Identification of Dissolved Nonreactive Phosphorus in Freshwater by

Jun 18, 2009 - Little information exists on nonreactive phosphorus (nrP) in the water column, because the concentration is much lower than that in the...
28 downloads 11 Views 977KB Size
Download PDF
Environ. Sci. Technol. 2009, 43, 5391–5397

Identification of Dissolved Nonreactive Phosphorus in Freshwater by Precipitation with Aluminum and Subsequent 31P NMR Analysis KASPER REITZEL,* HENNING S. JENSEN, MOGENS FLINDT, AND FREDE Ø. ANDERSEN Institute of Biology, University of Southern Denmark, Campusvej 55, 5230, Odense M, Denmark

Received April 14, 2009. Revised manuscript received June 2, 2009. Accepted June 9, 2009.

Little information exists on nonreactive phosphorus (nrP) in the water column, because the concentration is much lower than that in the sediment. Here we present a novel method for up-concentration and identification of nrP in lake water: nrP is precipitated with poly aluminum chloride and the precipitate is subsequently recovered and dissolved by NaOH. Additional up-concentrationbyrotaryevaporationincreasesPconcentrations up to 5500 times. Furthermore, there is only a low upconcentration of paramagnetic metals. The method is sensitive and easy to use. Bottom water from five Danish lakes was sampledinautumn2008andinfourofthefivelakesorthophosphate monoesters constituted the largest fraction of nrP (50-86%), whereas DNA-P was the largest fraction in the fifth lake (67%). The pyro-P/poly-P concentration varied between 0 and 33% of nrP in the lakes. Thus, most of the P compounds usually found in lake sediments were also found in the bottom water of these lakes.

Introduction During the last decades, increasing attention has been paid to the role of nonreactive phosphorus (nrP) in freshwater systems (e.g., 1-9). Hence, nrP has been found to constitute 20-30% of the total sediment P in a variety of Danish and Swedish lakes (e.g., 10-14). However, less is known about the nrP flux from the sediment, but unpublished data (H. S. Jensen) from a laboratory study with intact sediment cores kept under anoxic conditions indicate that around 10-15% of the P efflux from anoxic sediments can be made up by nrP in some Danish lakes. Nonreactive P is determined as the difference between total dissolved P (TDP) and dissolved inorganic reactive P (rP) in water samples, measured spectrophotometrically as a molybdenum blue complex (e.g., (15)). Two decades ago, Newman and Tate (16) introduced 31 P nuclear magnetic resonance (NMR) spectroscopy to the field of soil science. Later this technique was adapted to lake sediments (e.g., 1, 8). Here alkaline extraction of P is followed by a concentration step (e.g., rotary evaporation or lyophilization) in order to obtain a sufficiently high concentration for the NMR instrument, which requires P concentrations above 50 mg L-1 depending on the variety of P compound * Corresponding author. 10.1021/es900994m CCC: $40.75

