Temperature response of denitrification and anammox

Introduction Conclusions References


Introduction
Denitrification and anammox are the dominant processes for nitrogen loss in the ocean.These microbially mediated reactions convert dissolved inorganic nitrogen to N 2 and are limited to suboxic and anoxic regions of the ocean, including oxygen minimum zones and surficial sediments (Thamdrup and Dalsgaard, 2008).A large portion of nitrogen removal occurs in coastal sediments, where high primary productivity and export production provide abundant organic matter and dissolved inorganic nitrogen that fuels denitrification and anammox.Current nitrogen budgets attribute up to 50 % of the global ocean nitrogen loss to near shore coastal Published by Copernicus Publications on behalf of the European Geosciences Union.and continental shelf sediments (Codispoti, 2007).Continental shelves are dominated (up to 70 %) by relatively coarsegrained, relict sediments (Emery, 1968); however, the contribution of these permeable continental shelf sediments to nitrogen removal remains poorly constrained (Galloway et al., 2004).
Recent field experiments and modeling efforts have demonstrated the central role of pore water advection in controlling rates of N removal in sandy permeable sediments (Huettel et al., 2014).The transport of solutes and particles by pressure-gradient-driven pore water advection enhances mineralization processes in permeable sediments by delivering dissolved and particulate substrates and removing metabolic end products (Huettel and Rusch, 2000).Pore water advection supplies nitrate from the overlying water and may enhance coupled nitrification-denitrification as a result of an expanded oxic zone.Enhancement of denitrification by pore water advection has been shown to be highly influenced by the rate of pore water flushing and sediment topography.Denitrification rates generally increase with pore water flushing rate, but under high flushing rates or in highly permeable sands, the extent of anoxic zones may limit denitrification (Cardenas et al., 2008;Cook et al., 2006).Sediment topography may limit coupled nitrification-denitrification due to preferential flow of ammonium-rich water through anaerobic zones at ripple crests (Kessler et al., 2012).However, nitrification has been shown in some cases to be tightly coupled to denitrification at high flushing rates in sand under continuous advection (Gihring et al., 2010), indicating that zones of nitrification and ammonium production can be tightly coupled.
To parameterize large-scale models of nitrogen removal from continental shelves, the response of denitrification and anammox to environmental controls must be quantified.Major environmental controls of denitrification and anammox activity have been extensively studied in fine-grained sediments and include, but are not limited to, the availability of organic matter, nitrate concentration in the overlying water, O 2 penetration depth, and the extent of bioturbation (Cornwell et al., 1999).Temperature is often considered a master variable because it exerts a modulating effect on multiple physical and biogeochemical processes.The effect of temperature on microbial communities mediating denitrification is limited to a few studies.In previous studies from salt-marsh sediments, seasonal variation in the temperature response of denitrification rates showed that communities with lower optimal temperatures for denitrification became active during winter months (Kaplan et al., 1977;King and Nedwell, 1984).Ogilvie et al. (1997) demonstrated that lower temperatures led to the enrichment of denitrifying isolates, whereas higher temperatures enriched for bacteria that performed dissimilatory nitrate reduction to ammonium, demonstrating that temperature may have an effect on the dominant nitrate reduction pathway.Denitrification and anammox in Arctic fjord muds were shown to have optimal temperatures of 25 and 12 • C, respectively, indicating that the anammox bacteria were better adapted to cold conditions (Rysgaard et al., 2004).Together, these studies illustrate that the response of nitrogen removal pathways to the in situ temperature may be strongly dependent on the community structure.
Despite decades of research on the physiology and biochemistry of nitrate-respiring microorganisms, little is known regarding their metabolic response to temperature, especially under in situ conditions.In the present study, we examined the temperature adaptation of denitrifying and anammox communities in near-shore, permeable sediments spanning a 50 • latitudinal range.Permeable sediments were chosen because they dominate continental shelf sediments and may be important sites for nitrogen removal.We hypothesized that the temperature response of denitrification and anammox will vary according to the in situ temperature of the sediment.To this end, rate measurements employing the isotope pairing technique ( 15 N tracer) and temperature gradient block incubations were performed with permeable sediments from subtropical (northeastern Gulf of Mexico), temperate (German Wadden Sea), and Arctic (Svalbard, Norway) sites with annual mean temperatures of 23, 11, and −5 • C, respectively.The results demonstrate adaptation of denitrifying communities to in situ temperatures and suggest that marine anammox bacteria are psychrophilic to psychrotolerant.

