Application of the isotope pairing technique in sediments where anammox and denitrification coexist

Transcription

1 LIMNOLOGY OCEANOGRAPHY: METHODS Limnol. Oceanogr.: Methods 1, 003, , by the American Society of Limnology Oceanography, Inc. Application of the isotope pairing technique in sediments where anammox denitrification coexist Nils Risgaard-Petersen 1*, Lars Peter Nielsen, Søren Rysgaard 1, Tage Dalsgaard 1, Rikke Louise Meyer 1 National Environmental Research Institute, Department of Marine Ecology, Vejlsøvej 5, DK-8600 Silkeborg, Denmark University of Aarhus, Institute of Biological Sciences, Department of Microbial Ecology, Ny Munkegade Bldg., 540 DK-8000 Århus C, Denmark Abstract The isotope pairing technique (IPT) is a well-established N method for estimation of denitrification. Presence of anammox, the anaerobic oxidation of NH 4+ to N with NO results in violation of central assumptions on which the IPT is built. It is shown that anammox activity causes overestimation of the N calculated by the IPT. However, experiments with different additions of NO 3 will reveal the problems posed by anammox. Two alternative calculation procedures are presented, which enable a more accurate quantification of anammox denitrification activity in sediments where the processes coexist. One procedure is based on measurements of N-N in -amended intact sediment cores data addressing the contribution of anammox to total N estimated from slurry incubations. The other procedure is based on measurements of N in at least two parallel series of sediment cores incubated with different additions. The calculation procedure presented is used on field data from four studies where the IPT was used the potential anammox rate measured. The IPT overestimated total N-N rates by 0%,.5%, 31%, 8% relative to the revised estimates from the 4 different sites, where anammox accounted for 0%, 6%, 18%, 69.8%, respectively, of N. The overestimation of true denitrification was, however, up to several hundred percent. Our analysis suggests however that the IPT does not seriously overestimate N in estuarine sediments because anammox accounts for <6% of N in such sediments, according to present knowledge. The isotope pairing technique (IPT) (Nielsen 199) is a well-established N technique for estimating denitrification of bottom-water NO 3 + NO ( ) of produced via sedimentary nitrification. During the past decade, this technique has been used in numerous studies, including temperate (Cabrita Brotas 000; Dong et al. 000; Ogilvie et al. 1997; Ottosen et al. 001; Trimmer et al. 000a), tropical (Kristensen et al. 1998), arctic marine sediments (Glud et al. 1998; Glud et al. 000; Rysgaard et al. 1998), lake sediments (Mengis et al. 1997; Risgaard-Petersen et al. 1999; Svensson et al. 001), rivers (Pind et al. 1997; * nri@dmu.dk Acknowledgments This study was supported in part by grants from the Danish Natural Science Research Councils (Contract Nr Contract Nr ) in part by the ICON project under the European Union Fifth Framework Programme, project nr. EVK1-CT the Foundation for Strategic Environmental Research, Sweden (MIS- TRA). Three anonymous referees are acknowledged for their constructive comments. Trimmer et al. 000b), wetls (Davidsson et al. 1997; Hoffmann et al. 000; Stepanauskas et al. 1996). A review of the method its applications is given by Steingruber et al. (001). The IPT aims to quantify the genuine rate of N gas, i.e., N-N as it would occur without the addition of. According to the technique, this corresponds to the rate of 8 N ( N N) times plus the rate of 9 N ( N N) after addition of to the system. This (p ) is estimated from the rates of 9 N (p 9 N ) N (p N ) using the following expression of Nielsen (199): 9 p N 9 ( p N p N + ). (1) p N The supposition that the genuine rate of N equals 8 N + 9 N that this can be calculated from Eq. 1 is based on a number of assumptions. First, it is assumed that addition of NO 3 does not affect the of N-N. This assumption is valid if denitrification is 63

2 Anammox isotope pairing Fig. 1. Distribution of N isotopes in N labeling experiments where both anammox denitrification occurs. A 8 A 9 are the produced pools of 8 N 9 N originating from anammox, whereas D 8, D 9, D denote the pools of 8 N, 9 N, N produced via denitrification. p 8 N, p 9 N, p N are the integrated pools of 8 N, 9 N, N, respectively. Dark arrows represent the anammox process light arrows denitrification. Thin arrows represent NO 3 reduction. the only N -producing process if the process is limited. Under these conditions, addition will not affect the rate of reduction but only the manner in which reduced is distributed between 8 N 9 N. Second, it is assumed that the three isotopic N species produced, 8 N, 9 N, N, are binomially distributed (Nielsen 199). This requirement is met in denitrifying environments if the ratio between is constant throughout the NO x reduction zone, which is the third assumption of the IPT. Several tests including time NO 3 concentration series are available as documentation for these critical assumptions. If the assumptions hold, the results of the IPT should demonstrate the following: There should be a linear relationship between the concentration of added the rate of N- N. The of N-N as estimated from Eq. 1 should, however, be independent of the amount of NO x added (see Nielsen Glud 1996; Rysgaard et al. 1995; Dong et al. 000). Comparison of the rates measured in identical experimental systems, where denitrification was presumed to be the only N -producing process, has shown good agreement between estimates obtained by the IPT, the N flux method ( 1998a), the N /Ar method (Eyre et al. 00), respectively. This