Environ. Sci. Technol. 2002, 36, 3519-3524
Adsorption-Desorption of Phosphate on Airborne Dust and Riverborne Particulates in East Mediterranean Seawater G A N G P A N , * ,†,‡ M I C H A E L D . K R O M , § A N D BARAK HERUT# State Key Laboratory of Environmental Aquatic Chemistry, Research Center for Eco-environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China, and Eco-environmental Chemistry Laboratory, Qingdao University of Science and Technology, Qingdao 266042, China, and School of Earth Science, University of Leeds, Leeds LS2 9JT, UK, and I.O.L.R., National Institute of Oceanography, Haifa 31080, Israel
The potential importance of adsorption-desorption behavior of phosphorus (P) on the East Mediterranean (E. Med) P cycle was investigated. Contrasting adsorption behavior between Saharan dust (SD) and Nile particulate matter (Nile PM) was observed. SD was a source of P to the region, which released an average of 3.3 ( 0.3 µmolP/g into the surface seawater and showed no adsorption ability under the conditions close to the E. Med deep water. Saharan dust is therefore unlikely to be the reason for P limitation in the region. By contrast, Nile PM acted dual roles of a sink and source of P in different waters (surface seawater, deep seawater, and river water). A new crossovertype adsorption-desorption model explained the contrasting adsorption behavior and the dual nature of natural particles. The model indicates that when natural particles are transported between different waters, they can be a sink (adsorption) or a source (desorption) of phosphorus depending on the “specific concentration (λ)”, which is the ratio between the aqueous P concentration and the zero equilibrium P concentration (EPC0). EPC0 refers to the solute concentration value where the adsorption isotherm crosses over the aqueous concentration axis. When λ > 1, adsorption occurs, whereas when λ < 1, desorption occurs. The model added a general development to the methodology of adsorption isotherm, where, for the first time, effects of solute concentration, solid concentration, and aqueous medium (EPC0) on the adsorption and desorption of P in natural waters were simultaneously described by a single equation. Using the model, it was quantitatively reconstructed that particles emitted during the pre-1964 Nile floods could be a major source of P to Egyptian coastal waters (up to 4800 tonsP/yr), greater than the dissolved P flux (∼3200 tonsP/yr), but a trapper of dissolved phosphate in E. Med deep waters.
* Corresponding author phone: +(86) 10 62943436; fax: +(86) 10 62923543; e-mail:
[email protected]. † Chinese Academy of Sciences. ‡ Qingdao University of Science and Technology. § University of Leeds. # I.O.L.R., National Institute of Oceanography. 10.1021/es020516d CCC: $22.00 Published on Web 07/20/2002
2002 American Chemical Society
Introduction The Eastern Mediterranean is one of the most oligotrophic areas of the world’s oceans. It is also considered to be phosphorus limited (1, 2). Krom et al. (1) hypothesized that this unusual nutrient limitation was due to the removal (adsorption) of P, but not nitrate, by inorganic particles, since E. Med had one of the highest fluxes of atmospheric dust to the sea surface in the world (3, 4). The other major source of inorganic particles to the E. Med was, until 1964 when the Aswan Dam was completed, riverborne Nile PM. If the hypothesis were correct, these particles would have been a sufficiently large sink of P to the region. Whether this hypothesis is correct or not remains a question as it is difficult to reconstruct the role of pre-1964 Nile PM. It is also not clear how the biogeochemical cycling of P in the region is changed as the result of the dam. Here, we developed a new adsorption-desorption model and experimental design to study the relative impact of SD and Nile PM on P cycle in the region. Until now little is known on the adsorption-desorption of P on atmospheric dust in seawater (5). The effect of riverborne PM on the dissolved inorganic phosphorus (DIP) in estuaries has been investigated in much more detail, but results are diverse (6). Some