Environ. Sci. Technol. 2005, 39, 867-872
Sediment Depth Attenuation of Biogenic Phosphorus Compounds Measured by 31P NMR J O A K I M A H L G R E N , * ,† L A R S T R A N V I K , ‡ A D O L F G O G O L L , § M O N I C A W A L D E B A¨ C K , † KARIN MARKIDES,† AND EMIL RYDIN‡ Department of Analytical Chemistry, Uppsala University, Box 599, 751 24 Uppsala, Sweden, Limnology, Department of Ecology and Evolution, EBC, Uppsala University, Norbyva¨gen 20, 752 36 Uppsala, Sweden, and Department of Organic Chemistry, Uppsala University, Box 599, 751 24 Uppsala, Sweden
Being a major cause of eutrophication and subsequent loss of water quality, the turnover of phosphorus (P) in lake sediments is in need of deeper understanding. A major part of the flux of P to eutrophic lake sediments is organically bound or of biogenic origin. This P is incorporated in a poorly described mixture of autochthonous and allochthonous sediment and forms the primary storage of P available for recycling to the water column, thus regulating lake trophic status. To identify and quantify biogenic sediment P and assess its lability, we analyzed sediment cores from Lake Erken, Sweden, using traditional P fractionation, and in parallel, NaOH extracts were analyzed using 31P NMR. The surface sediments contain orthophosphates (orthoP) and pyrophosphates (pyro-P), as well as phosphate mono- and diesters. The first group of compounds to disappear with increased sediment depth is pyrophosphate, followed by a steady decline of the different ester compounds. Estimated half-life times of these compound groups are about 10 yr for pyrophosphate and 2 decades for mono- and diesters. Probably, these compounds will be mineralized to ortho-P and is thus potentially available for recycling to the water column, supporting further growth of phytoplankton. In conclusion, 31P NMR is a useful tool to asses the bioavailability of certain P compound groups, and the combination with traditional fractionation techniques makes quantification possible.
Introduction The effects of eutrophication, a major environmental problem, is in many aquatic systems controlled by the availability of phosphorus (P), making investigations of compounds containing this element in aquatic systems vitally important. Phosphorus is thus a key element for primary production in lakes, and the drainage area is the main P source. However, most of the P demand of lake biota is dependent on recycling of P, either within the water column or after sedimentation. Sediment P may either be buried permanently or recycled into the water column. This makes internal loading of P from sediments a key factor during eutrophication, and the * Corresponding author phone: +(46) 018 471 3680; fax: +(46) 018 471 3692; e-mail:
[email protected]. † Department of Analytical Chemistry. ‡ Limnology, Department of Ecology and Evolution, EBC. § Department of Organic Chemistry. 10.1021/es049590h CCC: $30.25 Published on Web 12/22/2004
2005 American Chemical Society
conditions that regulate recycling of P from sediments have thus received considerable interest. The mobility of sediment P is to some extent determined by association with metals [e.g., iron (1), aluminum (2), and sulfate reduction (3)]. Fractionation procedures for P, originally adopted from agricultural science, have contributed significantly to our understanding of the distribution and mobility of various inorganic forms of P in lake sediments. In contrast, the dynamics of organic P species are still not well investigated. The non-molybdenum-reactive P (nrP), extracted with NaOH during fractionation, is not necessarily of organic nature, it also includes polyphosphate (poly-P), pyrophosphate (pyroP), and other compounds that are the result of biological transformations. Hence, biogenic P includes organic P as well as, for example, inorganic polyphosphates and pyrophosphates. Very little information exists on the character of biogenic P stored in sediments. To our knowledge, poly-P (4) and phytate (5) are the only classes of biogenic P that have been investigated in freshwater sediments. Furthermore, there is a lack of information concerning the reactivity of various organic P species in the sediments, at what rates they are transformed and made available again for biological uptake, and which forms can be considered as refractory and thus will be buried in deeper sediment layers. Provided a constant sedimentation rate and no turbation after settling, transformation rates of the different biogenic P species can be derived from sediment concentration profiles as labile compounds should only be found in the newly accumulated, younger, uppermost sediments and not in the deeper, older, sediments where the more stable compounds, being transformed only slowly during continued sediment diagenesis, should dominate (6, 7, 2). This agrees with observations in the ocean water column where P-esters are selectively mineralized and thus decline during sedimentation of organic particles, while phosphonates appear resistant and remain at constant levels throughout the water column as seen with 31P NMR (8). Accordingly, using 31P NMR, Paytan et al. (9) identified selective hydrolysis of different organic P compounds in settling marine seston. Since the 1980s,31P NMR spectroscopy is one of the most used techniques for identifying organic P compounds in environmental samples (10, 11). A limited number of studies on mainly marine sediments (e.g., refs 12 and 13, identifying phosphate esters as the dominating form of organic P in a number of sediments), lake water (14), and plankton (15) have been presented since then. 31P NMR studies on freshwater sediments are limited to Hupfer et al. (4, 16), studies focusing mainly on poly-P associated with sediment bacteria, coming to the conclusion that poly-P is more rapidly decomposed than other P-containing compounds in the sediment, and Carman et al. (13), who investigated organic P compounds in brackish and lacustrine sediments with different redox conditions, identifying pyro-P and poly-P in oxic lake sediments only and concluding that orthophosphate monoesters were the major constituent in all sediments. We combined 31P NMR and P fractionation to identify labile and recalcitrant species of biogenic P in seston and sediment profiles from Lake Erken, which has been subjected to an almost constant external rate of P loading over the last century (2). We show that, in combination with the traditionally used P fractionation, 31P NMR brings new understanding to the behavior of biogenic P in aquatic samples.