is a strong indication of the validity of the technique. Anammox represents an alternative N -producing process, where NH 4+ is oxidized to N with NO under strictly anoxic conditions (Strous et al. 1997). The process was first discovered in a wastewater treatment plant by Mulder et al. (1995), recent studies have shown that anammox may be a significant process in marine sediments too, where it can account for more than 60% of anaerobic N (Thamdrup Dalsgaard 00). Presence of anammox is a challenge to the isotope pairing technique, because it leads to violation of both the assumption of independence between added NO 3 N-N the assumption of binomial distribution of produced 8 N, 9 N, N. Bacteria with anammox capacity will only produce two of three possible isotopic N species, 8 N 9 N, following addition of (Fig. 1). In systems where denitrification anammox coexist, the resultant pool of N produced after addition will therefore be a mixture of pools with different isotopic compositions (Fig. 1) binomial distribution of produced 8 N, 9 N, N can no longer be assumed. This implies that the genuine rate of N estimated as 8 N + 9 N cannot be calculated from Eq. 1. Violation of the assumption of independence between the amount of added NO 3 N-N can be illustrated with the following example: Consider a sediment without nitrification without in the overlying water. In such a system, no N would take place. If there is a capacity for denitrification alone, addition of will result in of only N. In this case, the IPT will correctly estimate the N-N rate as zero from Eq. 1. However, if the capacity for both anammox denitrification exists, addition will result in of both 9 N N (Fig. 1), the IPT would erroneously estimate a N-N according to Eq. 1. The problem is more general. In systems exhibiting anammox where anaerobic N is limited by the availability of, which is a prerequisite for extrapolation from N tracer results to the unamended state, addition will increase the rate of NH 4+ oxidation via the reaction NH 4+ + NO 9 N (Fig. 1). The alternative isotopic N species, 8 N, is produced by the reaction NH 4+ + NO 8 N. In such a system, the two reactions are independent of one another, the rate of 8 N therefore remains constant whereas the of 9 N increases as more is added. This implies that N-N estimated as 8 N + 9 N is correlated with the amount of added therefore does not reflect genuine N of the nonmanipulated sediment. In the present study, we focus on solutions to the problems associated with the application of the IPT in sediments where anammox denitrification coexist. We develop mathematical expressions that enable evaluation of results obtained with the IPT, we furthermore develop alternative calculation procedures that enable quantification of N-N from data obtained through isotope pairing experiments. It should be stressed that these equations are based on a set of assumptions, we cannot guarantee that these are valid in all systems because of the limited present knowledge of the distribution regulation of anammox. However, the equations facilitate test quality assurance procedures that can be applied to accept or reject results. 64

3 Anammox isotope pairing Table 1. List of variables used in the equation system Parameter Designation p Genuine N (units N units time 1 ) p IPT Genuine N (units N units time 1 ) estimated with the classical IPT p 9 N Total of 9 N p N Total of N Production of 8 N via denitrification D 8 D 9 D A 8 A 9 r r w ra V R 9 Ratio between p 9 N p N Production of 9 N via denitrification Production of N via denitrification Production of 8 N via anammox Production of 9 N via anammox Ratio between in the reduction zone Ratio between in the water column Contribution of anammox to N Ratio between concentration in the water for two different incubations with different amendments Within the concepts developed, we evaluate some of the very limited data comprising information on both anammox N measured by isotope pairing. These data originate from studies in the Skagerrak (Rysgaard et al. 001; Thamdrup Dalsgaard 00); two shallow-water Danish estuaries: Norsminde Fjord Rers Fjord (Risgaard- Petersen et al. pers. comm. unref.); sediments from the high-arctic fjord Young Sound, Greenl (Rysgaard et al. pers. comm. unref.). Materials procedures Definitions assumptions In the following discussion, we will use a series of terms that are listed in Table 1. A central variable in our discussion is p. This variable represents genuine N expressed in units of produced N atoms per unit time, i.e., the N-N which, in the IPT, is presumed to be unaffected by addition of NO 3. Our discussion is based on the following assumptions: 1. The ratio between N-labeled unlabeled species in the reduction zone is constant after amendment. As mentioned above, this assumption is also central to the IPT according to results from NO 3 concentration series experiments (Nielsen Glud 1996; Rysgaard et al. 1995; Dong et al. 000; Eyre et al. 00), it appears to be valid for a variety of sediments. The assumption implies that any isotopic N species produced via denitrification or anammox can be estimated from the mole fraction of N in undergoing dissimilatory reduction any other N isotopic species produced via a matching reduction process (see Appendix 1).. The mole fraction of N in the NO 3 NO pools undergoing dissimilatory reduction is similar. That is, we assume that NO x is supplied to anoxic sediment strata mainly as NO 3 then reduced further to N with NO as a free intermediate. Direct transfer of NO between aerobic ammonia oxidation anammox/denitrification is thus assumed to be insignificant. The absence of this shunt is also central