papers suggest that much of the adsorbed phosphate is released to solution when riverine PM meets estuarine waters (7, 8), while others suggest that removal of DIP occurs (9, 10). In some estuaries buffering of DIP, i.e., concentration of P remains relatively constant, has been reported (11, 12). So far, there are no physicochemical theories that can be used to quantitatively assess the role of natural PM as a sink or source of P in estuarine and sea waters. For phosphate adsorption in natural waters, sediments, and soils, adsorption isotherms often cross over the aqueous concentration axis. This is because natural particles often contain native adsorbed phosphorus (NAP) before they are used for the experiment. The concentration value where the isotherm crosses the aqueous concentration axis is termed “zero equilibrium phosphate concentration” (EPC0 (13, 14)). It represents the resulting aqueous P concentration when a solid containing sorbed P is added to water containing zero initial P. Recently EPC0 has been used to assess P retention capacity in soils (15) and bioavailable P in estuarine waters (16). However, little discussion on the physicochemical meaning of EPC0 is found in the literature. It is not well recognized that EPC0 is a critical parameter that can be used to distinguish the role of source or sink of P in natural waters and soils. No adsorption isotherm equation is available to describe such a crossover-type adsorption isotherm where EPC0 is incorporated. An adsorption isotherm is the traditional way of describing equilibrium characteristics of adsorption and has been regarded as independent of the solid concentration (Cs) and the kinetics of an adsorption process. However, over the last 20 years, an anomalous phenomenon that the isotherm changes with Cs has been observed in many natural water systems (17, 18). This phenomenon cannot be explained by classical thermodynamic theory of adsorption. So far, most researchers attribute the effect to a variety of experimental artifacts, such as the nonsettling colloids, aggregation of solids, competitive adsorption, or particle-particle interaction (19-22). However, there are still other observations of the Cs effect that cannot be explained by experimental artifacts (23). Recently, Pan and Liss found a fundamental deficiency in the theoretical foundation of adsorption thermodynamics VOL. 36, NO. 16, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3519
that adsorption density Γ (mol/m2) had been incorrectly used as a thermodynamic state variable in the past (24). Once this deficiency is removed from the existing theoretical foundation, the new theory, i.e., the metastable-equilibrium adsorption theory (MEA theory), does not exclude such a Cs effect (24, 25). According to MEA theory, Γ is not a thermodynamic state variable. The chemical potential of a given Γ depends on the metastable-equilibrium state of the adsorbed molecules and hence the history of an adsorption process. An adsorption isotherm, when expressed in terms of Γ, is therefore fundamentally affected by the kinetics of adsorption processes. By affecting MEA state or adsorption reversibility/rate, solid concentration can fundamentally influence adsorption isotherms. After the kinetic process of adsorption is finished, or a certain MEA state achieved, changes in Cs (e.g. centrifugation) will have no physicochemical effect on the adsorption isotherm. Here, we will use the theory to develop a new crossover-type adsorption isotherm equation, where effects of solute concentration, solid concentration, and aqueous medium (EPC0) on the adsorption and desorption of P can be simultaneously described by a single equation. We will use the model to explain the contrast adsorption-desorption behavior between a SD and a Nile PM samples, the dual role of Nile PM as a source and sink of P in different waters, and the unlikeness for SD being the reason for the P limitation in the region. A quantitative reconstruction on the role of Nile PM during the pre-1964 floods is discussed, which may be important for the understanding of environmental changes caused by the Aswan Dam.
