N2O production by mussels: Quantifying rates and pathways in current and future climate settings

Blue mussels (Mytilus edulis) are an abundant and economically important species across the North Sea. Partly because of their potent filter feeding and associated shell biofilm, they are able to influence and alter the surrounding marine ecosystem. As a result of proliferating offshore wind farms (OWFs), whose turbine foundations are rapidly colonised by suspension feeding artificial hard substrate communities dominated by M. edulis, as well as planned co-location strategies of these OWFs with mussel mariculture, their numbers will only increase towards the future. On top of these local stressors, global climate change is exerting additional pressure on the marine environment. This study focusses on the link between M. edulis, its microbial shell biofilm and the local nitrogen cycling by quantifying the magnitude and underlying pathways of mussel-associated nitrous oxide (N2O) production. A set of closed-core incubations established nitrifier denitrification as the main chemical pathway of M. edulis related N2O production, although ammonium, nitrite and nitrate all acted as possible precursors. Additional future-climate experiments revealed that blue mussel’s total N2O production, as well as its metabolic activity and the relative contribution of its shell biofilm, were affected by warming (+ 3°C), acidification (- 0.3 pH units), or the combination of both. Because the effects of temperature and acidity were often of an antagonistic nature, the results suggest a relatively small net effect on local N2O production in future-climate marine environments. However, N2O production rates were several orders of magnitude lower than other measured N species ( NH 4 + , NO 2 − and NO 3 - ), making substantial mussel-associated N2 production likely. This would greatly affect the local eutrophication levels or even bioavailable nitrogen concentrations.


Introduction
After carbon dioxide (CO 2 ) and methane (CH 4 ), nitrous oxide (N 2 O) is an impactful long-lived greenhouse gasses (GHGs), substantially contributing to global warming (estimated 6 %) and to the depletion of the stratospheric ozone layer (IPCC, 2021). It has a long atmospheric lifetime (average of 114 years) and its warming effect on the global climate through radiative forcing (Global Warming Potential of 273 for a 100-year timescale) therefore persists for over a century after its emission (Smith and Sharp, 2012;Myhre et al., 2013;IPCC, 2021). N 2 O is produced by (de) nitrifying bacteria and archaea in both terrestrial and aquatic environments, mainly associated with agricultural practices and invertebrate presence or activities (Mosier et al., 1998;Stein and Yung, 2003;Heisterkamp et al., 2010). The biogenic emission of N 2 O by freshwater invertebrates is correlated with their feeding type, due to the denitrifying activity of ingested bacteria in the anoxic digestive system (Stief et al., 2009). In the marine environment, N 2 O is emitted by denitrifying gut bacteria as well, but also by microbial biofilms on the surfaces of e.g. bivalves and gastropods (Heisterkamp et al., 2010;Heisterkamp et al., 2013). These shell biofilms can generate N 2 O through three main pathways ( Figure 1): (1) as a by-product of ammonium oxidation, the first step in (oxic) nitrification by ammonium-oxidising bacteria (AOB) and archaea (AOA), (2) as an intermediate in the reduction of nitrite during AOB-and AOAmediated suboxic nitrifier denitrification and (3) as an intermediate in the reduction of nitrate during anoxic denitrification by heterotrophic bacteria. The latter comprises the final steps in the pathway of coupled nitrification-denitrification (CND), in which (heterotrophic) denitrification is carried out by a different set of actors compared to (autotrophic) nitrifier denitrification and nitrification ( Figure 1).
An important shell-bearing organism in coastal ecosystems is the blue mussel Mytilus edulis. It has an important ecological relevance through its high filtration capacity and 'bio-engineering' role in habitat creation and food provision (Cranford, 2019;Degraer et al., 2020). In the North Sea, it is one of the main colonising species on artificial hard substrate [AHS] (De Mesel et al., 2015), readily available due to the proliferating European offshore wind farm [OWF] industry (IEA, 2019;GWEC, 2021). Partly due to the blue mussel's presence, these AHSs act as an artificial reef and as a biofilter, redirecting organic material from the water column towards the benthic environment (Slavik et al., 2019;Ivanov et al., 2021). Furthermore, this species has a high economical value because of its prominent role in aquaculture (Costello et al., 2020), a practice that might progressively be combined with OWF operations (Schupp et al., 2019;MSP, 2020) as a favourable way of minimising the impacted seafloor footprint (Buck and Langan, 2017;Steins et al., 2021).
All three of the N 2 O-producing pathways ( Figure 1) could potentially occur in and on the blue mussel. Firstly, M. edulis accommodates a selection of ingested denitrifying bacteria in its anoxic digestive system, where organic carbon and different Nspecies are readily available (Stief et al., 2009). Additionally, it has a microbial biofilm on the outer surface of its shell, within which a variable oxygenation allows different pathways of N 2 O emission (Heisterkamp et al., 2010;Heisterkamp et al., 2013). Furthermore, these biofilm emissions can be sustained through the pumping activity of the mussel replenishing its surroundings with nutrientenriched waters (Heisterkamp et al., 2013). This could potentially decouple the shell biofilm's N 2 O production from (outside) environmental nitrogen cycling and fluctuations, establishing a relatively stable capacity for N 2 O emission. In an OWF or aquaculture environment, where mussels are typically present in high densities, this could potentially generate high local N 2 O emissions (Voet et al., manuscript in preparation) The abundance of mussels could well rise towards the future due to their potential in offshore and integrated multi-trophic aquaculture projects (Avdelas et al., 2021). When investigating the production of a potent greenhouse gas by such a marine species, the link with marine climate change and its global effects on the oceanic environment needs to be taken into account. The IPCC predicts a sea surface temperature rise of ca. 3°C and a drop in oceanic pH of ca. 0.3 by the end of this century (Bindoff et al., 2019;Fox-Kemper et al., 2021;IPCC, 2021), changing the biochemistry in marine environments and potentially affecting the N 2 O emissions by mussel aggregations. The objective of this study was therefore to investigate the production of N 2 O by M. edulis and its associated shell biofilm in current conditions and to describe the potential effects of seawater temperature and pH on both. This was done by experimentally (1) quantifying the relative contribution of the microbial shell biofilm to M. edulis' entire N 2 O production and (2) unravelling the different chemical pathways and precursors of M. edulis' N 2 O production. Experiments were performed in a fully-crossed design of temperature and pH, representing current and future climate conditions.

