The significance of water column nitrification in the southeastern Bering Sea

Chinese Joumnal of Polar Science, Vol. 19 ,No.2, 185 - 192, December 2008The significance of water column nitrification in the southeasternBering SeaClara J Deal , Jin Meibing' and Wang Jia21 Intenational Arctic Research Center, Universily of Alaska Fairbanks, USA2 Great Lakes Environmental Research Lab, NOAA, USAReceived September 20, 2008Abstract Nitrate is considered the nutrient that limits new primary production in thesoutheastem Bering Sea shelf. Nitrate regenerated through biological nitrification hathe potential to significantly support primary production as well. Here we use meas-urements of the specific rate of water column nitrification in a 1-D ecosystem model toquantify the resupply of nitrate fromn nitrification in the middle shelf of the southeast-em Bering Sea. Model sensitivity studies reveal nitrification rale is an important con-trol on the dominant phytoplankton functional type , and the amount of nitrate in sum-mer bottom waters and in the winter water column. Evaluation of nitrification usingthe model supports the hypothesis that inereases in late-summer nitrate concentrationsobserved in the southeasterm Bering Sea bottom waters are due to nitrifcation. Modelresuls for nitrale replenishment exceed previously estimated rates of 20-30% based onobservations. The results of his study indicate that nitrification, potentially the sourceof up to ~ 38% of the springtime water column nitrate, could support ~24% of theannual primary production.Key words Arctice , Southeast Bering Sea, water column ntrification.1 IntroductionIn the southeastern Bering Sea, like other productive high latitude seas, most of theprimary production is fuelled by nitrate. Much of this nitrate is thought to be new nitrate 0-riginating from Pacific waters that flow through several Aleutian passes and deep basin wa-ters that upwell onto the shelf. Another source that has received little attention until re-cently is the biological regeneration of nitrate from ammonium, or nitrification. A recentsynthesis of open ocean measurements indicates that nitrification is a very important factorin the euphotic nitrogen cyclel'l. In this paper, we investigate the role of nitrification insupplying the nitrate needed for the primary production that sustains the world class fisher-ies of the Bering Sea and potentially the arctic food web beyond the Bering Sea.Observations suggest significant nitrification in bottom waters on the southeastern Be-ring Sea shelf. Starting at the time of spring bloom , ammonium regenerated from sinking or-ganic nitrogen accumulates in the middle shelf botto中国煤化工ons markedlydecrease in September-Octoberl2]. The coincidentalYHCNMHGbottom wa-186ClaraJ Deal et al.ters has been hypothesized to be the result of in situ nitification. Processes and Resourcesof the Bering Sea Shelf ( PROBES ) studies of benthic release show evidence of ammoniumconsumed as nitrate is released(3]. However, the net nitrate flux from mid-shelf sedimentswas found to play a minor role in seasonal replenishment of nitrate over the shelf. Tanaka etal. ( 2004)l4] have suggested that the slighly lighter isotopic composition of nitrate ob-served in the bottom water of the middle shelf might be due to new nitrate input by nitrifica-tion. With the onset of winter mixing, nitrate regenerated in summer bottom waters mixesupwards to the surface.Measurements of nitrification rates at Skan Bay, Alaska, show that nitrification occursthroughout the water column5]. In central North Pacific waters, Wada and Hattori( 1971)l6] observed that light does not inhibit nitrification. This observation has since beencorroborated by the synthesis of worldwide measurements that show no clear increase in spe-eifie nitrification rates with depth1]. Worldwide values are noted to vary substantially,with rates spanning four orders of magnitude. The median is 0.195 d -1. In this synthesisof observations , no significant relationship with either time