastal waters and sediments [Carini et al., 2003; Henriksen et al., 1981; Jensen et al., 1996]. Average (±S.E.) water column NH3 oxidation rates at the control pH were 2.4 ± 0.1 at PS and 3.1 ± 0.2 and 4.6 ± 0.5 nmol L−1 h−1 at ED after 2 and 10 weeks. These rates were approximately 4- to 8-fold higher than at Ischia (N1) at pH 8.2 (0.6 ± 0.2 nmol L−1 h−1). In the water column, NH3 oxidation generally decreased with decreasing pH, both in the long-term exposed Ischia samples and in the English Channel samples incubated under different pH treatments. In both the PS and ED experiments, NH3 oxidation rates in the water column declined significantly with the reduction in pH (1-way ANOVA; p<0.01; all statistical tests incorporated the relative uncertainties of both NH3 oxidation rates and pH). At Ischia, whilst NH3 oxidation rates declined from N1 to N3, this decline was not significant (1-way ANOVA; p>0.1) (Figure 1). Differences in water column NH3 oxidation rates for the two highest pH treatments were only significant for ED after 10 weeks (T-test; p<0.05) and PS samples (T-test; p<0.001).
 In sediments, there was no significant difference between the mean NH3 oxidation rates under different pH within the ED, PS mesocosms or at the Ischia site (1-way ANOVA; p>0.05). At the control pH, NH3 oxidation rates were 5–28 nmol L−1 h−1 ml−1 wet sediment at PS; 4–57 nmol L−1 h−1 ml−1 wet sediment at ED and 97 ± 32 nmol L−1 h−1 ml−1 wet sediment at Ischia. Three cores from the 2-week ED experiment showed ∼3-fold higher than average NH3 oxidation rates (189 ± 7 and 278 ± 3 nmol L−1 h−1 ml−1 wet sediment at pH 8.0 and 206 ± 4 nmol L−1 h−1 ml−1 wet sediment pH 7.5). Nevertheless, there was no significant difference between mean NH3 oxidation rates under different pH treatments for this experiment (1-way ANOVA; p>0.05). Furthermore, mean sediment NH3 oxidation rates for ED samples after 2- and 10-weeks incubation at pH 8.0 were not significantly different (t-test; p>0.05).
4.1. Water Column NH3 Oxidation
 Our results confirm previous observations of low-pH inhibition of NH3 oxidation in the water column, but show no evidence of such an effect in sediments. Furthermore, this difference is consistent across three experiments with contrasting characteristics, suggesting that our conclusions are widely applicable. Relatively few studies have focused on the effect of ocean acidification (high CO2) on the physiology, activity or community structure of marine microorganisms. Pelagic mesocosm experiments have shown increased bacterial growth and protein metabolism under high CO2, but this was thought to be an indirect response to changing phytoplankton dynamics (community structure and exudation of organics) rather than a direct effect of high CO2 [Grossart et al., 2006]. Here, CO2-induced acidification of seawater has been shown to have a pronounced and detrimental effect on water column NH3 oxidation activity. In agreement with previous work, we observed near-complete inhibition of NH3 oxidation activity at pH 6.5 [Huesemann et al., 2002]. Yet, other studies on single-species cultures of Nitrosococcus oceanus and Nitrosomonas europaea found that their activity decreased, but remained detectable even at pH 5.5 and pH 5.4 respectively [Jones, 1992; Stein et al., 1997]. Importantly, differences in NH3 oxidation rates between the two highest pH treatments were only significant for PS (ΔpH = 0.2) and ED sediments after 10 weeks of incubation (ΔpH = 0.4). This shows that a small decrease in pH or short exposure to low pH may not necessarily affect NH3 oxidation. A certain degree of resilience to low pH may be expected, given seasonal pH variability of up to 0.3 units in the western English Channel (V. Kitidis et al., Seasonal dynamics of the carbonate system in the western English Channel, submitted to Continental Shelf Research, 2011). It has been suggested that the inhibitory effect of OA on NH3 oxidation may be due to substrate (NH3) limitation of the ammonia monooxygenase enzyme, as the NH3-NH4+ equilibrium shifts towards the latter at low pH [Huesemann et al., 2002; Stein et al., 1997; Suzuki et al., 1974]. Alternatively, the observed pH dependence of NH3 oxidation activity in the water column may be related to changes in microbial community structure. Such changes may result from relative differences in susceptibility of different taxonomic groups to low pH, though this remains speculative in the absence of further information.