in the application of the IPT in sediments without anammox because it will not be traced by NO 3 amendment, leading to underestimation of denitrification. The good agreement between estimates obtained by the isotope pairing, N -flux, N /Ar methods ( 1998; Eyre et al. 00), when applied in identical experimental setups, suggests that direct NO transfer between aerobic ammonia oxidation anammox/denitrification is insignificant compared with the supply of NO 3. Porewater profiles of NO measured in biofilms with NO microsensors (de Beer 000) in freshwater sediments using fine-scale porewater extraction methods microsensors (Stief et al. 00) in marine sediments using biosensors (Meyer Risgaard- Petersen in prep. unref.) indicate that net NO takes place mainly in the anoxic zone of the sediment, as a result of NO 3 reduction. The data from Stief et al. (00) Meyer Risgaard-Petersen (in prep. unref.) also suggest a very efficient capacity for aerobic NO oxidation, leading to no net transport of NO to the anaerobic sediment strata. In addition, Stief et al. (00) observed an increased anaerobic NO after addition of NO 3 to the overlying water. Thus, there is good support for the assumption that NO 3 addition leads to the formation of similar mole fractions of N in both the NO 3 NO pools. However, if the concentration of NO in the water column is sufficiently high, it may penetrate the oxic zone, thus leading to anaerobic gas formation in the anoxic zone (Dong et al. 00). If NO 3 is used as a tracer in such systems the mole fraction of N in the NO 3 NO pools in the anoxic sediment strata will not be similar. 3. Anammox denitrification are both limited by the supply of NO 3 from above the oxicanoxic interface. In this case, addition of will result in an increase in both anammox denitrification activity will neither affect 65

4 Anammox isotope pairing N-N via denitrification (i.e., 8 N + 9 N ) nor 8 N via anammox. This assumption is confirmed with respect to denitrification in a variety of environments where anammox probably plays an insignificant role by the demonstration of a linear response of N-N rates to amendment constant N-N rates (Nielsen Glud 1996; Rysgaard et al. 1995; Dong et al. 000; Eyre et al. 00). In situ control of anammox denitrification in sediments exhibiting significant anammox activity is poorly described, no data are available at present on dependency of the processes in intact sediments. However, porewater profiles of NH 4+ measured in earlier studies of Skagerrak sediments (Canfield et al. 1993) where a significant capacity for anammox has been reported (Thamdrup Dalsgaard 00) do not indicate that NH 4+ supply is limiting to the anammox process, because NH 4+ was found in the oxic zone in the reduction zone in concentrations between 10 0 µm. The K m -value for NH 4+ in anammox bacteria has been reported to be less than 5 µm (Strous et al. 1999). Because addition of will exp the reduction zone as shown in microsensor studies (Christensen et al. 1989), bacteria with anammox capacity situated in deeper more NH 4+ rich strata (Thamdrup Dalsgaard 00) will be exposed, resulting in an increase of overall depth-integrated activity. We assume in this argument that the uptake kinetics for denitrifiers bacteria with capacity for anammox are similar, thereby that NO 3 addition will not change the proportion between anammox denitrification. The data from Dalsgaard Thamdrup (00) indicate that this presumption is valid. In a series of slurry experiments, these authors observed that K m for NO reduction in sediments was <3 µm. Nitrite reduction in these sediments was due to both anammox denitrification, according to Dalsgaard Thamdrup (00), the proportion between the processes was constant from the beginning of their experiment to the end, where NO was depleted. 4. Nitrification is not affected by the addition of. In sediments with efficient anammox, it is possible that an increase in anammox activity in response to NO 3 addition (see above) will reduce the flux of NH 4+ into the oxic zone. This may lead to NH 4+ limitation of nitrification thereby to a reduced N coupled with nitrification. However, moderate NO 3 additions will probably only have a minor impact on the NH 4+ flux into the oxic zone thus will not seriously affect the nitrification process. According to data of Dalsgaard + Thamdrup (00), the anammox process does not deplete NH 4 in slurry incubations. In earlier studies of these sediments (Station S9, Skagerrak), Canfield et al. (1993) observed that C-oxidation consequently NH 4+ was mainly the result of Mn reduction, suggesting that NH 4+ for anammox nitrification is mainly supplied from below the O NO x reduction zones. This may indicate that NH 4+ is in excess that a stimulation of the rate of anammox will have only a minor impact on the flux of NH 4+ into the oxic zone. The error imposed on the IPT in determination of sediment N In this section, we present a more formal discussion of the consequences of anammox for the estimate of N obtained by the IPT. Here we investigate the difference between the true rate the estimate of N obtained by the IPT. According to Nielsen (199), the isotope pairing technique estimates genuine N (p ) from the rates of 9 N (p 9 N ) N (p N ): (see Eq. 1 Appendix 1). If anammox is present, the measured of 9 N will integrate the 9 N from both denitrification anammox, whereas the measured N only represents denitrification. Thus, in these situations p 9 N is the sum of 9 N from anammox (A 9 ) denitrification (D 9 ) whereas p N is the of N from denitrification alone (D ) (Fig. 1). Eq. 1 then becomes Assuming that it was practically possible to distinguish between the N rates