Experimental Section Study Site. Lake Erken is a moderately eutrophic [total phosphorus (TP) concentration is 27 µg L-1] lake in Sweden. VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
867
Its surface area is 24 km2, and mean and maximum depths are 9 and 21 m, respectively. During summer stratification, bottom water occasionally turns oxygen-depleted. The drainage area (137 km2) is mostly forested and consists of nutrient-rich glacial and post-glacial clay deposits. The lake is well examined and has been moderately eutrophic ever since the measurements started in 1930. This together with the low and constant population density as well as the stable farmland area support the hypothesis of a constant external P load during the last century (2). Sampling. Ten sediment cores were collected with a core sampler (Willner sampler) at a depth of 16 m within an accumulation bottom area (50 m2) of the lake (59°50′32.0′′ N, 18°37′52.5′′ E) in September 2002. The cores were sliced into 1 or 2 cm segments immediately after sampling. Segments from the same depth of all cores were pooled and homogenized in order to obtain a representative sample of sufficient size for extraction. Samples were immediately refrigerated and analyzed. Seston was collected during a cyanobacterial bloom in September 2002 in a shallow bay of Lake Erken, using a 40 µm sieve. The phytoplankton was dominated by cyanobacteria (Microcystis spp.), with some diatoms present (Asterionella formosa and Stephanodiscus spp.). Fractionation and Chemical Properties. P fractionation followed Hieltjes and Lijklema (17), with an additional BD step according to Psenner et al. (18), using pro analysi chemicals and wet samples in triplicates. This sequential procedure separates the extracted P into five groups, and extracted P is measured as molybdate-reactive P (MRP, 19). Initially, porewater P and loosely sorbed P are extracted with NH4Cl. P adsorbed to iron and manganese, sensitive to low redox potential, is then extracted by bicarbonate dithionite (BD). P-forms that are extracted with OH- are dissolved with 0.1 mol L-1 NaOH. This MRP, mainly aluminum-bound P, is classified as NaOH-rP (NaOH-reactive P). NaOH also dissolves P-forms that become molybdate reactive only after digestion following Menzel and Corwin (20). This fraction is classified as NaOH-nrP (NaOH-nonreactive P) and assumed to be organic P and bacteria-incorporated P (e.g., poly-P). As this step is expected to include most of the organic P compounds, this step is the base for using NaOH as an extractant of P for NMR studies. Finally, HCl is used to extract forms of P sensitive to low pH, presumably mostly apatite. Residual P (res-P) is calculated by subtracting NaOH-nrP and MRP identified in each fractionation step from sediment TP. The NaOH extraction delivered the only fraction containing significant amounts of nrP. Sediment TP was analyzed in duplicate after acid hydrolysis at 340 °C, according to Murphy and Riley (19), and total carbon (C) and total nitrogen (N) were analyzed on freeze-dried sediments using a Carbo-Erba analyzer. NaOH Extraction for NMR Analysis. Solutions subjected to NMR analysis were extracted with 0.1 mol L-1 NaOH, without preceding extractions with NH4Cl and BD and the potential loss of more labile fractions of P. They thus include all NH4Cl-P, BD-P, and NaOH-P. A total of 100 g of sediment was mixed with 0.1 mol L-1 NaOH at a 1:3 volume-to-volume ratio. The mixture was shaken for 16 h and centrifuged at 2100 RCF for 10 min, after which the supernatant was concentrated 40-fold by rotary evaporation except for the seston sample, where the jelly-like consistence of the sample only allowed a 20-fold concentration. NMR Studies. The NaOH extracts were mixed (4:1, v/v) with a dithionite solution in order to reduce Fe(III) to Fe(II), as the paramagnetic Fe(III) otherwise would interfere with the NMR runs (21). The influence of sample pH on the NMR spectra was tested using a variety of solutions containing standard P compounds in a pH range between 7 and 14, with the emphasis on pH 868