Published on Web 06/18/2009

 2009 American Chemical Society

groups in the samples. Due to the relatively low sensitivity of the NMR instrument, freshwater related research has so far been limited to sediment studies, where it is relatively easy to extract and concentrate P for the NMR analysis. Normally, a 10-fold concentration of the extracted sediment sample is enough for a NMR spectrum with sufficiently high signal-to-noise ratio. In sediment studies NMR has been used to identify polyphosphates (poly-P) (1) and organic P compound groups (4, 17), poly-P and phospholipids in sediment extracts (5), degradation rates of various organic P compound groups (8, 18), utilization of organic P by aquatic vegetation in wetlands (19), and the effects of an aluminum treatment on the composition of organic P in a Danish lake (14). Furthermore, suspended seston has also been analyzed for nrP by alkaline extraction and NMR analysis (20), but a general feature for most of the projects that have used the NMR technique is that only P extracted from particles has been studied. The need for a methodology to study dissolved nrP in the water column was addressed by Nanny and Minear (2). They used tangential ultra centrifugation and reversed osmosis to concentrate the dissolved organic P for NMR analysis. However, most of the NMR samples from these studies had seemingly very low nrP concentrations, making it difficult to interpret the various peaks in the NMR spectra. Furthermore, what we believe to be line broadening, likely due to a coconcentration of paramagnetic metals along with upconcentration of nrP, disturbed the interpretation of the NMR spectra. The method proposed by Nanny and co-workers is relatively complicated and time-consuming, and there have not been further studies published deploying or refining this method. CadeMenun et al. (21) took another approach by using lyophilization of ∼4 L water followed by extraction in 20 mL NaOH-EDTA. However, spectra presented in their article show low signal-to-noise ratios, which suggests that the obtained concentration factor is too low. CadeMenun et al. (21) did not suggest lyophilizing of a larger sample volume than the 4 L, most likely because it would be very difficult to handle such large volumes of water in the freeze drier. Furthermore, the use of lyophilization has been shown to degrade commercial P products such as poly-P and glucose 6 phosphate (21). As with the method proposed by Nanny and co-workers, the CadeMenun method will also lead to an up-concentration of metals, which may cause line broadening in the NMR spectra. Since earlier studies on lake restoration by aluminum (Al) addition have demonstrated that Al hydroxide (Al(OH)3) precipitates both rP and nrP from lake water, we here examine the possibility to use this precipitation technique for concentrating natural and added nrP products from 20 L aliquots of lake water. The Al-P precipitate is redissolved in NaOH, and subsequently analyzed by NMR. The concentration method is easy to use, cheap, and uses water samples sufficiently small for routine work. The procedure can be easily modified to larger/smaller water volumes if needed. Since the up-concentration of P is performed by adsorption to a metal hydroxide specific for P, the negative effects on the NMR analysis by potentially hampering metals such as iron (Fe), manganese (Mn), and calcium (Ca) is dramatically reduced compared to existing concentration methods.

Materials and Methods Lake water was collected from near the bottom in five Danish lakes (Lake Søholm, Lake Søbygaard, Lake Vedsted, Lake Almind, and Lake Avn) ranging from mesotrophic to hyVOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5391

FIGURE 1. Mass balances of TP, rP, and nrP when water samples from three of the studied lakes are up-concentrated and analyzed by NMR. Numbers written in the columns are the P concentrations initially and after up-concentration. Numbers above the columns indicate percentages of P precipitated by the Al and percentages of P recovered in the final NMR sample. Error bars represent standard deviations. pertrophic. Initial concentrations of TDP, rP, and nrP in Lake Søholm, Lake Søbygaard, and Lake Vedsted can be seen in Figure 1. All lakes except the shallow Lake Søbygaard were stratified at the time of sampling, and all lakes, except for Lake Almind, had anoxic bottom water. Pre-concentration of Water Samples and Replicability. Lake water was collected from the deepest part of the lakes in September-October 2008, and brought to the laboratory where it was filtered through 1.2 µm GF/C filters. Twenty liters of filtered water was used for each sample. Six hundred µL of PAX 14 (a poly aluminum chloride solution, with an active substance of 3.67 mol Al L-1, Kemira) was added to the 20 L container and thoroughly mixed to create an Al flock, and adjusted to pH between 6.5 and 7. It should be noted that the Al:P molar ratio used in this study is calculated from TDP in the lake water applying a 10:1 Al:P molar ratio for precipitation. The Al solution can be substituted by any Al product forming an Al flock. The container was left overnight with a resulting precipitate of P and Al on the bottom of the container. The majority of the lake water was then removed by siphoning and the remaining water and Al floc containing P (∼300 mL) was collected and centrifuged for 10 min at 4000 rpm to isolate the flock from the residual lake water. After centrifugation the Al flock was redissolved in 20 mL of 1 M NaOH for 18 h to release P from the Al-P complex. By this process the original P concentration in the water sample was concentrated 1000 times. The NaOH sample was then concentrated by rotary evaporation at 35 °C to obtain an ∼4000 times higher concentration than in the original lake water sample. Finally, the sample was centrifuged for 10 min in a Sigma high speed centrifuge at 59 860g to remove any precipitate, resulting in a loss of P less than 5%. Prior to 31P NMR analysis 70 µL of D2O was added to 630 µL of sample. In some of the lakes there was a degree of line broadening, which could be overcome by adding 50 µL of 0.5 M tetrasodium-ethylenediaminetetraacetate (Na-EDTA) to the final NMR sample. This was subsequently done to all of the NMR samples to ensure comparability. 5392