Study sites
Near shore, permeable sediments were sampled at a total of 4 sites from subtropical, temperate and polar temperature regimes (Fig. 1, Table 1).The two subtropical sites were located at St. George Island (SGI), Florida (29.7 • N), which is a barrier island in the northeastern Gulf of Mexico.The annual mean water temperature at SGI is 23 • C (N.D.B.C. , Station SGOF1), and samples were taken in June 2010 from the subtidal zone located approximately 50 m from shore at 1.5 m depth.The temperate site was located at Hausstrand, the island of Sylt (55.0 • N), in the German Wadden Sea.The annual mean water temperature at Sylt in 2007 was 11.0 • C and comparable to the long-term mean  of 10.5 • C (Hedtkamp, 2005).Samples were taken on intertidal flats in March (mean temperature 5.6 • C) and June 2007 (mean temperature 17.9 • C).Polar permeable sediments were sampled in August 2010 on an intertidal flat at Ymerbukta (78.3 • N) on the island of Spitsbergen (Svalbard).This region has an annual mean temperature of −5 • C (Forland and Hanssen-Bauer, 2003), and sediments remain frozen from fall to spring (Sawicka et al., 2010).

Denitrification and anammox rates determined using intact cores
At each site, 15 cm-long sediment cores with 15 cm of overlying water were collected by hand using 3.5 cmdiameter core liners.Cores were immediately transported to temperature-controlled rooms and stored at the in situ temperature until incubation experiments were initiated (less than 24 h).All incubations were maintained at the in situ temperature measured at the time of core collection.Denitrification and anammox rate determinations were performed using the 15 N isotope pairing technique (IPT, Nielsen, 1992) with modification to account for anammox according to Risgaard-Petersen et al. (2003).For each site and sample date, IPT experiments were performed in the dark with and without pore-water perfusion as described in de Beer et al. (2005).
Overlying water was removed from the cores and replaced with air-equilibrated site water to which 50 µM Na 15 NO − 3 (99 atom %; Cambridge Isotope Laboratories, Andover, MA) was added.One set of cores was immediately capped without headspace and submerged in water collected from the field site.A second set of cores was perfused with 20 mL of Na 15 NO − 3 -amended site water by slowly pumping water with a syringe through a valve in the top stopper closing the core liners, while excess water was released through a valve in the bottom stopper of the core liners.The core diameter, sediment porosity, and volume of water perfused were used to calculate the depth of perfusion, which was approximately 5.0 cm for the subtropical and temperate sites and 7 cm for the polar site.Mixing of the water column in the cores was maintained by magnetic stir bars that were continuously rotated using an external magnet.Triplicate cores were sacrificed at each time point by the addition of 1 mL of ZnCl 2 (50 % w/v) to the water in the core liners, resealing the core without headspace, and inverting to homogeneously mixed sediment, ZnCl 2, and water.Water samples were then collected by a syringe while avoiding the introduction of air bubbles and transferred to 12 mL Exetainer ® vials containing 250 µL of 50 % ZnCl 2 added as a preservative.The total length and sampling interval of incubations was determined using preliminary data and was adjusted according to expected rates of N 2 production.Incubations lengths were 18, 55, and 72 h for Sylt, Ymerbukta, and SGI, respectively.
Linear regressions of excess 29 N 2 or 30 N 2 concentration (see analytical procedures) against time were used to calculate the rates of 29 N 2 and 30 N 2 production.All regression slopes were determined to be significantly greater than 0 (t test, p < 0.05), and the standard error of the regression slopes was used to calculate the error of denitrification and anammox rates.Volumetric calculations of denitrification and anammox were adjusted to areal rates using the sediment height, core diameter, and sediment porosity.At sites where anammox was negligible, the denitrification rate, D 14 , was calculated according to Nielsen (1992).Where anammox was detected, anaerobic slurry incubations were performed to determine the relative contribution of anammox to N 2 production (ra), and rates were calculated according to Risgaard-Petersen et al. (2003).The slurries were performed by adding 1 mL of homogenized wet sediment from the surface (0-5 cm) to a 12 mL Exetainer ® , which was then filled without headspace with site water containing 100 µM Na 15 NO − 3 .Vials were incubated for 24-48 h at the in situ temperature and sampled every 6 h in triplicate by addition of 250 µL of 50 % ZnCl 2 .The calculations of Thamdrup and Dalsgaard (2002) were used to determine the contribution of anammox.

