Isotopic evidence for biogenic molecular hydrogen production in 1 the Atlantic Ocean 2

18 Oceans are a net source of molecular hydrogen (H 2 ) to the atmosphere. The production of 19 marine H 2 is assumed to be mainly biological by N 2 fixation, but photochemical pathways are 20 also discussed. We present measurements of mole fraction and isotopic composition of 21 dissolved and atmospheric H 2 from the southern and northern Atlantic between 2008 and 22 2010. In total almost 400 samples were taken during five cruises along a transect between 23 Punta Arenas (Chile) and Bremerhaven (Germany), as well as at the coast of Mauritania. 24 The isotopic source signatures of dissolved H 2 extracted from surface water are highly 25 deuterium-depleted and correlate negatively with temperature, showing δ D values of (- 26 629±54) ‰ for water temperatures at (27±3) °C and (-249±88) ‰ below (19±1) °C. The 27 results for warmer water masses are consistent with biological production of H 2 . This is the first marine H 2 excess has directly attributed to biological production by isotope measurements. However, the isotope values obtained in the colder water masses indicate that beside possible biological production a significant different source should be considered. The atmospheric measurements show distinct differences between both hemispheres as well 5 as between seasons. Results from the global chemistry transport model TM5 reproduce the 6 measured H 2 mole fractions and isotopic composition well. The climatological global oceanic 7 emissions from the GEMS database are in line with our data and previously published flux 8 calculations. The good agreement between measurements and model results demonstrates that 9 both the magnitude and the isotopic signature of the main components of the marine H 2 cycle 10 are in general adequately represented in current atmospheric models despite a proposed 11 source different from biological production or a substantial underestimation of nitrogen 12 fixation by several authors.

first time that marine H 2 excess has been directly attributed to biological production by 1 isotope measurements. However, the isotope values obtained in the colder water masses 2 indicate that beside possible biological production a significant different source should be 3 considered. 4 The atmospheric measurements show distinct differences between both hemispheres as well 5 as between seasons. Results from the global chemistry transport model TM5 reproduce the 6 measured H 2 mole fractions and isotopic composition well. The climatological global oceanic 7 emissions from the GEMS database are in line with our data and previously published flux 8 calculations. The good agreement between measurements and model results demonstrates that 9 both the magnitude and the isotopic signature of the main components of the marine H 2 cycle 10 are in general adequately represented in current atmospheric models despite a proposed 11 source different from biological production or a substantial underestimation of nitrogen 12 fixation by several authors. Oceanic H 2 production is assumed to be mainly biological, as a by-product of nitrogen (N 2 ) 28 fixation (e.g. Conrad, 1988;Conrad and Seiler, 1988;Moore et al. 2009Moore et al. , 2014. H 2 is 29 produced during N 2 fixation in equimolar proportions, but also reused as an energy source. 30 The H 2 net production rate during N 2 fixation depends on environmental conditions and also N 2 fixation, abiotic photochemical production from chromophoric dissolved organic matter 1 (CDOM) and small organic compounds such as acetaldehyde or syringic acid has also been 2 found to be a source of hydrogen in the oceans (Punshon and Moore, 2008a, and references 3 therein). 4 Unfortunately, measurements that constrain the temporal and spatial patterns of oceanic H 2 5 emissions to the atmosphere are sparse. Vertical profiles display highest concentrations in the 6 surface layer (up to 3 nmol L -1 ) and a sharp decrease with depth towards undersaturation, 7 where the reasons for the undersaturation are not fully understood yet (e.g. Herr et al. May 2010, air and seawater samples were collected (see Fig. 1, Table 1). The cruises of RV 9 Polarstern were part of the OCEANET project ( for at least 12 hours; the N 2 remained in the flask at ambient pressure until the sampling. 30 During sampling the flasks were flushed for 4 minutes with ambient air at a flow rate of 12 L 31 min -1 using Teflon tubes and a membrane pump (KNF VERDER PM22874-86 N86ANDC). 32 The sample air was dried with Drierite® (CaSO 4 ). The flasks were finally pressurized to 1 approximately 1.7 bar, which allows duplicate measurements of the H 2 isotopic composition 2 of an air sample. 3 Table 1 gives an overview of the sampling scheme for discrete H 2 samples. In total 360 4 samples were collected, regularly distributed over the transects at 4 to 6 hour intervals. In 5 2009 the resolution of sampling was enhanced to one sample per two hours and focused on 6 five sub-sections of the transect, in an attempt to resolve dial variability. 