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

To observe replicability and whether there was a potential selective retention/loss of any specific P compounds by the Al flock, a study was conducted with addition of selected standard P compounds to artificial lake water. The replicability was tested on three replicates of 20 L of lake water from Lake Vedsted. Elemental Analyses. Samples from Lake Søholm, Lake Vedsted, and Lake Søbygaard were studied in more detail with respect to recovery of TDP, rP, and nrP during the procedure. Triplicate subsamples were collected before and after addition of Al, and also the final NMR samples after the centrifugation step were analyzed in triplicate. These samples were analyzed for TDP, rP, and nrP by the standard molybdenum blue method (15). Nonreactive phosphorus was determined as the difference between TDP and rP. Total Fe, Ca, and Mn was determined by inductively coupled plasma (ICP-OES) spectroscopy on a Perkin-Elmer Optima 2100 DV Instrument. Modifications of the Method. To test the up-concentration method on two different water sample volumes, phosphorus from 20 and 40 L of filtered water from Lake Vedsted was precipitated and concentrated, and the resulting spectra were compared. To demonstrate the usefulness of the method for smaller lab scale experiments, we concentrated phosphorus compounds from 4.4 L of water from Lake Vedsted. This water was collected from another experiment where mineralization of added IP6 in the sediment was followed. At the end of the experiment we sampled and upconcentrated the overlying water according to the proposed method, and NMR analysis was conducted. Addition of Standard P Compounds. Inositol hexakisphosphate (IP6) (phytic acid dipotassium, Sigma-Aldrich), pyrophosphate (pyro-P) (sodium pyrophosphate decahydrate, Sigma-Aldrich), and poly-P (tripolyphosphate, SigmaAldrich) were added to the lake water to follow the precipitation by Al and recovery in the NMR extract. In the test with artificial lake water, a 2 mM NaHCO3 solution was prepared to mimic the average alkalinity in Danish lakes. The three

standard P compounds were added in concentrations of about 1.5 mg P/L, then Al was added and pH was adjusted to 7 with NaOH. The relatively high P concentrations were used to save expensive NMR time, since the samples could then be run for about 1 h instead of 17 h as for the lake samples. Furthermore, we believe that any hampering effects of the up-concentration method observed at high P concentration would be expected to take place at lower P concentrations as well, which is supported by the findings in the experiment with IP6 to the sediment of Lake Vedsted, where a much lower IP6 concentration was up-concentrated. The solutions were thoroughly mixed and left overnight for flock formation. Then the flock was removed and redissolved in 20 mL of 1 M NaOH and the general procedure was followed. 31 P NMR Analyses. The 31P NMR spectra were recorded at 80.9 MHz on a Varian 200 MHz NMR spectrometer at ambient temperature. Spectra were recorded using a 90° observe pulse, acquisition time 0.4 s, and relaxation delay 1.5 s, acquiring around 32 000 transients (17 h). Chemical shifts were indirectly referenced to external 85% H3PO4 (at δ ) 0.0) via the lock signal. The NMR utility transform software (NUTS) was used to obtain peak areas from the raw spectrum. The areas of the peaks are proportional to the amount of the P binding groups. Thus, the percentage distribution of the single P compound groups can be calculated from the integrals of the peaks. By relating the TP in the extracts measured by, e.g. a colorimetric procedure or ICP, to the percentage distribution of the P compound groups the concentration of the individual P groups can be calculated. Spectra were plotted with a line broadening of 5 Hz. Peaks were assigned by comparisons with literature (e.g., (22)) and by comparison with added standard P compounds.