Organism collection and incubation set-up
In summer 2018 and 2020, a respective total of 800 and 240 adult blue mussels (Mytilus edulis) with a mean soft tissue dry weight (± SD) of 0.79 ± 0.28 g were sampled from a M. edulis longline in an offshore aquaculture pilot project approximately 10 km off the Belgian coast (51°11.02'N -02°39.88'E). All samples were stored in aerated seawater and transported to the experimental facilities within 4 hours.
On both occasions, organisms were evenly and randomly distributed across four identical aquaria (100 × 5 × 0 cm). These were equipped with a continuous flow-through mechanism, allowing the homogenisation of approximately 400 L natural seawater in circulation per system. All aquaria were continuously aerated and pre-set at laboratory conditions mimicking the in situ seawater salinity, temperature and pH at the time of sampling (34 PSU, 20°C and pH 7.96;LifeWatch Belgium, 2015). After an initial acclimatisation period of minimum 48 h under ambient conditions, both seawater temperature and pH were manipulated individually across the aquaria. This resulted in a 2 × 2 factorial design with four experimental treatments: a control treatment [CTRL: in situ temperature and pH], an ocean acidification treatment [OA: in situ temperature and lowered pH], an ocean warming treatment [OW: elevated temperature and in situ pH] and a climate change treatment [CC: combined elevated temperature and lowered pH].
For each treatment, stepwise manipulations were imposed over the course of three days, with a daily seawater temperature increase of 1°C and/or pH decrease of 0.1 pH unit. Seawater temperature was regulated using TECO TK2000 heaters and pH was manipulated through the controlled bubbling of 100 % CO 2 in the OA and CC treatment tanks using the IKS AquaStar aquaristic computer system, simultaneously logging temperature and pH records every 15 minutes throughout the six-week experiments (Table 1; Appendix A-Appendix B). This resulted in environmental conditions of +3°C and/or -0.3 pH units in the corresponding treatments compared to the control settings, reflecting the IPCC RCP-SSP8.5 projections for ocean warming and acidification towards the end of this century (Hoegh-Guldberg et al., 2014;IPCC, 2021). These conditions were maintained for six weeks and organisms were fed three times a week by adding 5 mL Shellfish Diet 1800 ® (Instant Algae ® mix by Reed Mariculture Inc.) to each aquarium. Glass pH electrodes were calibrated weekly using Hanna Instruments ™ NIST Reference Buffer Solutions (4.01 and 7.01) and each aquarium was sampled weekly to determine Total Alkalinity (TA) using a CONTROS HydroFIA ™ TA alkalinity system. One third of each flow-through system's water was renewed weekly. The carbonate chemistry of the seawater was calculated using CO 2 SYS software (Pierrot et al., 2006) with the thermodynamic constants of Mehrbach et al. (1973).