of year or latitude was found ,but the low number of observations could be obscuring any relationship that may exist. Themeasurements of Hattori et al. (1978)[5] and Wada and Hatorri ( 1971)6], are the mostrepresentative of the Bering Sea, being the closest to the location.This ecosystem modeling study was undertaken to examine the role of water column ni-trification in the Bering Sea. In particular , what is the resupply of nitrate from nitrification?And, how much of the annual primary production is supported by regenerated nitrate? Giv-en the uncertainties and sparse observations, we conducted a senstivity study to evaluate arange of nitrification rates and the consequences of parameter selection.2 MethodsA vertically resolved 1 -dimensional (1-D) ecosystem model incorporating nitrificationwas employed to examine this process on the middle shelf of the Bering Sea. The model wasapplied at the NOAA/PMEL M2 mooring site ( Figure 1, water depth = 74 m) where atime series of biophysical measurements has been generated for over the last decade. Smallmean flow at the M2 site supports the assumption that it is reasonable to use a 1-D verticalmodel in this regionl7,8J. The model was run for years 1997191 and 2003. These two yearsare representative of different sea ice conditions on the shelf. In year 1997, sea ice waspresent at M2 from mid-March until April 10. In 2003 annual sea ice did not extend this farsouth. For 2003, a time series of nitrate concentrations at the M2 mooring site from Maythrough December was available for model validationl10J.'The 1-D coupled ice- ocean ecosystem model of Jin et al. (2007)[9] has ten compart-ments: three microalgae ( pelagic diatoms, flagellates and ice algae) ，three zooplankton( copepods, large zooplankton, and microzooplankton), three nutrients ( nitrate + nitrite ,ammonium, and dissolved silicon) and detritus. The seawater ecosystem component isbased on the Physical Ecosystem Model ( PhEcoM)[8]danted frm Flinger et al.(2001)[12], and the ice algae ecosystem model中国煤化工A physicalmodel including a 2.5-level turbulence model is co;Y HC N M H Glel describedabove. The model is forced by tides ，wind，shortwave radiation, and surface heat and saltThe significance of water column nitnification in the southeastem Bering Sea187flux, and restored to available observed daily sea surface temperature and salinity. Initialtemperature and salinity conditions, and nitrate concentration ( 12 μM) were taken fromthe M2 mooring data.Alaska59-●M26-i5t身3--176 -174 -172 -170 -168 -166 -164 -162 -160 -158Fig. 1 Topography of southeastern Bering Sea. The NOAA-PMEL biophysical mooring site M2 is marked withan astenisk in the middle domain' 8.Nitrification is modeled simply as the‘decay ' of ammonium to‘regenerated’nitrateat a constant specific rate-- the simplest assumption, given the current uncertainties re-lated to the process. The nitrogen cycle processes of denitrification and nitrogen fixation,and horizontal transport of nitrate are not included in the model. A specific nitrification rateof0.015 d- 1 was chosen as the model standard value based on estimated in situ nitrite pro-duction rates from the North Pacific Oceanl6」and a study using spike abundance of the 15Nisotope tracked from ammonium to the nitrite and or nitrate pools in the Aleutian Islands ofAlaskal5l. Oxidation of ammonium was measured by tracking the abundance of labeled am-monium added at a concentration of ten μg-atoms N/l. The rates measured from the tracertechnique were similar to the rates estimated from 4- day concentration changes of nitrate,nitrite and ammonium within experimental bottles, and from 18 day in situ changes of thesenutrients in the water column of Skan Bay , Unalaska Island, Alaska. We define the specif-ic nitrification rate, 入nitrif， as the ammonium oxidation rale divided by the correspondingammonium concentration. The standard value of0.015 d-1 for λ nitif is at the lower end ofthe range of observed rates worldwide. The model was run for years 2003 and 1997 usingthis standard λ nti(. In addition to running the model with no nitrification for year 2003，two more λnirnf' sof0.03 and0. 