resulting from anammox denitrification, the correct estimate of p would be the sum of N-N from denitrification anammox, respectively. This corresponds to twice the 8 N formed via denitrification (D 8 ) plus the 9 N resulting from denitrification (D 9 ) plus twice the 8 N from anammox (A 8 ). As mentioned above, 9 N from anammox must be excluded as this represents induced oxidation of NH 4+ caused by addition. The correct procedure for determination of genuine N is therefore p D + D + A (). (3) According to assumption 1, the denitrification term D 8 + D9 D 9 equals ( D + D9), i.e., the classical IPT term. According to assumptions 1, the anammox term A 8 D equals A 9 r, where r is the ratio between NO x ( : ) in the reduction zone (see Appendix 1). By substitution, Eq. 3 becomes: D9 ( D + D9) + A9 r ( ) D. (4) The difference between the correctly calculated N-N the estimated by the IPT is thus the difference between Eqs. 4: (5) 66

5 Anammox isotope pairing This corresponds to the error that arises from use of the isotope pairing technique. We wish to express the error in terms that are easier to interpret, further rewriting is therefore necessary. Our strategy is to first express all terms in Eq. 5 as functions of the genuine N-N (p ), the contribution of anammox to N (ra) the : ratio in the reduction zone (r ), then substitute these new terms into Eq. 5. First we consider the denitrification terms, D 9 D. The of N-N formed via denitrification can be expressed as follows: p ( 1 ra) D + D. (6) 9 8 The of 8 N formed via denitrification can be expressed as a function of the of 9 N via this process, a consequence of assumption 1 (see Appendix 1). D D r 8 9. (7) Hence Eq. 6 is equivalent to p ra D D r ( 1 ) 9 + 9, (8) D 9 is isolated as follows: ( 1 ra) D9. (9) ( 1 + r) The of N via denitrification can also be expressed as a function of the of 9 N via this process (see Appendix 1). 1 D D9 r (10) D 9 is substituted with Eq. 9, D becomes ( 1 ra) D. (11) r ( 1+ r) Likewise, we can express A 9 in terms of p, ra, r. The genuine N-N via anammox can be expressed as follows: ra A8 (1) Because A 8 r A 9 (see Appendix 1), A 9 can be expressed as follows: ra A9. (13) r All terms in Eq. 5 can now be expressed as functions of p, ra, r, by substitution, Eq. 5 becomes () The magnitude of the error caused by the IPT calculation procedure relative to the amount of N-N being produced according to the revised calculation procedure is thus:. () Fig.. Overestimation of N as a function of N in the NO x pool undergoing dissimilatory reduction (upper X-axis) r (lower X-axis). The lines represent scenarios with different values of ra, the contribution of anammox to total N. Eq. is a nonlinear increasing function of ra a decreasing function of r, the N : N ratio of the substrate undergoing dissimilatory reduction (Fig. ). Presence of anammox will therefore lead to overestimation of N when estimated by the IPT, this overestimation is positively correlated with the concentration of added to the system. This is not surprising because the IPT includes 9 N from anammox in the calculation of N-N, this represents an additional oxidation of NH 4+, which would not have taken place in the absence of addition. In a previous study, it has been proposed that presence of anammox will not affect the genuine N as estimated by the IPT, but only complicate identification of the processes involved (Ogilvie et al. 1997). As shown in the present study, this is not correct. Estimating N-N correctly: an alternative approach In the previous section, we have focused on the problems associated with application of the IPT in sediments where anammox denitrification coexist. In the following, we propose two alternative calculation procedures for calculation of genuine N (p ). One of these relies on data from at least two sets of intact sediment cores incubated with different amendments. The other procedure relies on a single incubation measurements of ra, the contribution of anammox to N. Both procedures are based on calculations of p from measured rates of N-labeled N estimates of r 67

6 Anammox isotope pairing Table. Equations used for calculating genuine N Parameter Equation Estimate of N r [p 9 N +p N (1r )] Estimate of r if ra is known r (1ra) R9 ra ( ra) R 9 is the ratio between 9 N N, ra is the contribution of anammox to N. Estimate of r from incubation with two different concentrations (1) r p9 (1) N V p 9 () N p (1) N V p () N p 9 (1) N p (1) N are rates of 9 N N in incubation 1, p 9 () N p () N are rates of 9 N N in incubation a parallel incubation with a different NO 3 concentration in the water column. V is the ratio between concentration in the water column in incubations 1. V can also be estimated from: V p 9 (1) N + p (1) N p 9 () N + p () N the ratio between in the reduction zone both procedures estimate this ratio. The equations of interest are given in Table. Below we present the rationale. First we will develop the general formula, in which the only unknown term is r, then we will show how to estimate r with the two procedures. Expressing p from measurable parameters: p 9 N, p N, r The correct estimate of genuine N is given by Eq. 3. In the previous section, we expressed D as a function of D 9 r (Eq. 10). Because only denitrification produces N (Fig. 1), D equals p N D 9 is therefore equivalent to D r p N. (16) 9 In the previous section, we also expressed D 8 as a function of D 9 r (see Eq. 7). Using Eq. 16, D 8 can be expressed as follows: r D 8 D9 r p N. (17) The of 8 N via anammox (A 8 ) can be expressed as a function of the of A 9 r : A 8 r A 9. (18) The 9 N via anammox corresponds to the difference between total 9 N (p 9 N ) the of 9 N via