Experimental procedures
2.2.1 Relative contribution of shell biofilm to N 2 O production Weekly, triplicated individual closed-core incubations were set up in each experimental treatment (CTRL, OA, OW and CC) to measure N 2 O production by the mussel itself and its shell biofilm (see Appendix C: Experiment 1). Whole mussels [WHOLE] and dissected mussel shells [SHELL] were incubated separately, along with a control incubation without mussels or shells [EMPTY] to correct for background N 2 O presence. An additional control consisted of whole mussels in a 1 % ZnCl 2 seawater solution inhibiting all biological activity [ZNCL] (Heisterkamp et al., 2013). Each incubation core of 1.5 L held one individual or dissected shell of one individual (except EMPTY), along with manipulated seawater from the respective treatment, and was kept at the correct temperature throughout 3h the incubations. A magnetic stirrer ensured an evenly mixed water column in the cores. Discrete 30 mL water samples were taken at the start and end of the closed-core incubations and stored at room temperature in dark conditions. Each sample filled, to overflowing, a 30 mL airtight serum bottle and was fixed with 100 μL saturated HgCl 2 solution for N 2 O analysis. Quantification of dissolved N 2 O was done by gas chromatography (SRI Instruments, ECD) and the N 2 O production rates (PR) of whole organisms and dissected shells were calculated according to Equation 1: with V the volume (L) of the incubation core corrected for the biovolume of the whole organism or shell and with C 0 and C 1 the concentrations of N 2 O (nmol L -1 ) at the start and finish of the closed-core incubation, t 0 and t 1 (h), respectively. The relative contribution of the shell biofilm to whole animal N 2 O production was calculated by dividing each SHELL PR by the corresponding WHOLE PR in each week and treatment.
2.2.2 Pathways of N 2 O production: Nutrient flux and inhibition by NaClO 3 After three and six weeks of manipulations [WK3 and WK6, respectively], triplicated closed-core incubations with individual whole organisms were set up and sampled as described above (see Appendix C: Experiment 2). Along with biomass-specific N 2 O production rates [PRB] (calculated according to Equation 2), NH + 4 , NO À 2 and NO À 3 concentrations were measured through discrete sampling (25 mL; automated colorimetric analyses performed on SKALAR SAN++ CFA) at the start and finish of closed-core incubations, with and without the addition of 20 mM NaClO 3 to inhibit CND (Belser and Mays, 1980; see Figure 1). NaClO 3 inhibits the oxidation of nitrite (NO À 2 ) to nitrate (NO À 3 ) in the nitrification pathway (Figure 1), meaning N 2 O is produced either though nitrification (as a by-product during ammonium oxidation) or through nitrifier denitrification (as result of denitrification of nitrite; Tallec et al., 2008), as well as through denitrification of nitrate already present in the incubation core ( Figure 1).
Nutrient flux was calculated according to Equation 3; with parameters identical as in Equation 1, C the concentration of N 2 O (nmol L -1 ) and nutrients NH + 4 , NO À 2 or NO À 3 (μmol L -1 ) in Equation 2 and Equation 3, respectively, and DW the dry weight (g) of the mussel's soft tissue.  Table 2).  Before manipulations took place [WK0] and at WK3 and WK6, triplicated closed-core incubations for all three N-tracers were set up in all four experimental treatments (CTRL, OA, OW and CC; see Appendix C: Experiment 3). The 125mL incubation cores held one individual in seawater from its respective experimental treatment amended with the respective N-tracer. Cores were incubated for 4 hours on a shaking table to ensure adequate mixing. Hourly, 1mL samples were taken with a surgical syringe and transferred to 12mL He-flushed exetainers. These were prefilled with 100μL saturated HgCl 2 and 20 nmol natural N 2 O with known isotopic composition. The HgCl 2 was added to fixate the N 2 O, while the addition of unlabelled N 2 O ensured the detection limit of the isotope ratio mass spectrometer was reached. Since only 4 % of the core volume was sampled by the end of the incubation, under-pressure was not considered to be an issue, and the withdrawn volume was not replaced.

Shell biofilm oxygen microprofile
As a proxy for metabolic activity of shell biofilms, a vertical oxygen profile of the microbial biofilm on three replicate M. edulis shells was measured before seawater manipulations started [WK0] and in each experimental treatment after three and six weeks of manipulations [WK3 and WK6, respectively]. Individuals were carefully dissected with a scalpel to isolate the mussel shell without damaging the associated biofilm and dissected shells were placed in aerated seawater from the respective experimental treatments. A Unisense ™ MicroProfiling System was used to position the sensor tip of a PyroScience ™ retractable fiber oxygen microsensor on the shell surface and a vertical dissolved oxygen concentration profile was recorded in increments of 50 or 100 μm through the biofilm between 0-3000 μm above the shell surface. Oxygen microprofiles were measured at three random positions on the shell of each replicate organism and conducted in dark conditions to avoid net oxygen production.