06d'” 1 were tested. To evaluate the sensitivity of the mod-el results to these three Ahitri:'s (i.e. 0, 0.03 and 0.06 d-'), we compared simulated(1) annual net primary production (NPP)，(2) diatom NPP, (3) flagellate NPP, (4)year-end integrated water column nitrate and (5) year-end nitrate concentration at 11-15 mdepth to observations and model results using the standard λ .中国煤化工FYHCNMHG188ClaraJ Deal et al.Results and DiscussionModel results shown in Figure 2a ilustrate the very repeatable spring drawdown of ni-trate during the phytoplankton bloom in May followed by nearly depleted values in the sum-mer. These yearly events are recorded in a composite of mooring and water bottle samplesfrom 11-15 m depth at the M2 mooring site ( 1997-2005 ) ( Figure 2b from Stabeno el al.2006 10]). The record shows a fall enrichment period of nitrate concentrations starting a-round early-October, which is also evident in the 2003 model run ( Figure 2a).Feb- Mav01M2 MooringOeoiLMaro2MaV-Sep03Sep03-Apr04bottle(97-05)May-Sep03(1)，12-Apr-Aug05030 60 90 120150180210240270300330030 60 90120150180210240270 3030day of yearFig.2 Time series of (a) simulaled nitrate concentrations ( this study) and (b) nitrate measurements ( Stabe-no et al.2006)l10] at depth 11-15 m at the M2 site. The black diamonds are data from shipboard meas-urements.The standard λ nitnif of 0.015 d -1 produces the closest match of the model results to theobservational data. Figure 3 shows the comparison of mooring fluorometer data with simula-ted total phytoplankton at 12 m depth for year 2003. The comparison of mooring fluorometerdata with simulated total phytoplankton at 12 m depth for year 1997 is an even better match(see Figure4a in Jin et al. 2006b131 ). No nitrification and higher rates of nitrification(0.03 and 0.06 d-1 ) result in lower and higher concentrations of nitrate at 11-15 m depth(Table 1; 4.2, 11.4, and 12.3, respectively), compared to 10 μM observed in 2003(Figure 2b; day 0 value for year 2004) and 9.6 μM using the standard 入nirnif of 0.015d-l.. 16F. otu 720Algeg 12--D+F1.5多云1(-1.0鱼6-40.5盖0102030405060708091011 12Month of year 2003 (depth=12 m)Fig. 3 Time series of fluorometer observations and simulated l中国煤化Ipuhof 12 m.YHCNMH GSpecifce nitrification rate also appears to afct the proportion of phytoplankton net pri-The significance of water column nitrification in the southeastem Bering Sea189mary production ( NPP) among different phytoplankton types ( Table 1), but not totalNPP. Higher Anitnt' s favor diatom production in the model. This predilection is likely dueto input of ammonium from the senescing spring bloom and rapid conversion of ammonium tonitrate. Complete turnover in the ammonium pool in less than24 h may be possible, evenin surface watersll. Regeneration of nitrate in the model occurs early in surface waterswhen diatoms are the dominant bloomers. Grazing on the diatoms then routes the N throughsecondary producers and sinking detritus making fewer nutrients available in the eupohoticzone for flagellate production.Nitr alei mmol Nm'Ammoniumi mmol Nm' ;兰-50:200304 05060708910111201010203040506070809101112Niratet mmol Nm')Ammoniummmol Nm'y(b)-5/E200304 05 06 07081910 1 120102030405060708091011 12Nitratel mmol Nm')Ammoniumt mmol Nm')03 04 05 0607080910 11 12010203040506070809101112 01Fig.4 Depth-time contour plots of simulated nitrate for year (a) 2003 with nitification, (c) 2003 without ni-trification, and (e) 1997 with nitrification, and ammonium for year (b) 2003 with nitrification, (d)2003 without nitrification, and (f) 1997 with nitrification.Table 1. The infuence of selected parameter values on simulated total net primary production (NPP gC m-1y-2), diatom NPP, flagellate NPP, year-end integrated water column nitrate, and year- end nitraleconcentration at 11-15 m depth. The observed year-end nitrate concentration at 11-i5 m depth was10 μ M (day0 for year 2004 in Figure 2b) .Year-endYear-end ni-Net PrimaryNitrifcation RateYearDiatom NPPFlagellatenitratetrate at 11-15(d-l)NPP( mmolm depth(NPP)N m-(uμ M)No nitrification200390.419.870.62944.20. 