Results and Discussion By using Al to precipitate dissolved nrP from water samples, we successfully managed to develop a method suitable for sample preparation to NMR analysis. This method has the potential to resolve some of the missing links in the complicated P cycle. For instance, it should now be possible to link nrP changes in sediments to the accumulation of nrP in the overlaying water column, thereby gaining further insight into the P release from anoxic sediments. Compared to previously suggested methods, NMR spectra with high signal-to-noise ratio and narrow peaks were generally obtained when applying this method to 20 L aliquots of lake water (Figure 2), allowing identification of distinct P groups, such as mono-P, DNA-P, and poly-P as well as single P compounds such as orthophosphate, pyrophosphate, and IP6. The dissolved nrP in the water column from the five lakes that we studied was generally composed of P compound groups normally found in sediments. Mass Balances of P, Fe, Mn, and Ca. To assess the replicability of the method, we conducted a study with three replicate water samples of 20 L from Lake Vedsted, and compared the standard deviation of recovery in the NMR sample as well as the distribution of P groups (Table S1). The low standard deviation in all measurements demonstrates that the water samples were homogeneous and that the method produces reproducible results. Therefore, we consider our results on single samples of 20 L of lake water (Figures 1 and 2) to be representative. Lake Søbygaard was characterized by much higher TDP (531 µg/L) and rP (513 µg/L) concentrations than Lake Vedsted and Lake Søholm (∼170 µg/L TDP and ∼140 µg/L rP) (Figure 1). However, the nrP concentrations were very similar in the three lakes, ranging from 19 to 24 µg/L. The recovery of nrP in the final NMR sample was between 86% and 100% of the initial nrP (Figure 1). In Lake Søholm the

FIGURE 2. lakes.

31

P NMR spectra of lake water from five Danish

up-concentration resulted in a final sample concentration of 131 mg nrP/L, which corresponded to a 5458 times upconcentration (calculated from Figure 1). For Lake Søbygaard VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5393

a final nrP concentration of 86.5 mg/L was reached, corresponding to a 4550 times concentration. Finally, Lake Vedsted was up-concentrated to 65 mg/L, corresponding to a 3100 concentration factor. In Lake Vedsted and Lake Søholm the lower than 100% recovery of nrP was the result of an incomplete precipitation with Al. We subsequently conducted a test on water from Lake Vedsted and found that a complete precipitation could be obtained by adding Al in two steps with half a dose in each step instead of one step (Figure S1). However, there was no difference between the spectra from Lake Vedsted where 86% was precipitated and the optimized spectra where 100% was precipitated (spectra not shown), indicating that it was not a single P phase that was lost, but rather an evenly distributed loss from all the P phases. In addition to the partition of the Al dosage into two dosages, we also tested the effect of time for precipitation (Figure S1) and the effect of doubling the Al dose on the NMR analysis. The results showed that 2 h of precipitation is not sufficient for binding all the nrP, whereas 18 h was sufficient. Furthermore, the higher concentration of Al did not have any effect on the NMR analysis, as expected (spectrum not shown), which allows for the use of a higher Al concentration in cases where more nrP needs to be precipitated. For rP and TDP the final recovery in the NMR extract was in general lower than the recovery of nrP, except from Lake Søholm where all of the rP and TDP was recovered. These results demonstrate that the use of Al to precipitate nrP is highly efficient, and it actually recovers a higher percentage of nrP than of TDP and rP. This finding is in contrast to the findings of Browman et al. (23) who indicated that Al will not bind dissolved organic P. Fe, Mn, and Ca. Similar to the P mass balances, balances of Fe, Mn, and Ca were made (Figure S2). Less than 8% of Fe and less than 7% of Mn were recovered in the final NMR extracts, whereas basically no Ca was recovered. This is an important feature of the method, since both Fe and Mn are paramagnetic ions (24), which may cause line broadening in the NMR spectra, whereas Ca may play a role as a catalyst in the breakdown of poly-P (20). This specificity toward nrP has not been demonstrated in prior attempts to upconcentrate P in lake water (2, 21), and is a major benefit of the Al-precipitation method. However, despite the much lower concentration factor of the metals, the concentration of paramagnetic metals was still high enough for interference in some of the NMR samples. In Figure S3, the spectra from Lake Søholm, Lake Avnsø, and Lake Vedsted are shown with and without addition of EDTA to the NaOH extracts. In these lakes there is an obvious effect of the paramagnetic ions on the ortho-P peak which is very broad, covering a major part of the mono-P region. However, this negative effect can be overcome by adding EDTA to the samples as seen in Figure S3. But in the case of Lake Vedsted, the addition of EDTA did not improve the spectra significantly. Another factor that may contribute to line broadening could be adsorption of nrP onto colloids and macromolecules, as mentioned by Nanny and Minear (2), but we cannot conclude whether these factors led to some degree of line broadening in the sample from Lake Vedsted. However, the problem was not a general problem for the lakes studied in this project. Modifications of the Method. The fact that the present method is sufficiently sensitive for nrP concentrations down to 15 µg/L (Lake Almind, data not shown) enables us to detect changes in the organic P pool in most freshwater systems. Even in oligotrophic systems with lower dissolved P concentrations, it may be possible to adapt the method by simply increasing the volume of water. Thus, Vicente et al. (25) observed that phosphate would bind to freshly formed Al(OH)3 by a 1:50 molar ratio at concentrations as low as 9 µg/L. A test with a larger volume of water from Lake Vedsted is shown in Figure S4 where we compared the spectra 5394