Vertical profiles of denitrification and anammox
Vertical profiles of the potential rates of denitrification and anammox were determined in the Sylt sediments in June 2007.Cores were sectioned at 2 cm intervals, and sediment was slurried with bottom water in sealed gasimpermeable bags.The bottom water was amended with 100 µM Na 15 NO − 3 (99 atom %), and the bags were flushed with helium to ensure anaerobic conditions.The slurries were incubated at the in situ temperature.Samples for the determination of 29 N 2 and 30 N 2 production were taken at 3, 6, 12, and 20 h and preserved in 6 mL Exetainers ® with ZnCl 2 as described above.Potential denitrification and anammox were calculated according to the equations of Thamdrup and Dalsgaard (2002).Vertical profiles of pore-water nitrate, nitrite, and ammonium concentrations were measured at the time of core collection using Rhizon samplers (Rhizosphere Research Products) that were inserted into ports drilled at 1 cm intervals in polycarbonate core liners (9.5 cm diameter) (Seeberg-Elverfeldt et al., 2005).Rhizons were rinsed with the first 0.5 mL of extracted pore-water, and 2 mL were collected for each 1cm-depth interval for a total depth of 10 cm.

Temperature gradient block experiments
The temperature response of potential denitrification and anammox rates was determined using temperature gradient block incubators (cf., Sagemann et al., 1998) located at either the Max Planck Institute for Marine Microbiology (Bremen, Germany) or Florida State University (Tallahassee, Florida).The two temperature blocks were identical in size: each consisted of a 2 m-long (cross section 15 cm × 15 cm) insulated aluminum block that was electrically heated at one end and cooled at the other by a circulating water bath.Each block had 4 parallel rows of 30 vertical wells (1.8 cm diameter, 15 cm length), with 3 rows for samples and one row for temperature monitoring.
At the time of intact core collection, samples for temperature gradient experiments were collected from the upper 5 cm of the sediments into pre-sterilized plastic containers and stored at the in situ temperature.Potential rates of denitrification and anammox were determined in time series incubations of anaerobic slurries across a temperature range of   −1.0 to 42 • C. Slurries were prepared by adding 1 mL of wet sediment to a 12 mL Exetainer ® vial followed by filling of the vials with helium-degassed site water that was amended with Na 15 NO 3 to a final concentration of 100 µM.One set of samples was sacrificed at each time point by the addition of 250 µL of 50 % ZnCl 2 to each vial, followed by vigorous mixing.Four time points were sampled for each experiment over a period of 15-24 h.Parallel anaerobic slurries were incubated at the in situ temperature and sampled in triplicate at each time point to verify homogeneity between vials.The production of labeled N 2 was calculated as described for the core incubations by the regression of 29 N 2 and 30 N 2 against time, and the calculations of Thamdrup and Dalsgaard (2002) were used to partition 29 N 2 and 30 N 2 production into potential denitrification and anammox rates.An additional temperature block experiment was performed to determine anammox activity at each site, whereby production of 29 N 2 and the absence of 30 N 2 after addition of 100 µM 15 NH 4 Cl and 100 µM Na 14 NO − 2 tracer in anaerobic slurries indicated anammox activity.Nitrification can lead to artifacts when using a 15 NH + 4 tracer, but the absence of 30 N 2 production indicated that the slurries were sufficiently anaerobic to preclude nitrification.The rate of N 2 production by anammox was determined from the regression of 29 N 2 production against time, which was then corrected for the fraction of 15 N-labeled NH + 4 according to Holtappels (2011) The optimum temperature (T opt ) was visually determined from plots of rates against temperature.Additionally, a range for T opt was defined where rates did not fall below 90 % of the optimum rate.The slope of temperature response was modeled using the Arrhenius function.A regression through the linear range of the plot of inverse temperature against the natural log of denitrification (or anammox) rate provided an estimate of activation energy according the equation where k is the reaction rate, and T opt is the activation energy (kJ mol −1 ).In addition to the activation energy, the ratio of the potential rate at 5 • C to the rate at in situ temperature was calculated as an indicator of psychrophilic adaptation.