015 ( standard)90.628. 162.46659.60.0390.240.849.37850.0691.251.086012.30.015 ( tandard)199735.6Even though modeled total NPP is not impacted by the selection of Aniri(，, the amountof nitrate remaining in the simulated water column at the end of the year is. A factor of twoincrease in the rate results in a 15% increase in mid中国煤化工nitrate (Ta-ble 1). No nitrification halves the amount of nitrateMHCNMHG.06d-1 re-sults in a 23% increase in nitrate in the water colum.' Higher mid-winter nitrate is reflected in the higher concentrations of nitrate below the mixed layer. Ni-190ClaraJ Deal et al. .trate below the summer mixed layer will only impact the total NPP if mixed upwards into theeuphotice zone, by e. g. storm, wind or convective mixing. This implies that a change in cli-mate towards more storms may result in higher NPP. On the other hand , if warming trendsresulting in stronger ocean stratification continue these occurrences may be lessened.Ammonium concentrations begin to increase in bottom waters following the onset ofsinking phytoplankton and detritus. As summer progresses ，nitrate concentrations increaseas the ammonium concentrations decrease in late-summer ( Figures 4a and 4b). This pat-tern is not seen in the absence of nitrification ( Figures 4c and 4d). Contour plots for theannual cycle during PROBES display a similar pattern of declining ammonium and increas-ing nitrate21. In 1997, the maximum in the modeled ammonium concentration occurred inmid-May. This year was different from the others in that sea ice was present at the site fromMid-April through May. An earlier modeling studyl9」revealed that the phytoplankton bloomin 1997 was seeded by ice algae released from the sea ice. Decomposition of the sinking seaice algae to ammonium likely resulted in the earlier ammonium concentration maximum. Al-though, the timing of the maxima is different in individual years, the recurring patterm forthese representative years provides clear evidence of nitrification.The amount of simulated nitrate at 11-15 m depth in the water column at year-end va-ries as well, 9.5 vs. 8.7 μM nitrate on January 1, for year 2004 and 1997，respectively.Recent implementation of a moored nitrate sensor at M2 has recorded differences betweenyears ( Figure 2b). Mid-winter values of nitrate were observed to be at least 2 μM higher in2004 than in 2003. Timing and degree of temperature stratification of the water column ap-pear to be related to the individual differencesl 10J. It follows then that change in climatethat influence stratification will play a role in the availability of nitrate.The model results with and without nitrification ( Figure 5) ilustrate that by the end00-Standard| .... No nitrification800 t00 t公Z 600-300 t05 06 07中国煤化工Fig.5 Model time series comparing water cnuns with nitrification ( solid line) anJIYHCNMHGline) for year 2003.The sigificance of water column nitrification in the southeastemn Bering Sea191of the year, well over half of the nitrate in the water column is regenerated nitrate; an a-mount equal to ~ 38% of the initialized nitrate content when the model run started on Janu-ary 1, 2003. This amount is somewhat higher than the estimated 20 30% nitrate inputthrough nitrification3」 based on PROBES time series dala. To estimate how much the a-mount of regenerated nitrate would contribute to annual primary production, we initializedthe model run for 2003 with 38% less nitrate (i.e. 7.5 μM instead of 12 μM). The resultwas 24% less primary production over the year-long model simulation period.4 ConclusionsKnowledge of nutrient sources , including nitrification, is needed to understand the im-pact of a changing climate on the nitrate supply on which arctic marine life critically de-pends. Nitrification is a difficult process to quantifyll. More measurements of nitrificationrates are