4.2. NH3 Oxidation in Sediments
 In contrast to the water column, acidification had no significant effect on NH3 oxidation activity in surface sediments. Substantial variability in the magnitude of sediment NH3 oxidation rates was apparent between all PS and ED sediments under each pH treatment. This variability may have been caused by heterogeneity in organic matter content, which has been shown to suppress NH3 oxidation activity in lakes [Strauss and Lamberti, 2000]. However, we did not observe a correlation between organic matter content (% LOI) and nitrification rates here. Our data are apparently in contrast with our earlier studies which attributed a decrease in nitrate- and concomitant increase in ammonia-efflux from acidified sediments to a decrease in nitrification activity [Widdicombe et al., 2009; Widdicombe and Needham, 2007]. However, direct measurements of nitrification were not performed during these studies. Therefore, observed changes in nitrate and ammonia efflux during previous work may reflect differences in other N-cycling processes; dissimilatory nitrate reduction and denitrification; or a reduction in nitrification associated with the activity of macro fauna, which were adversely affected by low pH. Both microbial diversity [Laverock et al., 2010] and nitrification activity are enhanced in the burrow walls of bioturbating fauna [Henriksen et al., 1983; Welsh and Castadelli, 2004]. A decrease in nitrification may thus have occurred as a result of the deleterious effect of OA on macro-fauna during our previous work.
4.3. Potential Causes of Sediment/Water Column Difference
 It is possible that pH within sediments differed substantially from our pH measurements, made in the overlying water, However, this is unlikely given that only surface sediments (top 1 cm) were sampled here. The discrepancy between the effects of OA on NH3 oxidation in the water column and in sediments observed in our experiments may be explained by buffering of the pH perturbation within sediments and/or by low-pH adaptation of the sediment microbial community. This is a non-trivial distinction since pH buffering by sediments is a finite process dependent on their mineral content, while microbial adaptation would not cease under projected ocean acidification over the coming century. Firstly, the pH perturbation may have been buffered by alkalinity generation through the dissolution of carbonate minerals, including calcite, Mg-calcite and aragonite [Andersson et al., 2007]. Indeed, such alkalinity generation has been demonstrated for the PS sediments (H. Findlay et al., manuscript in preparation, 2011). Nevertheless, the experimental design maintained water pH at the nominal set values for this experiment. It is therefore unlikely that alkalinity generation negated the high CO2 perturbation. Secondly, the nitrifying microbial community present in sediments may be adapted to occasional, low pH exposure. Although this has not been shown in the marine environment, in soils there is evidence for a switch from bacteria-dominated to archaea-dominated nitrifying communities under acidic conditions [Nicol et al., 2008]. Furthermore, the community composition of nitrifying archaea inhabiting sediments has been shown to be distinct from that within the water column [Francis et al., 2005]. Given these differences in taxonomic composition between benthic and water column nitrifying communities, it is plausible that at least some benthic microorganisms may have evolved to cope with low pH while their water column counterparts do not. Following this argument, low-pH tolerant organisms may thereby be selected over time under future OA. Thirdly, cell aggregation may be important in maintaining nitrifying activity in acidic soils as shown by the reduction in the pH minimum of Nitrosomonas europaea grown in sand columns compared to liquid culture [Allison and Prosser, 1993].
4.4. NH3 Oxidation Under Long-Term CO2 Exposure
 It is possible that community composition changes or cell aggregation, may also take place under future ocean acidification. If so, our experiments suggest that the magnitude and function of benthic NH3 oxidation will remain unchanged. To test this hypothesis, we have examined NH3 oxidation activity in the water column and sediments off Ischia island, where continuous, long-term ebullition of thermogenic CO2 at the seafloor provides a natural analogue for future ocean acidification from anthropogenic CO2 [Hall-Spencer et al., 2008]. In agreement with our ED and PS experiments we observed a decrease in mean water column NH3 oxidation with decreasing pH for Ischia samples (70% reduction between pH 8.2 and pH 7.6). However, this decrease was not statistically significant possibly due to low pH adaptation of the NH3 oxidizing community (see above). Genetic adaptation or selection of the microbial community is certainly a realistic possibility given that fluid seepage (presumably including CO2) has been taking place in this region since Roman times. In agreement with our PS and ED sediment incubation experiments, we found that low pH had no impact on sediment NH3 oxidation (Figure 1). This agreement between mesocosm and in situ data for such contrasting environments suggest that NH3 oxidation in sediments is unlikely to be affected either by future ocean acidification or under an abrupt decrease in pH as might be expected from leakage of sub-sea floor stored CO2. This has implications for modeling studies investigating the impact of OA on marine ecosystems. For example, Blackford and Gilbert  predicted a 20% decrease in pelagic nitrification by 2100. Our data suggest that a separate parameterization of this effect is required for the benthic component of such ecosystem models. Furthermore, we highlight the need to identify the cause of NH3 oxidation resilience to acidification in marine sediments, in order to accurately predict the future response of this process.
 Our experiments have demonstrated that ocean acidification or CO2 leakage from sub-sea floor carbon storage will have a detrimental effect on water column ammonia oxidation activity. Yet in sediments, this effect was absent, i.e., NH3 oxidation rates were not affected by low pH. This pattern is repeated in samples collected from a natural analogue site where sediments and seawater are exposed to long-term (decades) CO2-induced acidification. It is possible that the nitrifying microbial community in sediments is to some extent adapted to low pH or that pH buffering occurred. However, it remains unclear how their function will respond to further environmental stress following exposure to low pH.
 This project was funded through the UK Natural Environment Research Council (NERC) Oceans 2025 program. B. Laverock was funded by a NERC algorithm studentship (NE/F008864/1).
 The Editor thanks Michael Beman and an anonymous reviewer for their assistance in evaluating this paper.