7 Samples were always taken at the downwind side of the ship to exclude a possible 8 contamination by ship diesel exhaust. One atmospheric sample was taken directly inside the 9 ship's funnel of RV Polarstern to determine the mole fraction and δD of ship diesel exhaust as 10 a possible contamination source. This first measurements for ship diesel exhaust gave an H 2 11 mole fraction of (930.6±3.2) nmol mol -1 and a δD of (-228.6±5.0) ‰. In the following, we 12 will use the abbreviation "ppb" = 10 -9 in place of the SI unit "nmol mol -1 ". 13 isotopic composition measurements of molecular H 2 . It consists of a glass vessel (10 L) and 20 an evacuation / headspace sampling unit. 21 The glass vessel was evacuated for at least 24 h before sampling, using a Pfeiffer vacuum 22 DUO 2.5A pump, with a capacity of 40 L min -1 (STP: 20° C and 1 bar). Water samples were 23 taken from 5 m depth (RV Polarstern cruises) or 10 m depth (RV L'Atalante cruise) using a 24 24-Niskin-bottle rosette with a volume of 12 L each. Sampling started immediately after 25 return of the bottle rosette on-board and from a bottle dedicated to the H 2 measurements. The 26 evacuated glass vessel was connected to the Niskin bottle by Teflon tubing, which was first 27 rinsed with approximately 1 L surface water. Then, 8.4 L water streamed into the evacuated 28 flask (Fig. 2), using a drip to enhance the dispersion of the sample water. After connection of 29 the headspace-sampling unit, the lines were first evacuated and then flushed with a makeup 30 gas several times. During the RV L'Atalante cruise a synthetic air mixture with an H 2 mixing 31 ratio below threshold was used as makeup gas. The makeup gas used during the RV 32

Atmospheric H 2 and δD (H 2 ) in discrete samples 24
The mole fraction and isotopic composition of molecular H 2 was determined using the  mole fractions of the air in these cylinders were determined by the Max Planck Institute for 6 the air into the RGA was not necessary and the flasks were simply connected to the RGA inlet 23 by Teflon tubing. The remaining pressure was mostly sufficient to perform 8 to 10 24 measurements. A slight memory effect was observed and thus only the last 5 measurements 25 were taken into account when stable. Samples with only three or less valid measurements 26 were not used for evaluation. The standards were the same as those used for the isotope 27 system. For both cruises (ANT-XXV/5 and ANT-XXVI/1), the mean measurement 28 repeatability was better than ±0.8 % (H 2 ) and ±2 % (CO). A comparison between the H 2 mole 29 fractions measured with the Peak Performer 1 RGA and the isotopic experimental setup 30 reveals on average slightly lower RGA values of (7.5±23.8) ppb (see Fig. 3). 31

Atmospheric H 2 measured continuously 1
For the on-board continuous measurements of H 2 mole fractions a Peak Performer 1 RGA 2 was used. The atmospheric air was drawn from the bridge deck to the laboratory in ¼ inch 3 Decabon tubing. The CO mole fraction was also measured in the same measurement and will 4 be reported here for information, but without further discussion. 5 In alternating order, 10 air samples and 10 aliquots of reference air were measured, using 6 synthetic air as carrier gas. Due to small memory effects, only the last 5 measurements of 7 each were taken into account when the values were stable. The mole fractions of H 2 and CO 8 were calculated by using the mean of the enclosing standard measurements, with an estimated 9 maximal error of ±5 %. For more details see Popa et al. 2014. The mean measurement 10 repeatability for the air samples was ±1.7 % for H 2 and ±3.6 % for CO in ambient air, 11 respectively ±0.8 % (H 2 ) and ±0.9 % (CO) for the reference air. Comparing the H 2 mole 12 fractions measured continuously on the RGA with discrete samples measured on the isotope 13 system and collected close in time, we found a mean offset of (-18.8±16.4) ppb for the RGA 14 results. 15 Defining the extraction efficiency η as 22 where V h and V w are the volume of the headspace and the water fraction, and c h the 24 concentration of H 2 in the headspace. The initial concentration of H 2 in seawater, c w0 , can be 25 calculated from 26 The concentration in the headspace, c h , was not measured directly, but can be derived from 28 the measured H 2 mole fraction in the sampling flask. The sampling procedure following gas 1. Expansion of the headspace into the gas transfer system 1 2. Addition of makeup gas 2 3. Expansion of the headspace / makeup gas mixture into a sample flask 3 As shown in the Appendix, the original concentration of H 2 in seawater (in nmol L -1 ) can be 4 calculated using the following equation 5 The extraction efficiency, η can be calculated from the following mass balance 11 Assuming that headspace gas phase and water phase are in equilibrium, the ratio of the H 2 13 concentration in water and in the headspace is given by the Ostwald coefficient (where the 14 concentrations refer to in situ temperature): 15 This gives for the extraction efficiency as defined in equation (2) 17 In the present case, α = α(H 2 ) was equal to 0.0163±0.0001, which gives η = 92.12 (±0.013)% 19 for V w /V h = 8.4/1.6 = 5.25. 