Theoretical Section Crossover-Type Adsorption Isotherm. According to the MEA theory, a metastable equilibrium inequality exists for solidliquid and gas-solid surface adsorption reactions when adsorption density is not treated as a thermodynamic state variable (24):
Kreal ) Kme × Keq
traditional reaction rate laws, an increase in Cs (a concentration of reactant) may increase the rate of the reaction and hence reduce the thermodynamic reversibility (Kme) of the process. By using a semiempirical assumption, Kme ) γ × C-n s , a Freundlich-type Cs effect isotherm equation can be obtained (32) β Γ ) Ksd × C-n s × C
where Γ is adsorption density (µmol/g), C is the equilibrium concentration of solute in solution, and Cs is solid concentration (mg/L). Ksd is specific adsorption coefficient, which represents the equilibrium characteristic of an adsorption reaction and is independent of solute and solid concentrations. n and β are empirical constants. n is called the Cs effect index, which is a measure of the degree of Cs effect. For phosphate adsorption in natural waters, the adsorption density (Γ) contains two parts that behave differently (14)
Γ ) NAP + Psor
β Psor ) Ksd × C-n s × C - NAP
Γ ) R × Kme × C
β
β NAP ) Ksd × C-n s × (EPC0)
9
(8)
For a given isotherm under fixed Cs condition,
Ksd × C-n s )
NAP (EPC0)β
(9)
Replace eq 9 into eq 7 and rearrange:
Psor ) NAP
[
]
Cβ -1 (EPC0)β
(10)
C EPC0
(11)
Define:
λ)
Psor ) NAP(λβ - 1)
(12)
where λ is called “specific concentration” which is the ratio between P concentration and EPC0. For isotherms under different Cs conditions, replace eq 8 into eq 12 and rearrange: β β Psor ) Ksd × C-n s × (C - EPC0 )
(13)
(4)
Equation 4 is a theoretical equation under the abovementioned theoretical assumption. R and β are constant under isothermal conditions, which can be calculated from the energy distribution of adsorption sites. Equation 4 indicates that adsorption isotherm declines to lower Γ values as Kme (e.g. adsorption reversibility) decreases. According to 3520
(7)
At EPC0 where C ) EPC0, Psor ) 0,
Kreal e Keq (< for irreversible process, ) for reversible process) (3) Keq is the equilibrium constant of an ideal reversible adsorption process. Kreal is the experimentally measured metastable equilibrium constant for a real adsorption reaction. Kme is metastable equilibrium coefficient, which is a measure of the deviation of a MEA state from the ideal equilibrium state. Equations 1 and 2 are obtained through rigorous thermodynamic deduction from first principles (24). Therefore, eq 3, the metastable-equilibrium inequality, represents a general principle in surface adsorption. Based on eq 1, by assuming an exponential distribution of adsorption energy, a Freundlich-type metastable isotherm equation can be obtained (for derivation, see the Appendix of ref 24):
(6)
where NAP is the “native adsorbed P” in natural particles, which is desorbable when P concentration is below EPC0. Psor is the amount of P adsorbed during the experiment (µmol/g). Replace eq 6 into eq 5:
(1)
0 e Kme e 1 (< for irreversible process, ) for reversible process) (2)
(5)
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 16, 2002
Equation 13 is the crossover-type adsorption-desorption model, which describes relationships between adsorption density, equilibrium concentration, solid concentration, and EPC0. Calculation of the Model Parameters. Calculation of β. According to eq 7, for two points of an isotherm under a fixed Cs condition
β Psor1 ) Ksd × C-n s × C1 - NAP
(14)
β Psor2 ) Ksd × C-n s × C2 - NAP
(15)
β β ∆P1 ) Psor2 - Psor1 ) Ksd × C-n s (C2 - C1 )
(16)
where (Psor1, C1) and (Psor2, C2) are experimental data of two points on the isotherm. Similarly, for another two points of the same isotherm β β ∆P2 ) Psor4 - Psor3 ) Ksd × C-n s (C4 - C3 )
(17)
∆P1 Psor2 - Psor1 C2β - C1β ) ) ∆P2 Psor4 - Psor3 C β - C β
(18)
4
3
Based on eq 18, by using the data of four points of an isotherm, β can be calculated. Calculation of n. According to eq 7, for two points with the same C value of two isotherms under different Cs conditions
Psor1 ) Ksd × Cβ × Cs1-n - NAP1
(19)
Psor2 ) Ksd × Cβ × Cs2-n - NAP2
(20)
After β is calculated from eq 18, NAP1 and NAP2 can be calculated from eq 12 under Cs1 and Cs2 conditions, respectively. Divide eq 19 by eq 20 and rearrange
( )
Psor1 + NAP1 Cs2 ) Psor2 + NAP2 Cs1
(
log n)
n
(21)
)
Psor1 + NAP1 Psor2 + NAP2 Cs2 log Cs1
( )
(22)
Based on eq 22, by using two points with the same C value of two isotherms under two Cs conditions, n can be calculated.