Data analysis
The effect of temperature (current or elevated) and pH (current or lowered) on the production of N 2 O in the WHOLE, SHELL, NaClO 3inhibited and non-inhibited incubations was investigated using linear mixed effects models, with incubation core and/or experimental week added as a random factor in the models. If none of the random factors could explain left-over variance, a linear regression model was fitted. The effect of temperature, pH and experimental week (WK3 or WK6) on nutrient flux was analysed using linear regression models. Significance of the two-way interaction 'temperature x pH' and post-hoc pairwise comparison of the group means was used to identify possible combination effects of increased temperature and lowered pH. Normality of the residuals and model assumptions were checked using Shapiro-Wilk, Gaussian error distributions were used and model selection was based on the parametric bootstrap and Kenward Roger methods for mixed model comparison (Halekoh and Højsgaard, 2014) or on the backwards selection procedure for linear regression models. Statistical analyses were conducted using R v3.6.1 (R Core Team, 2019), linear mixed effects models were built using the R package lme4 (Bates et al., 2015) and linear regression models were fitted using the stats package (R Core Team, 2019).
To test for differences between oxygen microprofiles between experimental treatments in WK3 and WK6 (to keep a balanced design), a multivariate matrix was constructed with the oxygen concentrations at each distance above the shell (Widdicombe et al., 2013). Differences in the shell biofilm oxygen microprofiles were tested using permutational ANOVA (PERMANOVA) with experimental treatment and week as fully-crossed factors and with all profiles on all shells considered as replicates within the Euclidian distance matrix. For all significant PERMANOVA factors, post-hoc pairwise testing was used and homogeneity of multivariate dispersion was tested using PERMDISP. The SIMPER routine was used to identify which distances from the shell contributed most to any observed differences. Multivariate analyses were carried out in Primer v6.0 with PERMANOVA+ add-on software (Clarke and Gorley, 2006;Anderson et al., 2008).

Results
Mean seawater temperature (± SD) in CTRL treatments was 19.97 ± 0.20°C and ranged between 23.07 ± 0.36°C and 23.15 ± 0.24°C in the warmed treatments (OW and CC). Mean seawater pH (± SD) in acidified treatments (OA and CC) ranged between 7.63 ± 0.05 and 7.65 ± 0.02, compared to an average CTRL seawater pH of 7.96 ± 0.02 (Table 1; Appendix A -Appendix B). Seawater salinity differed slightly between the treatments but was well within the natural occurring salinity range. The total alkalinity (TA) fluctuated according to the pH levels (1985-2123 μmol kg -1 ) and the average aragonite/calcite seawater saturation states (W; ± SD) ranged between 1.70 ± 0.25 -2.97 ± 0.52 in the non-acidified treatments and between 0.82 ± 0.16 -1.51 ± 0.24 in the acidified treatments (Table 3).
For an overview of ecophysiological effects of increased temperature and lowered seawater pH on M. edulis, see Voet et al. (2021). In this study, the same experimental treatments (CTRL, OW, OA and CC) were imposed for six weeks. Significant results included a decreased survival rate and growth in acidified treatments, enhanced growth rates in warmed treatments and an additive effect of higher temperatures and lower pH levels on respiration and clearance rates (Voet et al., 2021).