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005
FIGURE 1. 137Cs dating of sediment cores from Lake Erken. Note that the peak activity of the October 2004 measurements are found about 1 cm lower than that of the May 2003 measurement, confirming the 6 mm yr-1 accumulation rate. 11-13. A possible effect of different relaxation times on NMR spectras (22) was also investigated. However, neither the slight difference in pH between the samples due to the rotary evaporation nor the relaxation times in the NMR runs proved to influence the results notably. Assignment of peaks was done using standard solutions, added to one of the sediment extracts, as well as comparisons with literature. The standards used were adenosine-5monophosphate monohydrate (AMP, 99%, Aldrich Chemical Company, Inc., Milwaulkee, WI) for monoesters, RNA type IV from Torula yeast (Sigma-Aldrich Co.) for diesters, Na2HPO4‚7H2O for orthophosphate, and Na2P2O7‚10H2O for pyrophosphate. Other standards included tributyl phosphate (Analytical Services, Inc, US) and inositol hexaphosphate (40 wt % Aldrich Chemical Co.). 31P NMR spectra were recorded on a Varian XL-300 instrument at 121.4 MHz and ambient temperature. An amount of D2O sufficient to obtain a stable lock signal was added to the samples, and spectra were recorded using a 72° observe pulse, acquisition time 0.4 s, relaxation delay 0.2 s, acquiring between 70 000 and 100 000 transients (12-17 h). Chemical shifts were indirectly referenced to external 85% H3PO4 (at δ ) 0.0) via the lock signal. Data were processed using a line broadening of 5-20 Hz. To obtain peak areas, peaks in the raw spectrum were fitted with Lorentzian line shape using the deconvolution subroutine of the NMR software (Varian Vnmr version 6.1C). From these peak areas we calculated the contribution of the individual compound groups (orthophosphate, monoesters, diesters, and pyrophosphates) as compared to the TP detected in the extracts. Decay Rates of Identified Compounds. Apparent halflife times of the different P species were estimated by exponential fitting of calculated concentration versus age as derived from sediment depth. To derive the sediment age, an additional core was sampled in May 2003 at the same spot, and the 137Cs activity was recorded. The freeze-dried sediment samples were run through an Intertechnique model 4000 gamma counting system, equipped with a sodium iodine well detector to measure the 137Cs content. The peak in activity was estimated to the 5-6 cm sediment layer (Figure 1), apparently representing the Chernobyl nuclear accident in 1986. This peak was confirmed by another core sampled in October 2004, where the peak was found in the 6-7 cm sediment layer. This increase in sediment depth where the peak in activity shows could be explained by additional sediment accumulation between May 2003 and October 2004. The average deposition since 1986 was calculated to be 320 g dry matter m-2 yr-1, corresponding to an annual sedimentation of 6 mm yr-1. These findings corresponds to the results of Weyhenmeyer et al. (23), who
FIGURE 2. Phosphorus fractionations in the sediment core from Lake Erken used for NMR analysis. Error bars indicate triplicate standard deviation.
TABLE 1. Total Concentrations of Water, Carbon (C), Nitrogen (N), and Phosphorus (P) in a Seston Sample and in a Vertical Sediment Profile, Pooled from 10 Cores from 16 m Depth, from Lake Erken water C (mg N (mg P (mg C/N C/P (%) g-1 dw) g-1 dw) g-1 dw) (M) (M) seston sediment (0-1 cm) sediment (1-2 cm) sediment (2-3 cm) sediment (3-4 cm) sediment (4-5 cm) sediment (5-7 cm) sediment (10-12 cm) sediment (15-17 cm) sediment (20-22 cm) sediment (30-32 cm)
97 92 90 90 90 89 88 83 82 79 83
477 97 95 94 94 93 92 71 66 65 95
68 13 13 13 13 13 12 9 9 9 13
7.16 2.09 1.61 1.47 1.43 1.41 1.37 1.07 1.00 0.93 0.93
8.2 8.5 8.6 8.6 8.7 8.6 8.6 8.9 8.8 8.8 8.7
172 120 152 165 169 171 173 173 170 180 263
reported that the average sedimentation in Lake Erken is 1.4 mm yr-1 and that the maximum (due to sediment focusing) is 7.5 mm.