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

obtained from 40 L lake water to that obtained from 20 L lake water. There was no difference in the P distribution in the two spectra, but the signal-to-noise ratio is higher in the spectra obtained from 40 L lake water, reflecting the higher nrP concentration in this sample. As a consequence, this somehow allows the user to decide the detection limit. Similarly, the method can easily be adjusted to a lower water volume, if higher nrP concentrations are found, as seen in the case of the IP6 test described later (Figure 3g). Addition of Standard P Compounds. The test performed with addition of commercial P compounds (Figures 3 and 4) demonstrated that Al precipitated both organic nrP (monoP) and inorganic nrP (poly-P and pyro-P) completely. Furthermore, the recovery of the specific P compounds in the final NMR sample was 100% for IP6, 97% for Pyro-P, and 81% for poly-P, which demonstrate the high efficiency of Al in precipitating P. Prior to this study, the effectiveness of Al as an immobilizer of P has mainly been described in relation to rP (e.g., (25)). However, our results indicate that organic P in lake water might adsorb even stronger to Al than does ortho-P. This finding has not been described in relation to lake restoration, where Al has been used, although it is wellknown that some organic P compounds such as, e.g., IP6 might bind stronger to FeOOH than does ortho-P (7). A similar mechanism could play a role in the binding of nrP to the Al(OH)3 floc. As can be seen in Figure 3c and d, the proposed method did not alter the poly-P structure much. Thus, the three signals in the poly-P middle group region (furthest to the right in the spectra) were still present after upconcentrating the sample, indicating that this method does not affect the chain length of poly-P, which most likely can be ascribed to the use of rotary evaporation as a concentration tool instead of lyophilization. As shown in Figure 3d and h, the use of lyophilization for preparation of NMR samples causes complete loss of poly-P while the signal remains after rotary evaporation. Also CadeMenun et al. (21) mention that the use of lyophilization might degrade certain P compounds. We conclude that lyophilization should be avoided even if it is an efficient technique for concentrating samples. The signals of IP6 and pyro-P remained, both after lyophilization and rotary evaporation (Figure 3a, b, and e, f). Figure 4 shows increase in rP concentration in the poly-P NMR sample, suggesting that some hydrolysis of poly-P to ortho-P occurred during preparation by rotary evaporation. However, when using the NMR analysis to calculate the amount of ortho-P in the poly-P sample there is only a 4% increase in the ortho-P concentration after concentrating the poly-P sample. This corresponds to only 22% of the hydrolysis determined by the molybdenum blue method. We therefore suggest that the decline in poly-P measured as nrP in the final concentrated sample, by the molybdenum blue method, was due to hydrolysis caused by the strong acid used in the molybdenum blue mixed reagent. This potential hydrolysis has been described before (e.g., 26-29), supporting the finding that the acid environment required for rP analysis results in hydrolysis of poly-P. Compared to the original spiked samples (Figure 4), neither IP6 nor pyro-P was affected by the present concentration procedure. An additional test on water from Lake Vedsted was conducted to verify the detection of much lower amounts of IP6 compared to the spiked samples, and also to demonstrate that a much smaller water volume can be used in this method. In this experiment, IP6 was added to the sediment to follow its mineralization and possible release to the water above the sediment. After 6 months of incubation under anoxic conditions the concentration of nrP in the water was 60 µg/L and we wanted to identify the composition of the nrP. A water sample of only 4.4 L was collected and up-concentrated according to the proposed method. The NMR result are shown in Figure