Relative contribution of shell biofilm to N 2 O production
In all treatments, N 2 O was produced in both the WHOLE and SHELL incubations, whereas no N 2 O production was measured in the poisoned ZNCL incubations. In CTRL, whole animals produced on average (± SD) 4.55 ± 0.18 nmol N 2 O ind -1 h -1 and throughout the experiment, the shell biofilm contribution ranged between 65-75% ( Figure 2). In OW, the average WHOLE production rate gradually increased from 5.47 ± 0.23 to 15.84 ± 0.48 nmol N 2 O ind -1 h -1 over six weeks, while the shell biofilm contribution remained stable at 68% (Figure 2). In both acidified treatments, the relative contribution of the shell biofilm to whole organism N 2 O production decreased over time. In OA, the WHOLE N 2 O production rate was stable throughout the experiment (mean 5.79 ± 0.33 nmol N 2 O ind -1 h -1 ), but the contribution of the shell biofilm decreased from 63% to 23% over the course of six weeks (Figure 2). Whole animals in CC produced on average 9.10 ± 1.82 nmol N 2 O ind -1 h -1 , while the associated shell biofilm contribution decreased from 60% to 12% over time, after an initial rise to 65% in the second week of experiments. Overall, temperature and pH had a significant, antagonistic effect on the parts of the animal that emitted N 2 O, with an increasing production in warmer environments and a decreasing production in acidified or combined climate change conditions (Table 4).  ; partial pressure of CO 2 (pCO 2 ; μatm); total inorganic carbon concentration (C T ; μmol kg -1 ) and saturation state of the seawater with respect to aragonite (W A ) and calcite (W C ).
3.2 Pathways of N 2 O production 3.2.1 Nutrient flux and inhibition of coupled nitrification-denitrification When the CND pathway was inhibited by the addition of NaClO 3 , average N 2 O production rates were 4 to 16 times higher compared to incubations where the inhibitor was absent (Figure 3). Regardless of the presence of the inhibitor, temperature and pH had an antagonistic effect on the production of N 2 O by M. edulis and the highest average N 2 O production rates were measured in OW ( Figure 3 and Table 5).
Inhibition by NaClO 3 significantly lowered ammonium flux, which additionally decreased significantly in WK6 compared to WK3 (Figure 3 and Table 6). Lowered pH and elevated temperature antagonistically affected ammonium flux (production or oxidation rates; Table 6), with the lowest average NH + 4 flux in OW and a significantly higher flux when combined with lowered pH in CC (Figure 3).
There was no significant effect of the inhibitor NaClO 3 on nitrite flux and in general, nitrite flux decreased with time ( Figure 3 and Table 6). Elevated temperature and lowered pH significantly increased NO À 2 flux (Table 6). Nitrate flux was negative across all experimental treatments and not affected by the presence of the inhibitor (Figure 3 and Table 6). Nitrate consumption (negative flux) significantly increased with time and the significant decreasing effect of a lowered pH was further amplified when combined with an elevated temperature in CC ( , NO À 2 and NO À 3 all serve as potential precursors and nitrification, nitrifier denitrification and coupled nitrification-denitrification (respectively) all serve as potential pathways of N 2 O production by M. edulis and its shell biofilm (Table 7 and Figure 4).
In the NIT incubations, targeting the production of N 2 O through nitrification (precursor NH + 4 ), no or very low amounts of 46 N 2 O (0.00 -0.09 nmol g -1 DW h -1 ) were produced across all experimental treatments and weeks (Table 7 and Figure 4). The contribution of nitrification to the total 15 N-labelled N 2 O production ( 46 N 2 O + 45 N 2 O) by M. edulis and its shell biofilm in this series of experiments was very low, ranging between 1 % in CTRL to 3 % in OA. The production rate of 45 N 2 O, on the other hand, was ± 10 to 500 times higher (0.03 -2.95 nmol g -1 DW h -1 ), started after an initial lag phase of 3 hours (Figure 4 and Table 7) and was the result of either random isotope paring, nitrifier denitrification or CND ( Table 2).
The DNO2 incubations, targeting nitrifier denitrification (precursor NO À 2 ), showed the highest average 46 N 2 O production rates of all 15 N-tracer treatments, with highest average values (2.08 ± 0.96 nmol g -1 DW h -1 ) recorded in CC after 6 weeks (Table 7). Nitrifier denitrification contributed, on average, 46 % to the total labelled N 2 O production of whole animals in CTRL and up to 59 % the OA treatment. In general, the 45 N 2 O production rates were ± 1.2 to 3 times lower compared to 46 N 2 O and produced by the combination of 15 NO À 2 with 14 NO À 3 or 14 NO À 2 present or produced during the incubation (Table 7 and Figure 4).
Average 46 N 2 O production rates in DNO3 incubations, targeting the CND pathway (precursor NO À 3 ), was relatively high at the start of the experiment (WK0; 0.12 ± 0.18 nmol g -1 DW h -1 ) and contributed approximately 40 % of the total labelled N 2 O production in that week. In subsequent weeks, 46 N 2 O production rates decreased 2-to 42-fold and all manipulated treatments had higher average 46 N 2 O production rates compared to CTRL (Table 7 and Figure 4). On average, CND contributed between 4 % in CRTL up to 15 % in OW in the subsequent weeks. Average 45 N 2 O production rates were 2 to 46 times higher across all experimental treatments compared to 46 N 2 O and occurred in the DNO3 tracer treatment as a result of the random pairing of 15 NO À 3 with 14 NO À 3 or 14 NO À 2 that were present or produced during the incubation (Tables 2, 7).