20 Two alternative scenarios were considered to derive the δD of the dissolved H 2 , with scenario 21 1 assuming equilibrium isotopic fractionation between headspace and water, and scenario 2 22 assuming kinetic isotopic fractionation during extraction from Niskin bottle to glass vessel. 23 Scenario 1: The equilibrium isotope fractionation between dissolved phase and gas phase is ε = (37±1) ‰ 25 at 20 ºC [Knox et al., 1992]. 26 Scenario 2: The kinetic isotope fractionation during gas evasion is ε k = (-18±2) ‰ at 20 ºC [Knox et al., 1 1992]. The approximation is not used and only shown to illustrate the small difference 2 between δ w0 and δ h when η ≈ 1. 3 The temperature dependences of ε and ε k are unknown and were neglected here. 4 The air saturation equilibrium concentration, c sat (H 2 ), was determined using the 5 parameterization of Wiesenburg and Guinasso (1979). The H 2 saturation anomaly, Δ(H 2 ), was 6 calculated as the difference between the measured H 2 concentration, c(H 2 ), and c sat (H 2 ): 7 Δ(H 2 ) = c(H 2 ) -c sat (H 2 ) (10) 8 Meteorological and oceanographic parameters (radiation, air and water temperatures, salinity, 9 relative humidity) were measured using standard instrumentation and recorded and provided

Atmospheric H 2 transects 10
Our data set includes data of two hemispheres and two seasons between 2008 and 2010 (see 11 Table 2, Fig. 4). The mean mole fraction of H 2 ranged between (532.0±10.7) ppb and 12 (548.5±6.8) ppb. In spring, the mean values were almost equal between the hemispheres with 13 approximately 1 to 2 ppb difference, but they differed significantly in autumn. In this season, 14 the mean values in the northern hemisphere (NH) were approximately 16 ppb or 3 % lower 15 compared to the southern hemisphere (SH), with a distinct transition between the hemispheres 16 at around 8° N. In contrast, δD differed significantly between the hemispheres in both 17 seasons. In the southern hemisphere, absolute δD values were always between 9 and 27 ‰ 18 higher than in the northern hemisphere, and generally remained within a narrow range 19 between (140.5±21.1) ‰ and (145.4±5.3) ‰. In contrast to the mole fraction, isotope delta 20 differences between the hemispheres were less pronounced in autumn than in spring. These 21 two seasonal patterns, in the following defined as "summer signal" and "winter signal", are 22 mainly caused by biological processes and tropospheric photochemistry and driven by 23 during this season is lower in the NH than in the SH. Due to the general preference of 2 organisms for molecules with lighter isotopic composition, the δD values increase during 3 summer in the NH and the interhemispheric gradient becomes less pronounced. 4 The "winter signal" observed in April is defined by almost equal mole fractions and more 5 pronounced differences in δD values between the hemispheres. In winter, molecular hydrogen 6 is accumulating in the NH hemisphere, and the main source is fossil fuel combustion with a 7 The model results are less variable on small spatial scales, due to the low spatial resolution, 21 and possibly to local influences that are not included in the model (e.g. ocean emissions in the 22 model are less variable in time and space than they could be in reality). The largest 23 differences between the modeled and measured H 2 occur between 30° S and the equator. This 24 seems a systematic feature and could be due to a slight overestimation of sources or values within or only slightly outside a 2 σ range around the mean, except for the one 20 between 23.5° S to 15.7° S (Fig. 6a). Here the highest H 2 mole fractions of (631.9±3.2) ppb, 21 combined with the lowest δD values of (20.9±5.0) ‰, were found around 16° S. Due to the 22 limited spatial resolution and therefore low number of data points a Keeling plot analysis 23 ( Fig. 6b) of the data between 15° S and 18° S was made with either 5, 7, or 9 data points to 24 get a reasonable range for the source signature. It reveals a mean source signature of -561. respectively. The observed increase of δD seems reasonable when assuming oxidation by 22 HO • , but with respect to the HO • mole fraction and the slow reaction rate of H 2 + HO • it is 23 questionable whether the H 2 decrease here can be explained by this. 24

Dissolved H 2 25
A new method has been presented to extract H 2 from surface waters for isotopic 26 determination. Before discussing the measurement results, we will give an overview of the 27 possible main errors and their effects. To show the effect of the errors on the measurements, 28 we will present error factors, thus how much the final data differ by shifting the respective 29 parameter by 1 % and also the absolute assumed error.