Experimental Section Sampling. The dust sample was collected by a dust sampler in Eilat, Israel on May 28, 1997, following a dust storm. Details of the sampling method are described elsewhere (26). The grain size distribution of SD in seawater was between 0.4 and 10 µm. Since Nile PM from annual floods no longer exists, sediments laid down during the past floods in the delta were used to approximate the suspended particulates in the E. Med. Cores VII, using the notation given in ref 27, were collected from the Manzalah Lagoon in the Nile delta approximately 4 km S.W. of Port Said. The cores were sampled from a depth of 45-90 cm which, based on sedimentological evidence, were deposited prior to 1964. Surface seawater was collected from 15 km offshore of Haifa, Israel. pH change before and after the equilibrium experiment was less than 0.03 pH units (pH ) 8.07-8.12) as seawater is a natural buffer for pH and salinity. Sample Treatment. Cores VII, which contained coarsegrained silt and sands, were wet sieved through a 63 µm sieve using Milli-Q water. Particles less than 2 µm were collected using a conventional granulometric method (28). The < 2 µm particles were then filtered through a 0.45 µm membrane. Particles of 0.45-2 µm, designated as Nile PM here, were used throughout the experiments (only very fine particles from the Nile could be jetted out into the E. Med
and deposited in the delta). When these particles (0.45-2 µm) are filtered through a membrane of 0.2 µm filter during the adsorption experiment, the effect of nonsettling colloids, which is the major reason for a “pseudo Cs effect” (18-25), may be largely eliminated. The washing and particle selecting treatment may wash out part of particulate P in the sediment sample, which may help the sample be more representative since the sediment had been contacted with the anaerobic pore water, where P level was normally much greater than that in the bulk water, for many years. Because of the difficulties to get real Nile PM that was transported to the delta pre-1964, it was considered necessary to treat the sample in a consistent way to get repeatable experimental data and to return the sample as far as possible to its original deposition state. Seawater Leachable P Experiment (SD Only). Different amounts of dry SD were weighed into a series of 50 mL sterilized polypropylene centrifuge tubes. Forty milliliters of E. Med surface seawater was added to each tube to produce a series of suspensions with different Cs. The tubes were capped and shaken in the dark for 3 days. A kinetic experiment (data not shown here) indicated that the leaching process reached equilibrium within 48 h. Suspensions were filtered through 0.2 µm Whatman polycarbonate membrane filters, and the solutions were measured for P concentration. Kinetic Adsorption Experiment. Kinetic adsorption experiments under different Cs conditions were carried out according to ref 25. The total volume for each experiment was made up to 175 mL using E. Med seawater. For the Nile PM experiment, the initial P concentration was 0.8 µM for Cs1 (18 mg/L), 1.0 µM for Cs2 (48 mg/L), and 1.2 µM for Cs3 (120 mg/L). For the dust experiment, the initial P concentration was 1.0 µM for Cs1 (42 mg/L), 1.2 µM for Cs2 (120 mg/L), and 1.56 µM for Cs3 (228 mg/L). These conditions were chosen after one or more preliminary experiments so that reactions under different Cs conditions could reach a similar final P concentration. Kinetic curves obtained in this way can be used to directly confirm/explain a Cs effect (25). Adsorption-Desorption Isotherm Experiment. Adsorption isotherm experiments under different Cs conditions were conducted according to ref 25. The Cs condition examined was higher than that normally found in the surface water in order to simulate the dust storm and flooding. For SD experiments, dry dust was first weighed into the centrifuge tube and leached with E. Med surface seawater for 3 days. After the supernatant was removed, the remaining wet dust was used for the adsorption experiment. Tubes were