denitrification. Hence A 9 equals 9 A p N r p N. (19) 9 Using this relationship we can express A 8 as follows: 9 A8 r ( p N r p N). (0) All terms in Eq. 3 are now expressed as functions of p 9 N, p N, r, Eq. 3 becomes (1) In Eq. 1, all parameters can be directly measured, except for r for which calculation procedures will be described in the following section. Estimation of r, method 1: determination of the contribution of anammox to total N It is possible to estimate r thereby p from rates of 9 N N if ra, the contribution of anammox to N, is known. At present, published techniques enable simultaneous quantification of the potential for anammox denitrification in anoxic jars with sediment from defined sediment strata. From such data it is possible to estimate the contribution of anammox to N (Thamdrup Dalsgaard 00). Such an estimate can be considered as a measurable proxy for ra. We can express the contribution of anammox to N (ra) according to Eq. 1: A ra 8. () By combining Eqs. 0 1 with Eq., we can express r in terms of measurable parameters: Solving for r gives r ( 1 ra) R9 ra ( ra). (3) (4) where R 9 is the ratio between 9 N N. Thus, if ra is known, for instance through the technique described by Thamdrup Dalsgaard (00), the genuine N can be estimated by combining Eqs

7 Anammox isotope pairing Estimation of r, method : multiple incubation It is possible to estimate r p on the basis of 9 N N rates measured in at least two sets of parallel intact cores incubated with different concentrations. In this section, we will develop a set of equations for that purpose. In order to do so, we will use an assumption central to our theory: namely that the of 8 N via anammox (A 8 ) is independent of addition. Thus, using Eq. 0 we can express A 8 in each of two sets of sediment cores incubated with different additions as follows: (1) 9 (1) A 8 r (p N r p N ) (1) (1) (5) denitrification of bottom-water NO 3 NO 3 produced by sedimentary nitrification during incubation (Nielsen 199). The question is whether or not these two sources of denitrification can be estimated in sediments where anammox denitrification coexist. In the previous section, we showed that it is possible to estimate r, the ratio between, in the reduction zone. Using this information, it is a simple task to quantify the dependency between p the concentration of in the water column (p w) the dependency between p nitrification (p N): rw w r (31) ( ) 9 ( ) 8 A r ( p N r p N ) ( ) ( ) (6) rw n w ( 1 ) r (3) where p 9 (1) N p (1) N are rates of 9 N N resulting from one incubation with, p 9 () N p () N are the rates of the respective N species from a parallel series of sediment cores incubated with a different concentration of (1) in the water column. r () r represent the respective : ratios in the NO x reduction zone. In an experimental setup such as this, it is possible to express r for one incubation series (series 1) as a function of () (1) r in another incubation series (series ): r V r where V is the ratio between the concentrations of in the water column of the two incubations. This approximation is valid if addition of NO 3 only changes the concentration of in the reduction zone, whereas the concentration of in that zone remains constant (an implicit consequence of assumption 3). If there is a linear relationship between the concentration of added the amount of N-N produced (assumption 3) V equals [ NO ] 3 V NO [ 3 ] () 1 ( ) 9 p N 9 p N () 1 + p () 1 N ( ) + p ( ) N Eqs. 5 6 can then be expressed as follows: () 1 9 () 1 8 A r ( p N r p N ) ( 1) ( 1), (7) (8) A r () 1 9 ( ) 1 V ( p N r V p N ). (9) ( ) 8 From this set of equations, r can be determined from measurable terms: () 1 r () 1 9 ( ) () ( ) 9 p N V p N 1 p N V p N. () With this estimate of r, the genuine N-N can be calculated from Eq. 1. Estimation of N based on nitrate produced in the sediment (p n) or coming from the overlying water (p w) The classical IPT allows quantification of distinction between where r w is the ratio between in the water column. Procedures for evaluation of the classical the updated IPT Experiments with different NO 3 additions can be used to validate both classical IPT the alternative procedure for calculation of p. We recommend that such experiments be performed as a stard procedure when measuring N with N-isotopes. In this context, experiments with multiple ( 4) concentrations of NO 3 should be performed the responses of N N-N evaluated using stard techniques such as analysis of variance accompanied by a power analysis to evaluate the experimental design. In addition, the contribution of anammox to N (ra) should be measured with the method described by Thamdrup Dalsgaard (00). Data from experiments such as these will facilitate interpretation of data from the concentration series experiment. We will consider two main outcomes of these experiments: 1. There is a positive correlation between N-N estimated by the classical IPT the amount of added. In addition, there is a linear dependency between the of N-N (i.e., p N + p 9 N ) the concentration of NO 3 applied.. There is no significant correlation between N-N estimated by the classical IPT the amount of added but there is a linear dependency between the of N-N the concentration of NO 3 applied. If there is no correlation between N-N NO 3 addition, N is not NO 3 -limited NO 3 tracer techniques cannot be used to estimate genuine N. An experiment resulting in outcome 1 is indicative of unsuccessful application of the classical calculation procedure may suggest presence of anammox. This is clear from our finding in the section, The error imposed on the IPT in determination of sediment N, where we demonstrated the positive correlation between the fraction of N in the NO x undergoing dissimilatory reduction the degree of overestimation of genuine N when estimated with the 69