FIGURE 3. Spectra from commercial P compounds: inositol hexakisphosphate (IP6), polyphosphate (Poly-P), and pyrophosphate (Pyro-P). (a, c, and e) Spectra from samples dissolved in 1 M NaOH with up-concentration by rotary evaporation. (b, d, and f) the same samples without up-concentration. (h) Poly-P sample dissolved in 1 M NaOH and up-concentrated by lyophilization. (g) Spectrum for water from the experiment with Lake Vedsted sediment spiked with IP6. 3g, where a very clear signal from IP6 appears, but also three signals in the region of pyro-P/poly-P end groups are identified. This illustrates the versatility and flexibility of this method, since we managed to identify specific P compounds in only 4.4 L of water. P Compound Groups. In all lakes the signal from ortho-P was dominant (Figure 2), which was expected since rP constituted more than 83% of TDP in the samples from the different lakes. Table S2 shows the percentage of individual

nrP compound groups from the 5 lakes. A general feature is that mono-P comprises the largest fraction of the dissolved nrP in the lakes, which is rather similar to the distribution normally found in sediment extracts (e.g., (14)). The only exception to this finding is Lake Almind, where DNA-P makes up the largest fraction of the nrP. Whether this is related to the fact that Lake Almind was the only lake with oxic bottom water at the time of sampling cannot be concluded from this study. VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5395

FIGURE 4. Mass balances of TDP, rP, and nrP when the three commercial compounds, inositol hexakisphosphate, polyphosphate, and pyrophosphate, undergo the up-concentration and NMR analysis. Figures above the columns are percentages of P precipitated by the Al and percentages of P recovered in the final NMR sample. In the case of inositol hexakisphosphate, the concentration of rP was below the detection limit. In general, only one signal in the region of mono-P was seen. Nevertheless, there was a tendency for two additional signals in the mono-P region in the sample from Lake Almind, although the concentration was too low for quantification. The mono-P signal at 5.183 ppm was found in both Lake Søbygaard and Lake Vedsted, whereas Lake Søholm and Lake Avn both had a mono-P peak at 5.002 ppm. The mono-P signal from Lake Almind was located at 5.063 ppm. Based on the work of Turner et al. (22) and our own identification it was not possible to further identify these signals. In the case of DNA-P, only one major peak was observed in each sample. Only Lake Almind and Lake Søholm had similar signals (-0.187 ppm), whereas the other lakes had DNA-P signals at 0.537 ppm (Lake Søbygaard), -0.248 ppm (Lake Vedsted), and -0.067 ppm (Lake Avn). In addition to the signals from mono-P and DNA-P, Lake Vedsted, Lake Søholm, and Lake Søbygaard all had signals in the region of pyro-P/poly-P end groups (Figure 2, Table S2). Since there was no signal in the region of poly-P middle groups (around 20 ppm) we believe that the signals in the region of pyro-P/poly-P end groups are mainly from pyroP. The presence of pyro-P compounds could reflect the land-use in the catchment, where agriculture often contributes the largest share. Hence, fertilizers could be the source of the pyro-P peak as proposed by Sundareshwar et al. (30). Not much is known about the ecological role of the various P compound groups. However, previous studies (8, 9, 18) have revealed differences in degradation rates between the individual P compound groups. Thus, it is generally accepted that poly-P and pyro-P constitute a P pool which is rapidly mineralized, whereas a major part of the mono-P are more recalcitrant in nature, due to their high charge density (31). However, the scope of this study was not to gain detailed information on the 5396

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009

ecological role of each P compound group, but rather to provide a method which subsequently can be used to enhance the knowledge in this field. In conclusion, we find that the combination of Alprecipitation and rotary evaporation provides reliable and reproducible results with a recovery of nrP close to 100%. This opens for more detailed studies of the role and turnover of dissolved organic P-compounds in freshwater ecosystems in the future.