Oxygen microprofile
The biofilm oxygen microprofiles differed significantly between experimental treatments (PERMANOVA; pseudo-F=93.98; p=0.001), while the effect of week or the 2-way interaction of both was not significant (PERMANOVA; pseudo-F=0.92 and 1.23, respectively; both p>0.29). Pairwise testing revealed significant differences in oxygen microprofiles between all experimental treatments (p=0.001), except for the profiles measured in OA and CC (p=0.24). SIMPER revealed that at least 90% of the differences between CTRL and the other experimental treatments could be attributed to the first 600μm above the shell surface. In OW, oxygen concentrations above the shell declined faster and stronger, while in both acidified treatments (OA and CC), the onset of the decline in oxygenation was closer to the shell surface (biofilm thickness declined) and the magnitude of the decline was lower (net biofilm oxygen consumption decreased) ( Figure 5).

Discussion
This study confirmed the contribution by the abundant and economically important blue mussel Mytilus edulis to marine N 2 O production (Stief et al., 2009;Heisterkamp et al., 2010;Bonaglia et al., 2017), as well as the relative importance of shell biofilms in this process (Heisterkamp et al., 2013). Our results on the magnitude and underlying processes of this N 2 O production show that nitrification, nitrifier denitrification and coupled nitrification-denitrification all served as potential pathways in N 2 O production. In contrast to earlier research by Heisterkamp et al. (2013) on dissected shell biofilms, nitrification of NH + 4 was not an important contributor to the total production by both N 2 O production rates (nmol g -1 DW h -1 ± SE) and nutrient flux (µmol g -1 DW h -1 ± SE) of whole animals incubated with and without the addition of NaClO 3 , inhibiting the oxidation of nitrite to nitrate, in four experimental treatments [CTRL, control; OW, ocean warming; OA, ocean acidification and CC,: climate change]. Note different scale and unit in left panels. Significance is indicated in bold.
the mussels and its shell biofilm. Additionally, we suggest that future climate change will not substantially affect total N 2 O production by blue mussels as the increase caused by higher temperatures is counteracted by a decreased shell biofilm activity due to ocean acidification and overall, N 2 O production declined over time in the climate change environment.

Pathways and contributions to N 2 O production
The current climate [CTRL] N 2 O production of blue mussels (including their associated shell biofilm) was in line with other studies (e.g. Heisterkamp et al., 2010;Heisterkamp et al., 2013;Gaŕate et al., 2019) and exclusively biological, as ZnCl 2 poisoning fully inhibited N 2 O production. The shell biofilm contribution to total N 2 O production by M. edulis was important (± 70 %) and remained stable through time, as did the total N 2 O production rate. Although still a considerable contribution, the partitioning measured in this study was lower than previously reported by Heisterkamp et al. (2013), where N 2 O production originated almost exclusively from the shell biofilm (± 94 %). Possibly, methodological differences such as the use of natural seawater and a regular feeding regime in our experiments could have prompted the adept filter feeder M. edulis, with an established rich intestinal microbial system capable of considerable N 2 O production, to continue to do so throughout the duration of this experiment.
Nitrifiers, a combination of ammonium-oxidising bacteria and archaea, produce N 2 O during nitrification in oxic environments and during nitrifier denitrification in suboxic environments. Another set of  bacteria, denitrifiers, produce N 2 O during anaerobic denitrification in anoxic environments, comprising the final steps of CND. These three pathways are considered to be the main oceanic sources of N 2 O production, while the main sink is the further reduction of N 2 O to N 2 by (nitrifier) denitrification in sub-or anoxic environments (Bange et al., 2010). In this study, each of these three N 2 O producing pathways were observed. Contrary to Heisterkamp et al. (2013), where only dissected shells were used in a similar experiment, nitrification apparently did not play as an important role when whole animals (incl. shell biofilm) were incubated. Most likely, the presence of the living and digesting animal itself caused the resulting total N 2 O production to lean towards a more prominent inclusion of anoxic or suboxic pathways. Denitrification of nitrite, targeted by the DNO2 precursor incubations, was the most dominant pathway of N 2 O production by M. edulis in the current climate scenario, corroborating the findings of Heisterkamp et al. (2013). Production of N 2 O in the presence of NaClO 3 , actively preventing the oxidation of nitrite to nitrate, was on average an order of magnitude higher than without the inhibitor present. As such, we suggest that the availability of nitrite might be a limiting factor in the production of N 2 O and that incomplete (nitrifier) denitrification of nitrite, where further reduction to N 2 has not (yet) happened, is indeed a major pathway of N 2 O production by M. edulis and its biofilm. An overestimation of said denitrification due to anaerobic ammonium oxidation (anammox), where ammonium oxidises to N 2 using nitrite under anoxic conditions, was assumed negligible since no anammox activity has been associated with mussel biofilms before (Marzocchi et al., 2021).
Denitrification of nitrite can be associated with the presence of M. edulis in different ways, both in and on the organism. The incomplete denitrification of nitrite to N 2 O by denitrifiers in the mussel's anoxic digestive system could be the result of a high lysozyme activity, cutting short the denitrification pathway (Birkbeck and McHenery, 1982;Heisterkamp et al., 2013;Gaŕate et al., 2019). Additionally, this last step in the nitrifier denitrification pathway has only be attributed to a certain genus of nitrifying bacteria, Nitrosomonas sp., whose biofilm presence and/or (differential) activity is related to oxygen and nutrient availability (Schramm et al., 2000;Shrestha et al., 2002) and substrate affinity and availability (Foesel et al., 2008;Zhu et al., 2013). With the variable oxygenation of the shell biofilm measured in this study in mind, this could further explain why the produced N 2 O is not always readily reduced to N 2 in a heterogeneous microbial biofilm on M. edulis's outer shell. However, as the N 2 O flux was several orders of magnitude lower than that of the other measured N species (NH + 4 , NO À 2 and NO À 3 ), reduction of N 2 O to N 2 must still have occurred to close the nitrogen budget in the experimental incubations. Research by Marzocchi et al. (2021) suggests that mussel-associated N 2 production, although currently often overlooked, might be substantial. The high CTRL nitrate consumption rates observed in this study comply with this assumption.