Results Physical and Chemical Properties. The TP concentration was markedly higher in the seston than in the uppermost cm of the sediment (7.2 mg of P g-1 dw as compared to 2.1 mg of P g-1 dw). TP decreased with depth throughout the sediment profile from 2.1 mg of P g-1 dw in the upper cm to 0.9 mg of P g-1 dw in deeper sediment layers (Table 1). The decrease in the TP content between seston and surface sediment can mainly be attributed to the NH4Cl-P fraction [4.5 mg g-1 dw in seston, 0.017 mg g-1 dw in surface sediment (0-1 cm)], whereas the NaOH-nrP, residual P, and BD-P fractions contributed most to the decreasing TP concentrations in the sediment (Figure 2). HCl-P was the only fraction that tended to increase with increasing sediment depth. res-P was 1.1 mg of P g-1 dw in seston and between 0.2 and 0.6 mg g-1 dw in the vertical sediment profile. The coefficient of variation of measurements of sediment P was generally low (below 10%) (Figure 2). NMR Results. All NMR-spectra show peaks in the areas for inorganic ortho-P (ortho-P, 6-7 ppm), monoester ortho-P (monoester-P, 4-5 ppm), and diester ortho-P (diester-P, 0 ppm) (Figure 3). In addition, spectra from the seston and the upper 5 cm of sediment suggest the presence of pyrophosphate (-4 to -5 ppm) and generally exhibit larger peaks and better resolution than those from deeper layers. The ortho-P and the monoester-P peaks frequently overlap, making them hard to distinguish from each other, especially in the deeper sediments. The ortho-P detected as MRP in the fractionation procedure decreased in the extracts with increasing sediment
FIGURE 3. 31P NMR spectra of 0.1 M NaOH extracts from a seston sample and a sediment profile from Lake Erken. depth and was markedly lower in sediment than in the seston sample. In contrast, ortho-P measured by NMR was higher in the upper sediment layers than in the seston sample, varying between 35 and 55% of extracted P. The monoester-P content was slightly lower, making up between 25 and 45% of extracted P from the sediment. The diester-P portion was fairly stable around 20% throughout the sediment profile. Monoester-P and diester-P constituted 60% and 1%, respectively, of the content in the seston sample. Pyrophosphate was only found in the upper centimeters of the sediments and in the seston sample (Table 2). VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
869
TABLE 2. P Composition of the Sediment Profile According to 31P NMR Analysis, Compared to MRP from Fractionationa 31P
NMR analysis
P fractionation
sediment ortho-P monoester-P diester-P pyro-P depth (cm) (%) (%) (%) (%) seston 0-1 1-2 2-3 3-4 4-5 5-7 10-12 15-17 20-22 30-32
27 56 35 32 53 52 52 34 39 44 48
62 22 39 44 26 24 26 41 39 39 33
1 17 20 20 17 23 22 25 21 17 19
10 6 6 4 4 0 0 0 0 0 0
MRP (% of TP) 88 36 26 21 19 20 18 18 24 21 15
a MRP refers to the percent of total P identified as orthophosphate extracted by NH4Cl, BD, and NaOH in P fractionation.
FIGURE 4. Concentration of phosphorus groups identified using 31P NMR in the Lake Erken sediment profile. The sum of biogenic P is equivalent to the sum of pyrophosphate and the ester compounds. The concentration of ortho-P, as derived from NMR peak areas, was about 1.6 mg g-1 dw in the seston sample and decreased from 0.7 to 0.2 mg g-1 dw with increasing sediment depth. Monoester-P concentration was above 3.5 mg g-1 dw in the seston sample and declined from 0.3 to 0.1 mg g-1 dw in the sediment profile. The concentration of diester-P was about 0.08 mg g-1 dw in the seston sample, but 0.2 mg g-1 dw in the surface sediments, and decreased slowly down through the sediment profile. Pyrophosphate was abundant in seston (0.6 mg g-1 dw) but was rapidly attenuated with increasing sediment depth and was undetectable below 5 cm sediment depth.
Discussion The results show generally decreasing concentrations of ortho-P and biogenic P (mono- and diester-P, and pyrophosphate) with increasing sediment depth (Figure 4). Together with the loss of TP (Table 1), this may reflect degradation processes over time as well as loss of sediment P to the overlaying water column. The decrease of sediment biogenic P, generally regarded to be equivalent to P extracted as NaOH-nrP during standard P fractionation (e.g., ref 18), provides the primary source of sediment P to be recycled to the water column if the generated ortho-P is not permanently adsorbed to solid phases. Quantification of these groups of compounds with NMR has to be performed carefully, given the difficulty to make an exact integration of the peak areas in NMR. Nevertheless, the general trends are clear, and our study illustrates that the method is suitable for assessing the relative proportions of the P groups. 870
9
TABLE 3. Calculated Half-Life of Biogenic Phosphorus Compounds in the Sediments of Lake Erkena
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005
monoester-P diester-P pyrophosphates
half-life (yr)
R2
p-value
23 21 13
0.67 0.89 0.80
0.002