needed, especially in Arctic waters where virtually none exist.When measurements are limited, using a model is an especially good way to gain in-sights and help focus scientific efforts on critical questions. One question raised by thismodeling study is, how does nitrifcation rate impact the dominant phytoplankton type? Wefound that different λ nirnit' s do not significantly impact total primary production (i. e. simu-lated NPP), but rather the contribution to primary production from specific phytoplanktoritypes supported by the available N shifts. It therefore follows that any change in environ-mental or ecological factors that impact the two different groups of bacteria that convert am-monium to nitrite and nitrite to nitrate in seawater has the potential to influence the domi-nant phytoplankton functional type. The results of this modeling study also suggest that at even the relatively low nitrification rates observed in the vicinity of the Bering Sea and insub-arctic Pacific waters, nitrification is an important source of the nitrate available for pri-mary production.Acknowledgements This study was supported by North Pacific Research Board ( NPRB)grant 607 awarded to Jin, Deal, and W ang. The International Arctic Research Center sup-ported this study through the JAMSTEC-IARC Research Agreement and NSF - IARC Coop-erative Agreement. We thank NPRB and PMEL NOAA for supporting the collection of dataat site 2. This is GLERL Contribution No. 1498.References[ 1] Yool A, Martin AP, Fermandez C, Clark DR( 2007 ): The significance of nitrification for oceanic newproduction. Nature, 447, doi :10.1038/ nature05885.[2]Whitledge TE, Luchin VA( 1999) : Summary of chemical distributions and dynamics in the Bering Sea.In; Loughlin TR & Ohtani K Eds. , Dynamics of the Bering Sea, 217-250. Fairbanks, Alaska: Univer-sity of Alaska Sea Grant.[3] Whitledge TE, Reeburgh WS, W alsh JJ( 1986) : Seasonal inorganic nitroen distributions and dynamicsin the southeastem Bering Sea. Continental Shelf Resea中国煤化工[4] Tanaka T, Guo L, Deal C, Tanaka N, WhitledgeT, Mo:MHCNMHG*well oxygena-ted cold bottom water over the Bering Sea shelf; InfuenuoU s... iinContinentalShelf Research, 24(2): 1271-1283.192ClaraJ Deal el al.[ 5] Hattori A, Goering JJ, Boisseau DB( 1978 ). Ammonium oxidation and its significance in the summercycling of nitrogen in oxygen depleted Skan Bay , Unalaska Island , Alaska. Marine Science Communica-tion, 4(2): 139-151.[6] Wada E, Hattori A ( 1971) : Nitrite metabolism in the euphotic layer of the central North Pacifie Ocean.Limnology and Oceanography, 16: 766-772.[ 7] Coachman LK( 1986) : Girculation, water masses, and fuxes on the southeastern Bering Sea shelf. Con-tinental Shelf Research, 5: 23-108.[8] Jin M, Deal CJ, Wang J, Shin KH, Tanaka N, Whitledge TE, Lee SH, Gradinger R( 2006a) : Controlsof the land fast ice-ocean ecosystem oflshore Barrow, Alaska. Annals of Glaciology, 44: 63-72.[9]Jin M, Deal C, Wang J, Alexander V, Gradinger R, s. Saitoh, T. lida, Z. Wan, and P. Stabeno(2007). lce-associated phytoplankton blooms in the southeasterm Bering Sea. Geophysical ResearchLelters, 34, L06612, doi:10. 1029/ 2006GL028849.[10] Stabeno P, Napp J, Whitledge T ( 2006 ). Long-term observations on the Bering Sea shelf ( 2004-2005): Biophysical moorings at sites 2 and 4 as sentinels for ecosystem change , NPRB Project 410 FinalReport.[1] Wang, J, Deal CJ, Wan z, Jin M, Tanaka N, Ikeda M(2003). User' s Guide for a Phyical Ecosys-tem Model ( PhEcoM) in the Subpolar and Polar Oceans. IARC-FRSGC Technical Report 03-01 Version1: 75.[12] Eslinger DL, Cooney RT, McRoy CP, Ward A, Kline T, Simpson EP, Wang J, Allen JR (2001):Plankton dynamics: Observed and modeled responses to physical forcing in Prince William Sound. Alas-ka Fisheries Oceanography., 10, suppl. 1, 81-96.[13] Jin M, Deal CJ, Wang J, Tanaka N, Ikeda M(2006b): Vertical mixing efecte on the phytoplanktonbloom in the southeastem Bering Sea mid-shelf. Jourmal of Geophysical Research, I1, C03002, doi;10. 1029/ 2005JC002994.中国煤化工MYHCNMHG