Analytical procedures
Dissolved nitrate + nitrite (NO − x ) and nitrite (NO − 2 ) were determined by chemiluminescence detection after reduction to NO gas on a Thermo model 42i NO x analyzer (Thermo Scientific).Nitrate + nitrite were reduced with an acidic VCl 3 solution (Braman and Hendrix, 1989), and nitrite was reduced using an acidic iodide solution (Garside, 1982).Dissolved ammonium (NH + 4 ) was determined by the colorimetric method of Bower and Holm-Hansen (Bower and Holm-Hansen, 1980).Concentrations of 29 N 2 and 30 N 2 in excess of natural abundance were measured by either isotope ratio mass spectrometry (IRMS) (Sylt) or membrane-inlet mass spectrometry (St.George and Ymerbukta).For IRMS, a gas sample from the helium-equilibrated (2 mL) headspace of each Exetainer ® was analyzed.The membrane inlet mass spectrometer was configured and calibrated according to An et al. (2001).An air-equilibrated standard of 0.2 µm filtered site water or seawater was maintained at in situ temperature in a thermostatic water bath and was analyzed every 20 min to correct for instrument drift.Oxygen was removed from the vacuum system inlet before entering the mass spectrometer using a heated (600 • C) quartz column with copper turnings.

Results
Sediment characteristics for each site are shown in Table 1.Sediments were comprised of fine-to-medium grained sands, except at the polar site, which contained a mix of sand and very fine gravel (Table 1).The median grain size was 211-223 µm at the subtropical sites and 338 and 824 µm at the temperate and polar sites, respectively.Permeabilities ranged from 1.0 × 10 −10 to 4.8 × 10 −11 m 2 and were sufficiently high to permit advective pore-water exchange (Huettel and Rusch, 2000).The organic carbon contents were < 0.1 % at the subtropical and temperate sites, and C : N ratios for these sites were between 7.1 and 8.4.The polar site had an organic carbon content of 0.49 % and a C : N ratio of 22.6.

Rates and pathways of N 2 production
The isotope pairing technique, modified according to Risgaard-Petersen et al. (2003), was used to estimate the rate of unlabeled N 2 production, p14, in intact 3.5 cm diameter cores.Areal rates of p14 varied between 1.9 and 359 µmol m −2 d −1 (Table 2).Simulated pore-water advection by perfusion led to substantial increases in N 2 production rates at all sites.An approximately 2-fold increase in p14 was observed upon perfusion of the subtropical sediments.Perfusion led to a larger increase of p14 in the winter temperate and polar (5-fold) and summer temperate (7-fold) sediments.The highest denitrification rate (359 µmol m −2 d −1 ) was observed with sediment perfusion at the temperate site.For all sites, denitrification was the dominant pathway for N 2 production.At Sylt, the contribution of anammox to N 2 production was 14 % in winter and 6 % in summer, and anammox was responsible for 6.5 % of N 2 production at Ymerbukta.The contribution of anammox was below the detection limit in sediments from St. George Island.

Vertical profiles of denitrification and anammox
Pore-water DIN profiles and potential rates of denitrification and anammox at Sylt (June) revealed a vertical structure of N-removal mechanisms within the sediment (Fig. 2).Pore-water DIN profiles showed elevated ammonium and nitrate concentrations below 4 cm.Concentrations were constant from 4 to 10 cm, and were approximately 12 µM for both nitrate and ammonium.Potential denitrification rates exhibited a maximum of 120 µmol L −1 d −1 between 2 and 6 cm, whereas, anammox gradually increased with depth to a maximum rate of 20 µmol L −1 d −1 at 8-10 cm.The anammox trend with depth was confirmed in parallel incubations with 15 NH + 4 tracer additions (data not shown).