Climate change effects on N 2 O production
By nature, adult blue mussels have a wide thermal tolerance window (e.g. Seuront et al., 2019;Kamermans and Saurel, 2022). Since this species has adapted to survive in such variable conditions, this study opted to push towards the boundaries of its natural tolerances in order to get the most realistic stress response to variable climate scenarios. In the North Sea, the highest mean temperatures occur over the summer period (when organisms were sampled) when the mean seawater pH is fairly stable (LifeWatch Belgium, 2015), meaning the adopted climate manipulations should elicit a relatively accurate physiological response outside M. edulis' natural tolerances. Results show that the effects of seawater temperature rise and ocean acidification were generally of an antagonistic nature. N 2 O production rates of whole animals, as well as the production rate of dissected shells with an intact microbial biofilm, significantly increased with a higher temperature and significantly decreased in an acidified environment (whole animals) or a combined high temperature/low pH climate change environment (biofilm).
These experiments were set up to mimic North Sea summer conditions in CTRL, with additional manipulations in terms of temperature and/or pH levels according to the experimental treatments. Although this summer setting is the most relevant in terms of studying potential stress responses due to climate change, the temperature-dependent nature of physiological and microbial processes suggest that the absolute values of these responses would differ between the seasons (Lesser et al., 2010;Bouletreau et al., 2012;Muǵica et al., 2015). However, whether the observed patterns and nature of the interaction between ocean warming and acidification varies seasonally remains to be confirmed.
The increasing effect of elevated temperature on N 2 O production rates by M. edulis and its associated shell biofilm are most likely linked to the mussel's concomitantly increasing filtration rates in combination with temperature-induced increased biofilm activity. As bivalve filter feeding increases with temperature (Kittner and Riisgård, 2005;Ong et al., 2017;Voet et al., 2021), more denitrifying microorganisms might be ingested and/or the concentration of available N precursors in the digestive system might increase. In addition, general microbial activity increases with warmer temperatures Bouletreau et al., 2012). This study showed that even though the N 2 O production by whole animals increased with temperature, the relative contribution of the microbial shell biofilm remained stable over time in the ocean warming [OW] scenario. Indeed, this would Vertical oxygen profiles of the microbial biofilm on Mytilus edulis shells, 1500µm above the shell surface, before seawater manipulations started (WK0) and after three and six weeks of manipulations (WK3 and WK6, respectively) in each experimental treatment [CTRL: control, OA: ocean acidification, OW: ocean warming and CC: climate change]. Profiles were measured on three random locations per replicate M. edulis shell (n=3) per timepoint and treatment.
suggest a similar positive effect of warmer environments on both the animal's and the shell biofilm's N 2 O producing capacities. Moreover, increased feeding rates will result in higher biodeposition rates and in turn, a higher nutrient load, potentially further stimulating (de) nitrification activities in the microbial shell biofilm.
On the other hand, the relative contribution of the shell biofilm to N 2 O production evidently decreased over time in both acidified scenarios, on top of the overall decrease in whole animal N 2 O production rates. A negative effect of ocean acidification on the marine nitrogen cycle was previously observed by Beman et al. (2011), reporting reduced oceanic nitrification at even relatively small experimental pH reductions (0.05-0.14). Together with a profound effect of reduced pH on the community composition of marine biofilms (Espinel-Velasco et al., 2021), this could potentially explain the observed lower N 2 O production rates under acidified conditions, as well as the decreasing relative contribution of the microbial biofilm. The latter is most likely also tied to a thinner, patchier or even disappearing shell biofilm, as is suggested by this study's significantly affected oxygen microprofiles in both acidified conditions. The microprofile analysis indeed confirmed an adverse effect of pH on the metabolic activity (respiration) and composition of the shell biofilm. In both acidified treatments, perceived biofilm thickness considerably decreased, as well as the net biofilm oxygen consumption. Furthermore, net oxygen consumption at the shell surface sometimes dropped to (near) zero under acidified conditions, indicating either an incomplete biofilm cover or a profound shift in functional metabolism and/or the biofilm's microbial makeup. A more in-depth functional analysis of M. edulis' shell biofilm under climate change conditions is described in more detail by Dairain et al. (manuscript in preparation) 2 . Nitrifier denitrification, shown to be the dominant N 2 O producing pathway for M. edulis (e.g. this study and Heisterkamp et al., 2013), gained even more importance in the OA treatment. Affirmingly, this could (in part) be a result of the increasing prevalence of suboxic conditions provided within a thinning and/or receding shell biofilm.