capped and shaken in the dark at room temperature for 7 days. A kinetic experiment (data not shown here) indicated that adsorption reached equilibrium in 2-6 days depending on the Cs condition. Five milliliters of suspension was taken and filtered though 0.2 µm membrane for P analysis. Analytical Procedure. Phosphate was determined according to ref 29 by a Technicon AA-II autoanalyzer. The precision (7.8 nM; 2 s) was determined by replicate standards, and the limit of detection was 6.8 nM (2s; n ) 6). All solid samples were analyzed for Si, Al, Fe, Mn, Mg, Ca, Na, and K content by XRF. Analysis of a certified standard (SARM42) was better than 6.3% for all elements with a total recovery of 99.6%. The SEDEX method (30) was used for the measurement of particulate P speciation in the dust and sediment. The extraction procedure consists of five sequential leaching steps which extract the following: (1) exchangeable and loosely bound P (Pex), (2) iron-bound P (PFe), (3) authigenic and biogenic apatite and calcium carbonate bound P (Paut), (4) detrital apatite and any other remaining inorganic P phases (Pdet), and (5) organic P (Porg). Details of the extraction were given elsewhere (31). VOL. 36, NO. 16, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3521
FIGURE 1. Adsorption kinetics of Nile PM and SD in E. Med seawater under different Cs conditions. (a) Solid points are for Nile PM. Cs: 18 mg/L (solid circle), 48 mg/L (solid square), and 120 mg/L (solid triangle). Open symbols are for Saharan dust. Cs: 42 mg/L (open square), 120 mg/L (open circle), and 228 mg/L (open triangle). All the Nile PM experiments ended at a similar final P concentration of 0.7 µM. (b) Adsorption kinetics of Nile PM (solid points in Figure 1a) extended to a week.
Results Seawater Leachable P from SD. When dry SD was leached with E. Med surface seawater (blank P concentration < 0.03 µM), an average of 3.33 ( 0.26 µmol/g (1s; n ) 20) of phosphate was released to the seawater. The concentration of dust used in the experiment did not affect the leachable P (µmol/g) within the range tested (Cs: 2-250 mg/L). Adsorption Kinetics. The initial adsorption rate of Nile PM (containing 0.8-1.2 µM initial P) increased as the solid concentration increased (Figure 1a, solid symbols). However, the increased adsorption rate under higher Cs conditions did not lead to a greater final adsorption density, since adsorption under lower Cs conditions proceeded continuously, although more slowly, to reach a higher final adsorption density (Figure 1b). In contrast to Nile PM, leached SD showed no P adsorption within the first 4 h when they were placed in the seawater containing 1.0-1.56 µM initial P (Figure 1a, open symbols). The SD experiment was therefore stopped after 4 h. Adsorption Isotherms. Adsorption isotherms for both Nile PM and SD declined as Cs increased (Figure 2). They both crossed over the aqueous concentration axis. Within the P concentrations of surface water ( 1 or C > EPC0, Psor > 0 (adsorption area) (24)
TABLE 1: P Speciation and Major Elements Analysis of Nile PM and SD Samples Nile PM SD
Nile PM SD
Pex, µmol/g
PFe, µmol/g
Paut, µmol/g
Pdet, µmol/g
Porg, µmol/g
Pinorg, µmol/g
Ptotal, µmol/g
5.05 3.13
21.43 7.23
7.397 77.66
3.13 21.75
10.05 1.90
37.01 109.80
47.07 111.70
Si, wt %
Al, wt %
Mg, wt %
Fe, wt %
K, wt %
Ca, wt %
Na, wt %
21.53 18.50
9.87 4.72
1.84 2.37
9.16 3.33
1.39 0.73
0.49 12.77
0.14 1.29
when λ < 1 or C < EPC0, when λ ) 0 or C ) 0,
Psor < 0 (desorption area) (25)
Psor ) -NAP
(26)
When P concentration of a water is higher than EPC0, the particle will be a sink of P in the water (λ > 1, adsorption). When P concentration is lower than the EPC0, the particles will be a source of P in the water (λ < 1, desorption). If P concentration of the water equals EPC0, the particles will have no obvious effect on DIP (λ ) 1). EPC0, NAP, and the Effect of Aqueous Medium. Among the total phosphorus contained in a natural solid sample (Ptotal), various forms of particulate P can behave differently in terms of particle-water interaction.