Temperature response of denitrification and anammox
Potential rates of denitrification and anammox were determined between −1  block incubator.The optimum temperature for denitrification (T opt ) varied with the in situ temperature, with a T opt of 21 • C observed at the polar site and 35-36 • C observed at the subtropical sites, respectively (Fig. 3).The temperate site exhibited a seasonal pattern in the T opt of denitrification, with a T opt of 26 • C observed in the winter and 34 • C observed in the summer.Both the temperate and polar site had broad T opt peaks with widths of 7-9 • C (Table 3), whereas the subtropical site had well-defined, narrow peaks (2 • C or less).
At the polar site, a relatively high rate of denitrification was observed at 5 • C, corresponding to 34 % of T opt ; in contrast, rates at 5 • C comprised only 1 % of the T opt rate at the subtropical site (Table 3).The ratio of rates at 5 • C relative to the T opt at the temperate site was similar during both summer and winter field samplings (11-17 %).Activation energies (E a ) calculated from Arrhenius plots also indicated latitudinal variation in the temperature response of denitrification (Table 3).The E a values at the temperate and polar sites were similar at 52-65 kJ mol −1 .In contrast, the activation energies for denitrification at the subtropical site ranged from 102.7 to 123.5 kJ mol −1 , corresponding to a 1.5-2-fold higher activation energy than at the temperate and polar sites (Table 3, Supplementary material).Normalization of the temperature response curves to the denitrification rate at T opt revealed two trends: T opt increased with increasing environmental temperature, and temperate and polar sediments had higher relative rates at low temperatures as compared to subtropical sediments (Fig. 5).
The temperature response of anammox was only determined for Sylt and Ymerbukta, as anammox could not be detected at the subtropical SGI sites.Rates from the temperature gradient block experiments using 15 NH + 4 tracer additions are reported here rather than rates determined using a 15 NO − 3 tracer.Because denitrification rates were at least an order of magnitude higher than anammox rates in temperature block experiments, underestimation of the denitrification rate (estimated by production of 30 N 2 ) could produce a false anammox signal in experiments with 15 NO − 3 tracer.The use of 15 NH + 4 tracer is generally accepted as providing more conclusive evidence of anammox activity (Holtappels et al., 2011), and thus, these experiments were chosen as the best approach for estimating the anammox temperature response.Temperature optima of 9 and 26 • C were observed for anammox at the polar and temperate sites, respectively (Fig. 4).For both sites, anammox rates at 5 • C comprised 35 % of the T opt rate.The activation energy for anammox was 35 kJ mol −1 at the temperate site (Table 3), but activation energy could not be accurately calculated from the temperature response curve at the polar site.Normalization of the temperature response curves to the anammox rate at T opt clearly shows a higher T opt at the temperate site versus the polar site, but similar relative rates at lower temperatures are observed (Fig. 5).