For the extraction method several error sources could be identified: the determination of 1 pressure, especially in the sampling vessel before adding the make-up gas and during 2 extraction, the temperature of air and water, respectively the difference between them when 3 the sample is extracted from the headspace, and the volume of the set-up and the sample. The 4 determination of pressure in the sampling vessel would be one issue of further improvement, 5 because the error caused by pressure deviations for the total pressure after adding the make-6 up gas is about a factor of 0.7 for concentrations and 0.2 for the isotopic values. The error 7 based on temperature of air, water and sample is negligible due to high-precision 8 measurements and the short handling time between water sampling and headspace extraction. 9 The error for the volume parameter for the set-up is negligible due to the high volume, the 10 precise determination of the glass vessel volume by weighing, and the calculation of the 11 tubing volume. The main error source is the water volume of the sample, which counts by a 12 factor of 5.9 for the concentration, but with negligible effect on the isotopic values. Although 13 the relative error factor is quite high the absolute value is assumed to be around 0.5% due to 14 the sample size, which has also been weighed at the home lab. The H 2 measurement 15 procedure is the same as for atmospheric samples and Taking measurement and handling errors during the extraction as well as errors in the 21 determination of the dry mole fraction into account, we assume a robust overall uncertainty of 22 ± 6.9 % for the dissolved H 2 mole fractions and ± 4.7 % for the isotopic values by calculating 23 the root of the sum of the squared uncertainties. 24 As shown in Table 4 we also tested the effect of equilibrium isotopic fractionation and kinetic 25 isotopic fractionation. The effect is less than 0.2%. 26 Therefore, recommendations for the extraction method are to additionally measure parameters 27 such as the initial pressure in the glass vessel and to ensure a precise determination of the 28 sample volume. Besides this we recommend high-precision IRMS measurements and to 29 consider multiple sampling for better statistics on the data. 30 In total 16 headspace samples were taken during the RV Polarstern cruise in April / May 1 2010 along the transect 32.53° W / 18.79° S to 13.00° W / 36.54° N and 6 samples during the 2 RV L´Atalante cruise in February 2008 between 23.00 -17.93° W to 16.9 -19.2° N to 3 analyse the H 2 mole fraction and the isotopic composition (see Table 4). 4 Although our setup was a prototype with possibilities for improvement, the mole fractions are 5 in line with previously published data. The H 2 excess, Δ(H 2 ), exceeds 5 nmol L -1 , the 6 saturation differ from close to equilibrium to 15-fold supersaturation. Highest supersaturation 7 was found in the southern hemisphere between 16° S and 11° S and in the northern 8 hemisphere around the Cape Verde islands and the coast of Mauritania (Fig. 7a, Table 4). From the measurement of the isotopic composition of H 2 in the headspace we calculate the 7 isotopic composition of H 2 that was originally dissolved in the sea water as described in 8 section 2.4.3 and in the Appendix, using two different assumptions for fractionation between 9 dissolved H 2 and H 2 in the gas phase. The results shown in Table 4 reveal δD values for the 10 dissolved H 2 that vary within a wide range of -112 ‰ to -719 ‰ for both fractionation 11 scenarios. Interestingly, δD shows two distinct groups of samples that can be separated by the 12 water temperature (Fig. 7b). In water masses with a temperature above 21 °C the δD values 13 are (-629±54) ‰ (n = 14), in water masses with a temperature of 20 °C or below δD values 14 are (-249±88) ‰ (n = 8). There is no correlation of δD with salinity (Fig. 7c), but the high 15 temperature (and low δD) waters show also a generally higher saturation than the low 16 temperature (high δD) waters (Fig. 7d). 