Acknowledgments We thank Dr. Martin Søndergaard (NERI) for providing chemical data on Lake Avn and Lake Almind. We thank Dr. Paul Stein (University of southern Denmark) for technical assistance with the NMR analysis. The study was supported by Villum Kann Rasmussen Centre of Excellence: Centre for Lake Restoration (CLEAR).

Supporting Information Available Four figures, two tables, and information about P compound groups, Al dose, extraction times, replicability of the method, effects of EDTA on line broadening, and recovery of metals. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Hupfer, M.; Ga¨chter, R.; Ru ¨ egger, H. Polyphosphate in LakeSediments - P-31 NMR-Spectroscopy as a Tool for Its Identification. Limnol. Oceanogr. 1995, 40 (3), 610–617. (2) Nanny, M. A.; Minear, R. A. Characterization of soluble unreactive phosphorus using P-31 nuclear magnetic resonance spectroscopy. Mar. Geol. 1997, 139 (1-4), 77–94. (3) Rydin, E.; Brunberg, A. Seasonal dynamics of phosphorus in Lake Erken surface sediments. Arch. Hydrobiol. Spec. Issues

Advanc. Limnol. Beih. Ergebn. Limnol. 1998, 51, 157–167. (4) Carman, R.; Edlund, G.; Damberg, C. Distribution of organic and inorganic phosphorus compounds in marine and lacustrine sediments: a P-31 NMR study. Chem. Geol. 2000, 163 (1-4), 101–114. (5) Selig, U.; Hu ¨ bner, T.; Michalik, M. Dissolved and particulate phosphorus forms in a eutrophic shallow lake. Aquat. Sci. 2002, 64, 97–105. (6) Watts, E. E.; Dean, P. A. W.; Martin, R. R. 31P nuclear magnetic resonance study of sediment microbial phospholipids. Can. J. Anal. Sci. Spectrosc. 2002, 47, 127–133. (7) Turner, B. L.; Frossard, E.; Baldwin, D. S. Organic phosphorus in the environment; CABI Publishing: Wallingford and Cambridge, 2005. (8) Ahlgren, J.; Rydin, E.; Gogoll, A.; Waldeba¨ck, M.; Markides, K.; Tranvik, L. Depth attenuation of biogenic phosphorus compounds in lake sediment measured by 31P-NMR. Environ. Sci. Technol. 2005, 39 (3), 867–872. (9) Reitzel, K.; Ahlgren, J.; Gogoll, A.; Rydin, E. Effects of aluminum treatment on phosphorus, carbon, and nitrogen distribution in lake sediment: A P-31 NMR study. Water Res. 2006, 40 (4), 647– 654. (10) Rydin, E. Potentially mobile phosphorus in Lake Erken sediment. Water Res. 2000, 34, 2037–2042. (11) Reitzel, K.; Hansen, J.; Jensen, H. S.; Andersen, F. Ø; Hansen, K. S. Testing aluminum addition as a tool for lake restoration in shallow, eutrophic Lake Sonderby, Denmark. Hydrobiologia 2003, 506 (1-3), 781–787. (12) Hansen, J.; Reitzel, K.; Jensen, H. S.; Andersen, F. Ø. Effects of aluminum, iron, oxygen and nitrate additions on phosphorus release from the sediment of a Danish softwater lake. Hydrobiologia 2003, 492 (1-3), 139–149. (13) Jensen, H. S.; Caraco, N.; Hansen, J.; Christensen, K. K. Humicbound phosphorus in soil and sediment. In Phosphate in Sediments; Golterman, H. L., Serrano, L., Eds.; Backhuys Publishers: Leiden, 2005. (14) Reitzel, K.; Ahlgren, J.; Gogoll, A.; Jensen, H. S.; Rydin, E. Characterization of phosphorus in sequential extracts from lake sediments using P-31 nuclear magnetic resonance spectroscopy. Can. J. Fish. Aquat. Sci. 2006, 63 (8), 1686–1699. (15) Koroleff, F. Determination of nutrients. In Methods of Seawater Analysis; Grasshof, K. ; Ehrhardt, M., Kremling, K. Eds.; Verlag Chemie: Weinheim, 1983. (16) Newman, R. H.; Tate, K. R. Soil phosphorus characterisation by 31 P nuclear magnetic resonance. Commun. Soil Sci. Plant Anal. 1980, 11, 835–842. (17) Baldwin, D. S. The phosphorus composition of a diverse series of Australian sediments. Hydrobiologia 1996, 335 (1), 63–73.