Controls of N removal in permeable sediments
The controls of N removal at the temperate site at Sylt were the most extensively investigated based on previous results, indicating high rates of N removal.The Sylt-Romo basin receives nitrate from the Elbe and Weser rivers, with nitrate concentrations in the List tidal basin ranging from up to 60 µM in winter to less than 1 µM in summer (van Beusekom et al., 2008).In a recent study of sediment nitrogen dynamics in the same area, N 2 effluxes measured under diffusive conditions ranged from 144 to 1128 µmol N m −2 d −1 in February and April but N 2 production was not detected in summer when water column nitrate concentrations were low (Deek et al., 2012).In the present study, intact core incubations Rates were measured with a 15 NH + 4 label.Estimates of standard error were not made because rates were calculated from two time points.Note different y axis scaling.
showed higher rates of N 2 production in March relative to June under both diffusive and simulated pore-water advective conditions.These results indicate that seasonal variation in nitrogen loading is likely the dominant driver of the seasonal variability in N removal.The production of N 2 in winter months at Sylt is likely sustained by the advective transport of nitrate-rich water into the denitrification zone, whereas in summer, nitrate must be supplied from within the sediment.In June, pore-water profiles showed that 10-12 µM nitrate was present in the upper 10 cm of sediment, despite being nearly absent in the overlying water.This nitrate is likely supplied via nitrification, enhanced by bioturbation and bioirrigation of macrofauna (Huettel, 1990), as well as groundwater sources that have been shown to locally contribute inorganic nitrogen to the pore water at Sylt (Hedtkamp, 2005).Rates of anammox were approximately 10 % or less of denitrification, which agrees well with measurements in similar temperate sediments (Evrard et al., 2013).The rates from the present study add to a limited number of observations of N 2 production in advection-dominated permeable sediments.For sites with low bottom water nitrate (< 5 µM), previous isotope pairing technique incubations in sandy sediments under diffusive conditions have measured rates of denitrification between 34 and 108 µmol N m −2 d −1 (Deutsch et al., 2010;Gihring et al., 2010;Lohse et al., 1996;Vance-Harris and Ingall, 2005).Simulated advection in a subset of these studies has produced rates of denitrification between 99 and 480 µmol N m −2 d −1 (Cook et al., 2007;Gihring et al., 2010).In contrast to sites with low bottom water nitrate, rates from two distinct sites in the Wadden Sea with seasonally high water column nitrate (40-85 µM) were between 450-1128 µmol N m −2 d −1 under diffusive conditions, and substantially higher rates were observed (4000-25 000 µmol N m −2 d −1 ) with simulated pore water advection (Deek et al., 2012;Gao et al., 2012).Rates of sediment-water N 2 flux under continuous advective conditions are generally higher than the rates obtained by IPT (480-1680 µmol N m −2 d −1 ; Kessler et al. ( 2012) and references therein), indicating that the simulated pore water advection technique may underestimate N 2 removal rates from permeable sediments.It has yet to be demonstrated whether the rates observed using IPT at sites with high bottom water nitrate concentrations are higher under continuous advection.
Microbial community structure has an important role in determining N 2 removal under the dynamic redox conditions experienced in permeable sediments.A recent investigation of the kinetics of denitrification in permeable sediments found that the efficiency of denitrification (as indicated by the ratio of O 2 consumption to maximum denitrification rate) was much lower than the stoichiometric ratio of 0.8 expected for denitrification (Evrard et al., 2013).It was postulated that the microbial community was likely not able to immediately shift from oxygen to nitrate respiration, and thus, the rate of carbon metabolism was temporarily reduced.It follows that under low bottom water nitrate concentrations, dynamic topography and variation in flushing create a limited and shifting denitrification zone, which may preclude the establishment of a robust community of facultative anaerobes capable of denitrification.However, at sites with a higher concentration of bottom water (and pore water) nitrate (e.g., Sylt), it is likely that the community of denitrifying facultative anaerobes will be well established, and the kinetics of nitrate reduction at these sites may allow for rapid rates of N removal, possibly under aerobic conditions (Gao et al., 2010).More measurements of kinetics and community structure are needed to adequately examine this hypothesis.