8 Anammox isotope pairing Table 3. Production rates of 9 N (p 9 N ), N (p N ), the contribution of anammox to total N as estimated in slurry incubations (ra) the ratio between in the reduction zone estimated with Eq. a p 9 N p N ra Locality (µmol N m h 1 ) (µmol N m h 1 ) (%) r Norsminde Fjord (1 m) b 18.8 (.7) 131 (8.6) 0 b 0.55 (0.0) Rers Fjord (1 m) b 1.5 (17.4) 44.7 (4.9) 6. b 1.33 (0.03) Young Sound (36 m) c 7.54 (4.01) 7.7 (3.73) 18.6 e (0.) S9, Skagerrak (695 m) d 1.91 (0.83) (0.08) 69.8 f 1.79 (0.37) a SE of the mean are given in parentheses, n 5. b Raw data from pers. comm. unref. c Raw data from Rysgaard et al. (1998). d Raw data from Rysgaard et al. (001). e Data from Rysgaard et al. unpubl. data. unref. f Thamdrup Dalsgaard 00. classical IPT (Fig. ). A similar pattern can however be observed if the assumption of a constant ratio between NO x in the reduction zone is violated (Nielsen 199), but in theory the two error sources are distinguishable. If the assumption is violated, it is reflected in the estimated p converging to a constant value (Nielsen 199) in increasing of N-N with increasing addition. If presence of anammox is the cause of the problem, an increase in both the estimated p the of N-N will be observed with increasing NO 3 addition. However, whether or not anammox is the problem, it should be evaluated with results from slurry incubations to assess the contribution of anammox to N (Thamdrup Dalsgaard 00). If capacity for anammox is present the assumptions are valid, the results of incubations with different additions should demonstrate the following: (1) a linear dependency between the of N-N (i.e. p N + p 9 N ) the concentration of NO 3 applied () a positive correlation between the applied NO 3 concentration the of N-N estimated by the classical IPT, but not with the N-N estimated using the new equations. If both NO 3 concentration experiments slurry incubations are applied to assess the contribution of anammox to N, both procedures for estimating genuine N described in the previous sections should be compared. A full validation of the results requires that the estimate of anammox (or ra) obtained in the concentration experiment is not smaller than the estimate obtained with the alternative method. An experiment resulting in outcome is indicative of (a) absence of anammox successful application of the classical IPT, (b) presence of anammox violation of one to several assumptions of our theory, (c) presence of anammox, but a contribution of anammox to N too small to cause a significant response in p IPT, or (d) insufficient number of replicates to account for the spatial heterogeneity of the system being investigated. Results from slurry incubations assessing the contribution of anammox to N (Thamdrup Dalsgaard 00) should be used to accept or reject hypothesis a. If hypothesis a is rejected significant anammox is present but the expected response to addition is not observed, central assumptions may be violated (hypothesis b). Unlabeled NO supplied from nitrification might, for instance, be the main substrate for anammox (i.e., assumption is violated) or nitrification may be reduced by addition of (i.e., assumption 4 is violated). Both of these situations can cause lack of correlation between the of N-N estimated with the classical IPT the concentration of NO 3 applied. It is also possible that the contribution of anammox to N is too low to cause significant response in p IPT to NO 3 additions (hypothesis c). The sensitivity of the experiment can be examined as follows. According to Eq. 4, we have the following relation:. (33) We have shown how r for one incubation series could be () expressed as a function of r in another incubation series. r (1) V r where V is the ratio between the concentrations of in the water column of the two incubations. The difference in R 9 for two incubation series can then easily be expressed in terms of r ra, R 1 1 r () ( ra)( V 1) R9 1 ra ( ) ( ) 9 [ ] (34) where R 9 (1) R 9 () are the ratios between 9 N N in incubation series 1, respectively, r (1) is the ratio between in the reduction zone in incubation series 1, V is the ratio between the concentrations of in the water column of the two incubations. If r is estimated from Eq. 4 using data from incubation series 1, it is possible to evaluate the response in R 9 for different values of V thereby evaluate whether or not the contribution of anammox to N is high enough to have a significant effect. Finally, insignificant response of p IPT to addition might be due to the number of replicates being insufficient to 70

9 Anammox isotope pairing Fig. 3. Estimates of N calculated from isotope pairing (white columns) from the procedure outlined in the text (gray columns). The estimated rate of anammox is given as hatched columns. Error bars represent stard error of the mean (n 5). account for the spatial heterogeneity. The appropriate number of replicates can be assessed by power analysis. Assessment We estimated the error imposed by the presence of anammox on calculation of N-N according to the classical IPT for 4 different sites where anammox has been reported to account for 0% to 69.8% of N (Table 3). The rates of 9 N N measured after addition of NO 3 to intact cores were used as values of p 9 N p N. The reported data sets did not include measurements with different additions, Eq. was therefore used to estimate r. The measured contribution of anammox to total N in anoxic jars incubated with homogenized sediment was used as a proxy for ra (Table 3). The difference between