Towards ecosystem-level effects
The blue mussel is by far the most abundant mollusc species colonising North Sea artificial hard substrates (Coolen et al., 2020) and these numbers will multiply even further by planned co-location strategies involving offshore wind farms and bivalve mariculture (Schupp et al., 2019;MSP, 2020;Steins et al., 2021). Considering the prevalence of these organisms and the potential biofilm-substrate they provide are on the rise, their role in the local emission of the potent GHG nitrous oxide could be noteworthy and a valuable research topic. On the other hand, this study also suggested that reduction of this GHG to inert N 2 could be prevalent in or around these mussels, as was the case in other recent mussel microbiome research (Marzocchi et al., 2021). This would suggest that colonising and aquaculture mussel communities could play a role in the removal of nitrogen from the water column, actively counteracting marine eutrophication. Contrary to climate change-induced changes in (i.a.) survival or ecophysiology of blue mussels (Voet et al., 2021), the antagonistic effects of temperature and pH on N 2 O production described in this study would most likely not amount to large-scale enhancements of future mussel-associated N 2 O emissions. With the expansion of the offshore renewable energy effort in a necessary attempt to abate GHG emissions causing global climate change, insights such as described in this study are paramount. Complementary, marine climate science would benefit greatly from further research into the link between impacts of future climate effects on current marine communities and the potential effects of adaptation and ecophysiological acclimatisation over longer time periods.

Conclusion
The blue mussel Mytilus edulis and its microbial shell biofilm produce N 2 O through three potential pathways, i.e. nitrification, denitrification (CND) and nitrifier denitrification (using ammonium, nitrate and nitrite as a precursor, respectively), but nitrifier denitrification proved to be the main pathway. The relative importance of these pathways and the relative contribution of the shell biofilm to the blue mussel's total N 2 O production, as well as the biofilm thickness, metabolic activity and coverage are all affected by warming, acidification, or the combination of both. The effects of temperature and pH were often of an antagonistic nature, predicting a relatively small net effect on local N 2 O production in future climate environments. In a future setting with increasing blue mussel densities, such as proliferating offshore wind farms in possible multifunctional co-use with bivalve mariculture, animal-associated N 2 O and N 2 production will play a major role in local nitrogen cycling and have potential knock-on effects on local greenhouse gas emissions, eutrophication levels and bioavailable nitrogen concentrations. An understanding of the underlying pathways and inner dynamics of the OWF-and aquaculture-bound N cycling will therefore be indispensable to face and mitigate the additional impact of a changing marine climate.

Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Funding
The research leading to results presented in this publication was funded through BELSPO BRAIN-be project BR/175/A1/PERSUADE, additional funding obtained from Ghent University's Special Research Fund (BOF) through GOA project 01G02617 and carried out with infrastructure funded by EMBRC Belgium FWO project GOH3817N. This work was further supported by logistic support from the Research Vessel Simon Stevin, the vessel Stream (Brevisco), the scientific diving teams of VLIZ and RBINS and by access to the ILVO aquaculture pilot project Value@Sea (with partners Belwind, Brevisco, Sioen Industries, Colruyt Group, C-Power, DEME, RBINS -OD Nature, Lobster Fish and Ghent University). Practical and analytical support by the MarBiol Research Group (UGent), VLIZ, the Chemical Oceanography Unit (COU; University of Liège) and the Isotope Bioscience Laboratory (ISOFYS; UGent) was indispensable in the realisation of this study.