Temperature adaptation of denitrification
The adaptation of denitrifying microbial communities to in situ temperatures was inferred from multiple parameters of the temperature response curves, including the temperature optima (T opt ), activation energies (E a ), and the ratio of denitrification at 5 • C to T opt , all of which varied directly with in situ temperatures.It is important to note that the T opt of growth rates are used to define psychrophiles and mesophiles according to the classification scheme of Morita ( 1975), but T opt for respiration rates were measured in the present study.Temperature gradient block incubations were carried out over short (< 24 h) timescales, and T opt values most likely reflect the optimal respiratory enzyme activity of the in situ microbial community, which may deviate from the T opt for growth.
The T opt for denitrification fell into two ranges: a T opt of approximately 20-25 • C for polar and winter temperate sites (Sylt winter) and a T opt of approximately 35 • C for subtropical and summertime temperate sites (Sylt summer).All T opt values were greater than the in situ sediment temperature, although the subtropical sites had the smallest difference between T opt and the in situ temperature, suggesting that these communities may be closer to their maximum efficiency at in situ temperatures than are the polar and cold temperate communities.However, many purified psychrophilic enzymes have optimal catalytic rates between 15 and 30 • C, and the primary adaptation in psychrophilic enzymes is the maintenance of relatively high specific catalytic rates at low temperature (Feller and Gerday, 2003).The T opt values for permeable sediments examined in this study are consistent with previous results from cold Arctic muds, where T opt values of 20-28 • C have been observed for aerobic respiration, denitrification, and sulfate reduction (Arnosti et al., 1998;Rysgaard et al., 2004).The community at the polar site (where T opt was lowest at 21 • C) may still be psychrophilic based on T opt values for growth, as denitrifying isolates obtained from this site had optimal growth rates between 5 and 15 • C (Canion et al., 2013).Similar results from pure cultures of psychrophilic sulfate-reducing bacteria corroborate our observations: values of T opt for growth were found to be up to 10 • C lower than those for sulfate respiration (Isaksen and Jorgensen, 1996).Results indicate that the properties of respiratory enzymes may be important in determining the T opt for growth of mesophiles, whereas metabolic pathways other than respiration limit the T opt for growth in psychrophiles.
Activation energy (E a ) can also be used as an indicator of adaptation to low temperatures in conjunction with the T opt for denitrification.The E a values for the denitrification temperature response integrate multiple enzymatic reactions of denitrifiers with a range of thermal adaptation, although lower values of E a generally indicate high relative rates at low temperatures as a result of psychrophilic adaptation.Subtropical sediments had E a values of 102-123 kJ mol −1 , approximately twice that of the polar and temperate sites, which is interpreted as the presence of an exclusively mesophilic community.In contrast, the lower E a values at the polar and temperate sites were indicative of a significant contribution of psychrophilic to psychrotolerant populations.The range of E a we observed for polar and temperate sites (50-65 kJ mol −1 ) is comparable to previous observations of E a of 60.6 kJ mol −1 for denitrification in Arctic marine muds (Rysgaard et al., 2004).The ratio of denitrification at 5 • C to T opt also confirms an increasing contribution of psychrophilic communities to denitrification across the gradient from subtropical to polar sediments.Specifically, polar sites exhibited rates of denitrification at 5 • C that were 34 % of the rate at T opt , whereas the temperate and subtropical sites had rates at 5 • C that were 14 and 1 % of T opt , respectively.
Seasonal variation in T opt was observed at the temperate site, Sylt, and the broadening of the T opt in summer, rather than a clear shift towards a higher T opt , suggests that psychrophilic to psychrotolerant denitrifying communities remain active in the summer when mesophilic activity is also apparent.Similarly, the E a for denitrification at Sylt was similar between summer and winter months, indicating that psychrophilic/psychrotolerant populations still contributed to denitrification in the summer.Seasonal variation in T opt for nitrate reduction has been observed in temperate salt-marsh sediments, where two distinct temperature optima for nitrate reduction at 12.8 and 28 • C were observed in winter and spring, respectively (King and Nedwell, 1984).These two temperature optima were interpreted as evidence of two nitrate-reducing communities with distinct temperature adaptation.In the present study, such a low temperature optimum in the winter was not observed; however, the T opt of 12.8 • C in the study of King and Nedwell (1984) may have been the result of a seasonal development of an anammox community (see below), but the study was conducted prior to the discovery of anammox.(Dalsgaard and Thamdrup, 2002;Rysgaard et al., 2004).Nitrification, which supplies nitrite for the anammox reaction, was found to have T opt values of 8-16 • C in Svalbard fjord muds, closely corresponding to the T opt for anammox (Thamdrup and Fleischer, 1998).The T opt of 26 • C at the temperate site is similar to previously reported T opt values of approximately 25 • C for enrichments of marine anammox bacteria performed at 15-30 • C (Kawagoshi et al., 2012;van de Vossenberg et al., 2008).The temperate T opt of 26 • C for anammox indicates a psychrotolerant rather than a mesophilic response (see the previous section for rationale).Based on previous observations of the T opt for respiration and growth on sulfate in pure cultures of mesophiles (Isaksen and Jorgensen, 1996), a T opt for anammox higher than 30 • C would be expected if the anammox community at the temperate site were dominated by mesophiles.Anammox bacteria detected in the marine environment belong almost exclusively to the genus Scalindua, within the Planctomycetes, and evidence for temperature-selected ecotypes has been found in enrichments from temperate fjord sediments (van de Vossenberg et al., 2008).However, reports of anammox activity in marine samples are generally limited to environmental temperatures of 25 • C or less.The compilation of the marine anammox activity measurements in Schmid et al. (2007) includes oxygen minimum zone water columns, temperate sediments, and polar sediments with a range of in situ temperatures between −1.5 and 17 • C, with one observation at 20 • C. Anammox activity has been detected in temperate estuarine sediments with in situ temperatures ranging from 15-19 • C (Dale et al., 2009; Teixeira  et al., 2012), and rates were depressed or undetectable in summer months when temperatures were above 20 • C. The presence of anammox has been confirmed in subtropical and tropical mangrove sediments (Fernandes et al., 2012;Meyer et al., 2005); however, anammox communities in brackish mangrove sediments may be more metabolically diverse than those of marine sediments exposed to higher salinities (Amano et al., 2011).Along with the evidence from temperature block incubations in the present study, these studies suggest that marine anammox bacteria (Scalindua) are predominately psychrophilic to psychrotolerant.However, more surveys of the activity and diversity of anammox bacteria are required to thoroughly test this observation.