N-N estimated by the isotope pairing technique the revised estimate suggested here was 0%,.5%, 31%, 83% of the revised estimate for Norsminde Fjord, Rers Fjord, Young Sound, the Skagerrak, respectively (Fig. 3). According to a paired t test based on the individual cores, the apparent overestimation of N was, however, only significant for the Skagerrak sediment (P 0.04) due to the variation in the recorded activity (Fig. 3). However, it is clear that if the isotope pairing technique is assumed to measure denitrification only, the technique significantly overestimates the process in the Skagerrak Young Sound sediments. The true denitrification rate is the difference between the correctly calculated p the anammox rate (Fig. 3), this rate is overestimated by several hundred percent when the classical IPT is used. The IPT has been applied extensively in temperate estuarine sediments (Nielsen et al. 001; 1999; 1998b; Rysgaard et al. 1995; Rysgaard et al. 1996; Sundback Miles 000), the question is whether these data should be revised given the possible overestimation introduced by the anammox process. As already pointed out, there is a positive correlation between the concentration of added p estimated from stard IPT equations if anammox is present. With regard to studies where experiments with different NO x additions have shown independence between the estimated N-N the concentration of NO x dependency between the of N-N NO 3 (e.g., Dong et al. 000; Nielsen Glud 1996; Rysgaard et al. 1995; Eyre et al. 00), we find no reason to question the reported denitrification rates. Data from studies where such tests have not been performed might be subject to criticism. However, the still-limited knowledge about the biogeography of anammox suggests that the process accounts for only 0% to 6% of N in estuarine environments (Thamdrup Dalsgaard 00; pers. comm. unref.). According to Eq. 13, this means that the relative error imposed by the IPT is less than 1.5% for r > 0.5 (corresponding to a N fraction in a substrate undergoing dissimilatory reduction of 80%), which is not critical. Lower r values will produce higher relative errors, but a low r also implies that the availability of is low. Thereby the of N-N (p ) is low even a high relative error will have little quantitative importance in such environments. Comments recommendations In the present communication, we have explored the limitations to application of the IPT as formulated by Nielsen (199) in sediments where denitrification anammox coexist. We conclude that the assumptions of the IPT are violated in such systems that the values of N-N determined by the technique are overestimated. However, the overestimation of N in estuarine sediments does not seem to be a serious problem. We have developed alternative procedures for estimating the of N-N from measurable variables, namely rates of N-N the contribution of anammox to N (Table ). These procedures allow quantification of distinction between anammox denitrification as well as the dependencies of these processes on NO x supplied from either the water column or nitrification. This improvement of the IPT thus reveals a detailed picture of the N cycle with only minor additional experimental effort. According to our equations, incorrectness of the IPT induced by anammox can be revealed by a simple test-incubation of 71

10 Anammox isotope pairing sediment cores with different additions of NO 3. A positive correlation between the estimated of N-N the added concentration of NO 3 is indicative of unsuccessful application of the method. In the same experiment, the true N-N calculated by the equations presented here should be independent of the N-N the concentration of NO 3. The equations are based on several assumptions that seem plausible according to present knowledge of anammox experience with the IPT. However, we cannot guarantee that the assumptions are valid for all types of sediment. We therefore recommend measuring the potential for anammox by the method of Thamdrup Dalsgaard (000), furthermore, setting up parallel experiments with different NO 3 concentrations ( 4). The results of these very simple additional experiments will expose any problems imposed on the classical IPT, furthermore, will validate the estimates of N obtained with the alternative calculation procedure suggested here. References Cabrita, M. T., V. Brotas Seasonal variation in denitrification dissolved nitrogen fluxes in intertidal sediments of the Tagus estuary, Portugal. Mar. Ecol. Prog. Ser. 0: Canfield, D. E., B. B. Jorgensen, H. Fossing, R. Glud, J. Gundersen, N. B. Ramsing, B. Thamdrup, J. W. Hansen, L. P. Nielsen, P. O. J. Hall Pathways of organic-carbon oxidation in 3 continental-margin sediments. Mar. Geol. 113:7-40. Christensen, P. B., L. P. Nielsen, N. P. Revsbech, J. Sørensen Microzonation of denitrification activity in stream sediments as studied with a combined oxygen nitrous oxide microsensor. Appl. Environ. Microbiol. 55: Dalsgaard, T., B. Thamdrup. 00. Factors controlling anaerobic ammonium oxidation with nitrite in marine sediments. Appl. Environ. Microbiol. 68: Davidsson, T. E., R. Stepanauskas, L. Leonardson Vertical patterns of nitrogen transformations during infiltration in two wetl soils. Appl. Environ. Microbiol. 63: de Beer, D Potentiometric microsensors for in situ measurements in aquatic environments, p In J. Buffle G. Horvai [eds.], In situ monitoring of aquatic systems: chemical analysis speciation. Wiley. Dong, L. F., D. C. O. Thornton, D. B. Nedwell, G. J. C. Underwood Denitrification in sediments of the River Colne estuary, Engl. Mar. Ecol. Prog. Ser. 03:109-1 Dong, L. F., D. B. Nedwell, G. J. C. Underwood, D. C. O. Thornton, I. Rusmana. 