(18) Reitzel, K.; Ahlgren, J.; DeBrabandere, H.; Waldeback, M.; Gogoll, A.; Tranvik, L.; Rydin, E. Degradation rates of organic phosphorus in lake sediment. Biogeochemistry 2007, 82 (1), 15–28. (19) Pant, H. K.; Reddy, K. R.; Dierberg, F. E. Bioavailability of organic phosphorus in a submerged aquatic vegetation-dominated treatment wetland. J. Environ. Qual. 2002, 31 (5), 1748–1756. (20) Hupfer, M.; Rube, B.; Schmieder, P. Origin and diagenesis of polyphosphate in lake sediments: A P-31-NMR study. Limnol. Oceanogr. 2004, 49 (1), 1–10. (21) CadeMenun, B. J.; Navaratnam, J. A.; Walbridge, M. A. Characterizing dissolved and particulate phosphorus in water with 31 P nuclear magnetic resonance spectroscopy. Environ. Sci. Technol. 2006, 40, 7874–7880. (22) Turner, B. L.; Mahieu, N.; Condron, L. M. Phosphorus-31 nuclear magnetic resonance spectral assignments of phosphorus compounds in soil NaOH-EDTA extracts. Soil Sci. Soc. Am. J. 2003, 67, 497–510. (23) Browman, M. G.; Harris, R. F.; Armstrong, D. E. Interaction of Soluble Phosphate with Aluminium Hydroxide in Lakes; Technical report 77-05; Water Resources Center, University of Wisconsin: Madison, WI, 1977. (24) CadeMenun, B. J.; Liu, C. W.; Nunlist, R.; McColl, J. G. Soil and litter phosphorus-31 nuclear magnetic resonance spectroscopy: Extractants, metals, and phosphorus relaxation times. J. Environ. Qual. 2002, 31, 457–465. (25) de Vicente, I.; Huang, P.; Andersen, F. Ø.; Jensen, H. S. Phosphate adsorption by fresh and aged aluminium hydroxide. Consequences for lake restoration. Environ. Sci. Technol. 2008, 42, 6650–6655. (26) Rigler, F. H. Further observations inconsistent with the hypothesis that the molybdenum blue method measures orthophosphate in lake water. Limnol. Oceanogr. 1968, 13, 7–13. (27) Stainton, M. P. Errors in molybdenum blue methods for determining orthophosphate in freshwater. Can. J. Fish. Aquat. Sci. 1980, 37, 472–478. (28) Baldwin, D. S. Reactive “organic” phosphorus revisited. Water Res. 1998, 32, 2265–2270. (29) Hudson, J. J.; Taylor, W. D.; Schindler, D. W. Phosphate concentrations in lakes. Nature 2000, 406, 54–56. (30) Sundareshwar, P. V.; Morris, J. T.; Pellechia, P. J.; Cohen, H. J.; Porter, D. E.; Jones, B. C. Occurrence and ecological implications of pyrophosphate in estuaries. Limnol. Oceanogr. 2001, 46 (6), 1570–1577. (31) Celi, L.; Lamacchia, S.; Marsan, F. A.; Barberis, E. Interaction of inositol hexaphosphate on clays: Adsorption and charging phenomena. Soil Sci. 1999, 164 (8), 574–585.

ES900994M

VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5397