Implications for N removal on continental shelves
Rates of nitrogen removal from sands under simulated pore water advection in the present study corroborate previous re-sults that have shown permeable shelf sediments to be dynamic sites of N removal.Furthermore, shelf environments with consistently high concentrations of nitrate in bottom waters generally have higher rates of N removal, which may be related to microbial community structure.To our knowledge, we provide the first report of denitrification and anammox activity from permeable sediments of a polar or permanently cold ecosystem.The contribution of Arctic permeable shelf sediments to global N removal is likely significant, given the vast continental shelf of the Arctic Ocean.
Microbial community adaptation to in situ temperatures has been previously demonstrated in fine-grained marine sediments for aerobic respiration, hydrolysis, nitrate respiration, and sulfate reduction (Rysgaard et al., 2004;Arnosti et al., 1998;Thamdrup et al., 1998).In permeable continental shelf sediments, we observed a strong temperature adaptation of microbial communities that mediate denitrification to in situ temperature, and marine anammox bacteria appear to thrive in sediments exposed to colder conditions.Thus, other environmental parameters are likely to outweigh temperature as the dominant control on sediment N loss rates from permeable sediments.

a
Grain size and permeability from Chipman et al. (2012), b sediment characteristics from Hedtkamp (2005), c permeability estimated for Ymerbukta site according to the Carman-Kozeny equation.

Fig. 3 .
Fig. 3. Potential denitrification rates measured in temperature gradient block experiments for (A) Ymerbukta, (B) Sylt (March 2007), (C) Sylt (June 2007), (D) SGI Gulf, and (E) SGI Bay.The seasonal temperature range is indicated by gray rectangles (note that the seasonal minimum for panel A is −20 • C).Error bars represent the standard error of the regression slope calculated at each temperature.Note different y axis scaling.

Fig. 4 .
Fig. 4. Potential anammox rates measured in temperature gradient block experiments for (A) Ymerbukta and (B) Sylt (June 2007).Rates were measured with a 15 NH + 4 label.Estimates of standard error were not made because rates were calculated from two time points.Note different y axis scaling.

Fig. 5 .
Fig. 5. Left: normalized temperature response of denitrification for subtropical, temperate, and Arctic sites.Right: normalized temperature response of anammox for temperate and Arctic sites.Curves were visually fit and error bars are omitted for clarity.

Table 2 .
Thamdrup and Dalsgaard (2002)4) rate measurements of denitrification and anammox.The production of unlabeled N 2 (p14) was calculated according toRisgaard-Petersen et al. (2004), and the contribution of anammox (% Anammox) was determined in anaerobic slurries according toThamdrup and Dalsgaard (2002).Error estimates for p14 were determined from the standard error of the regression slope.

Table 3 .
Optimum temperatures (T opt ), apparent activation energy (E a ), and ratio of rates at 5 • C relative to T opt for anammox and denitrification in permeable sediments.The range of the T opt peak where rates did not fall below 90 % of the optimum rate is noted in parentheses.Values of Q 10 were calculated between T opt and T opt −10 • C.

www.biogeosciences.net/11/309/2014/ Biogeosciences, 11, 309-320, 2014 of
Multiple lines of evidence in this study indicate that the anammox process is well-adapted to cold conditions.At both the polar and temperate sites, low T opt and E a values anammox relative to denitrification and the ratio of the anammox rate at 5 • C to T opt (35 %) all support cold adaptation.Anammox activity was also absent from the subtropical sites.A T opt of 9 • C at the polar site is consistent with the optima of 12-15 • C previously observed in Arctic fjord and deep North Sea muds, with in situ temperatures of −1.3 and 6 • C, respectively