17 The very depleted isotope signature of the H 2 in the warmer water masses is consistent with 18 the values expected for biological production. The slight enrichment compared to the value of 19 ≈-700‰ that is expected for biologically produced H 2 in equilibrium with ocean water 20 (Bottinga, 1969 dominance of biological formation at higher temperatures is qualitatively consistent with the 31 general understanding of the temperature dependence of N 2 fixation rates for N 2 fixers such as 32 e.g. Trichodesmium spec., which exhibit highest N 2 fixation rates within a temperature range 1 between 24 °C to 30 °C (Breitbarth et al. 2007, Stal 2009). In fact, the saturations also show a 2 correlation with temperature, but less clear than for δD (Fig. 7d), presumably due to 3 simultaneous uptake and consumption processes in a complex microbial community. . We suggest that a source of H 2 must exist in these surface waters, which produces H 2 11 that is out of isotope equilibrium with the water. This can be either one single source with an 12 isotopic signature of approximately -250 ‰, or an even more isotopically enriched source that 13 mixes with the depleted biological source. water might be an alternative source of H 2 excess, which is not isotopically equilibrated with 20 water, especially in regions with high radiation and biological activity, and less N 2 fixation. 21 Given the fact that the two groups of warm and cold waters are relatively well separated and 22 there is not a continuous mixing curve between two end members, the explanation of a single 23 different source seems more straightforward. Isotope analyses are a powerful tool to 24 distinguish this source from biological production. Additional measurements are needed to 25 determine the isotopic signature of such a source and investigate to which extend 26 photochemical production contributes to the oceanic H 2 budget in colder water masses, and 27 also update the current models. However, with an isotopic signature of approximately -250 28 ‰, or an even more isotopically enriched, such a source would not significantly impact the 29 current models. 30 Based on their H 2 measurements, Moore et al. (2014) suggested a substantial underestimation 31 of oceanic N 2 fixation, especially due to high H 2 supersaturations measured in the southern 32 hemisphere. By using direct measurements of N 2 fixation rates a systematic underestimation 1 by approximately 60 % was also proposed by Großkopf et al. (2012) who suggested a global 2 marine N 2 fixation rate of (177 ±8) Tg N a -1 . In order to identify a possible significant 3 mismatch between N 2 fixation rates and total marine H 2 production, we calculated the 4 climatological global oceanic emissions from the GEMS database using the protocol of 5 Identifying sources is important to consider budgets and gain insight in production and 30 consumption processes. Although H 2 has been assumed reasonably to be produced mainly 31 biologically in the oceans, direct evidence was lacking. Our results verify a biological 32 production as a main source of H 2 in oceanic surface water, especially in warmer water 1 masses. As seen from the transects, local sources are difficult to spot due to their patchiness, 2 this should be taken into account when planning the sampling strategy. 3 The unexpectedly high δD values in colder temperate water masses indicate the significant 4 influence of processes other then biological production, and additional information e.g. by 5 isotopic composition is needed to distinguish and verify possible sources and supersaturations 6 of dissolved oceanic H 2 . Especially the investigation of the isotopic composition of possible 7 production pathways such as abiotic photochemical H 2 production needs further attention and 8 should be an upcoming issue. 9 The pattern of mole fractions and isotopic composition of H 2 along a north-south Atlantic 10 transect clearly depends on season and hemisphere and are consistent with previous published 11 data and models. A possible significant underestimation of N 2 fixation as assumed by several 12 authors could -providing a net H 2 release rate -go along with higher H 2 emissions. However      b) Comparing the δD (H2) at different water temperatures, the respective H2 saturations are color coded, sample dots marked with a diamond belong to the RV L´Atalante cruise, sample dots without to the ANT-XXVI/4 cruise; y = -35.2x + 360.9, R 2 = 0.66, n = 22 c) Distribution of δD (H2) (color coded) in correlation between water temperature and salinity d) Correlation between water temperature and H2 saturation, the δD (H2) is color-coded, the exceptional high saturation has been excluded from the correlation calculation, y = 0. 1 Figure 8: Oceanic H2 emissions used in the TM5 model simulations (mmol m -2 a -1 , based on the distribution provided by the project GEMS (Global and regional Earth-system (Atmosphere) Monitoring using Satellite and in-situ data) and scaled to a total oceanic source of 5 Tg a -1 (Pieterse et al. (2013))