00. Nitrous oxide formation in the Colne estuary, Engl: the central role of nitrite. Appl. Environ. Microbiol. 68:0-9 Eyre, B. D., S. Rysgaard, T. Dalsgaard, P. B. Christensen. 00. Comparison of isotope pairing N-:Ar methods for measuring sediment-denitrfication assumptions, modifications, implications. Estuaries 5: Glud, R. N., O. Holby, F. Hoffmann, D. E. Canfield Benthic mineralization exchange in Arctic sediments (Svalbard, Norway). Mar. Ecol. Prog. Ser. 173: Glud, R. N., N. Risgaard-Petersen, B. Thamdrup, H. Fossing, S. Rysgaard Benthic carbon mineralization in a high-arctic sound (Young Sound, NE Greenl). Mar. Ecol. Prog. Ser. 06:59-71 Hoffmann, C. C., S. Rysgaard, P. Berg Denitrification rates predicted by nitrogen- labeled nitrate microcosm studies, in situ measurements, modeling. J. Environ. Quality 9:00-08 Kristensen, E., M. H. Jensen, G. T. Banta, K. Hansen, M. Holmer, G. M. King Transformation transport of inorganic nitrogen in sediments of a southeast Asian mangrove forest. Aquat. Microb. Ecol. : Mengis, M., R. Gachter, B. Wehrli, S. Bernasconi Nitrogen elimination in two deep eutrophic lakes. Limnol. Oceanogr. 4:-43 Mulder, A., A. A. van de Graaf, L. A. Robinson, J. G. Kuenen Anaerobic ammonium oxidation in a denitrifying fluidized bed reactor. FEMS Microbiol. Ecol. 16: Nielsen, K., N. Risgaard-Petersen, B. Somod, S. Rysgaard, T. Bergo Nitrogen phosphorus retention estimated independently by flux measurements dynamic modeling in the estuary, Rers Fjord, Denmark. Mar. Ecol. Prog. Ser. 19:5-40 Nielson, L. P Denitrification in sediments determined from nitrogen isotope pairing. FEMS Microbiol. Ecol. 86: Nielsen, L. P. R. N. Glud Denitrification in a coastal sediment measured in situ by isotope pairing applied to a benthic flux chamber. Mar. Ecol. Prog. Ser. 137: Ogilvie, B., D.B. Nedwell, R.M. Harrison, A. Robinson, A. Sage High nitrate, muddy estuaries as nitrogen sinks: the nitrogen budget of the River Colne estuary (United Kingdom). Mar. Ecol. Prog. Ser. 0:17-8 Ottosen, L. D. M., N. Risgaard-Petersen, L. P. Nielsen, T. Dalsgaard Denitrification in exposed intertidal mudflats, measured with a new N ammonium spray technique. Mar. Ecol. Prog. Ser. 09:35-4 Pind, A., N. Risgaard-Petersen, N. P. Revsbech Denitrification microphytobenthic NO 3 -consumption in a Danish lowl stream: diurnal seasonal variation. Aquat. Microb. Ecol. 1:75-84 Risgaard-Petersen, N., L. P. Nielsen, T. H. Blackburn. 1998a. Simultaneous measurement of benthic denitrification with the isotope pairing technique the N flux method in a continuous flow-through system. Water Res. 3: Risgaard-Petersen, N., S. Skarup, L. P. Nielsen Denitrification in a soft bottom lake: evaluation of laboratory incubations. Aquat. Microb. Ecol. 17:

11 Anammox isotope pairing Risgaard-Petersen, N., T. Dalsgaard, S. Rysgaard, P. B. Christensen, J. Borum, K. McGlathery, L. P. Nielsen. 1998b. Nitrogen balance of a temperate eelgrass Zostera marina bed. Mar. Ecol. Prog. Ser. 174:81-91 Rysgaard, S., P. B. Christensen, L. P. Nielsen Seasonal variation in nitrification denitrification in estuarine sediment colonized by benthic microalgae bioturbating infauna. Mar. Ecol. Prog. Ser. 16: Rysgaard, S., N. Risgaard-Petersen, N. P. Sloth Nitrification, denitrification, nitrate ammonification in sediments of two coastal lagoons in Southern France. Hydrobiologia 39:133-1 Rysgaard, S., H. Fossing, M. M. Jensen Organic matter degradation through oxygen respiration, denitrification, manganese, iron, sulfate reduction in marine sediments (the Kattegat the Skagerrak). Ophelia 55:77-91 Rysgaard, S., B. Thamdrup, N. Risgaard-Petersen, H. Fossing, P. Berg, P. B. Christensen, T. Dalsgaard Seasonal carbon nutrient mineralization in a high-arctic coastal marine sediment, Young Sound, Northeast Greenl. Mar. Ecol. Prog. Ser. 175:61-76 Steingruber, S. M., J. Freidrich, R. Gächter, B. Wehrli Measurements of denitrification in sediments with the N isotope pairing technique. Appl. Environ. Microbiol. 67: Stepanauskas, R., E. T. Davidsson, L. Leonardson Nitrogen transformations in wetl soil cores measured by N isotope pairing dilution at four infiltration rates. Appl. Environ. Microbiol. 6: Stief, P., D. De Beer, D. Neumann. 00. Small-scale distribution of interstitial nitrite in freshwater sediment microcosms: the role of nitrate oxygen availability, sediment permeability. Microb. Ecol. 43: Strous, M., J. G. Kuenen, M. S. M. Jetten Key physiology of anaerobic ammonium oxidation. Appl. Environ. Microbiol. 65: Strous, M., E. vangerven, J. G. Kuenen, M. Jetten Effects of aerobic microaerobic conditions on anaerobic ammonium-oxidizing (Anammox) sludge. Appl. Environ. Microbiol. 63: Sundback, K., A. Miles Balance between denitrification microalgal incorporation of nitrogen in microtidal sediments, NE Kattegat. Aquat. Microb. Ecol. :91-0 Svensson, J. M., A. Enrich-Prast, L. Leonardson Nitrification denitrification in a eutrophic lake sediment bioturbated by oligochaetes. Aquat. Microb. Ecol. 3: Thamdrup, B., T. Dalsgaard. 00. Production of N through anaerobic ammonium oxidation coupled to nitrate reduction in marine sediments. Appl. Environ. Microbiol. 68: Trimmer, M., D. B. Nedwell, D. B. Sivyer, S. J. Malcolm. 000a. Seasonal organic mineralisation denitrification in intertidal sediments their relationship to the abundance of Enteromorpha sp. Ulva sp. Mar. Ecol. Prog. Ser. 03:67-80 Trimmer, M., D. B. Nedwell, D. B. Sivyer, S. J. Malcolm. 000b. Seasonal benthic organic matter mineralisation measured by oxygen uptake denitrification along a transect of the inner outer River Thames estuary, U.K. Mar. Ecol. Prog. Ser. 197: Submitted April 003 Revised 4 June 003 Accepted September