Ptotal ) Pinert + NAP + Pdissolve
(27)
where Pinert represents the fraction that is not released into the natural water unless chemical extraction reagents are used. Pdissolve represents the fraction that can be dissolved into the natural water. Although both NAP and Pdissolve may be “available” to natural waters, Pdissolve is determined by solubility, while NAP by adsorption-desorption properties. NAP can remain in the solid phase even though P concentrations in solution are far below the solubility of P-containing mineral phases. NAP refers to the P that is natively adsorbed in the solid phase when C G EPC0 and desorbed into the solution when C < EPC0. Pdissolve may be released into the solution when C G EPC0. Thus, NAP and EPC0 for the same solid sample may vary with the property of the solution (e.g. pH, salinity, and Cs conditions). According to eq 13, changes in EPC0 can alter the role of natural particles in terms of adsorption or desorption of P especially when λ is around 1. So far, we can only experimentally measure the EPC0. To theoretically predict EPC0 will be a task for future studies. For Nile PM in E. Med seawater (Figure 2a), EPC0 values are 0.005, 0.06, and 0.32 µM under Cs conditions of 6, 18, and 120 mg/L, respectively. These values, according to eq 8, correspond to NAP values of -0.37, -1.12, and -1.2 µmol/g under the same Cs conditions, respectively. For SD (Figure 2b), EPC0 values are 0.28, 0.61, 0.71, and 1.375 µM under the Cs conditions of 25, 52.5, 100, and 252 mg/L. The corresponding NAP values for SD are -0.92, -1.19, -1.21, and -1.47 µmol/g under the same Cs conditions, respectively. From Table 1 and the result of leachable P from SD it can be seen that much of particulate P in the SD is held by CaCO3 (Paut and Pdet counted for ∼90% of Ptotal), of which only less than 3% can be counted as NAP and/or Pdissolve. The high Pinert in the dust sample represents a high chemical potential of P in the solid phase, which may be responsible for the low adsorption capacity of SD. The larger grain size of SD may be partially responsible for its lower adsorption capacity than that of Nile PM. Contrasting Adsorption Behavior between SD and Nile PM. The modeling in Figure 2 explains why SD acts a source of P, while Nile PM a dual role in the same water. EPC0 values for SD are generally higher than the P concentration in the
FIGURE 3. Adsorption (solid lines) and desorption (dotted lines) isotherms of Nile PM in E. Med seawater. Square symbols: 6 mg/L. Triangular symbols: 120 mg/L. E. Med deep water so that λ < 1 when P < 0.3 µmol/L, making SD a source of P. By contrast, EPC0 values for the Nile PM in E. Med are generally less than 0.3 µmol/L, making Nile PM a sink of P (λ > 1) under the deep seawater conditions. In E. Med surface seawater where P concentration is close to their EPC0, little P is adsorbed or desorbed by Nile PM. The contrasting kinetic adsorption results between SD and Nile PM in Figure 1 can now be easily understood from Figure 2. Conditions of P concentration and Cs used for Nile PM in the kinetic experiment are all in λ > 1 (adsorption) area, while those for SD are in λ < 1 (desorption) area. It is worth to note that EPC0 values for Nile PM in E. Med seawater could be much lower than that in Nile river water (where P levels were high) since the NAP of Nile river PM could have been decreased as it was transported to the E. Med where P level was much lower. Clearly, depending on the λ, which is controlled by solute concentration, solid concentration, and aqueous medium conditions, natural particles can change its role in terms of sink and source of P in different waters. The validity of λ discriminance (eq 13) needs to be verified with more samples and elements/pollutants in natural waters and soils. Solid Concentration Effect. According to MEA theory, the Cs effect is an inherent property of adsorption. Since the MEA theory is published (24, 25), several studies have confirmed this aspect of the theory (33-38). One of the inferences of MEA theory is to predict a general rule for the existence of a Cs effect. If an increase in solid concentration causes a decrease in adsorption reversibility (which is often a result of increased adsorption rate), then a Cs effect should exist. However, if a change in Cs causes no changes in adsorption reversibility or MEA states, then a Cs effect should not physicochemically exist in such a system. Experimental data here agree with this prediction. In Figure 3, adsorption isotherms declined remarkably as Cs increased from 6 to 120 mg/L. Correspondingly, the hysteresis, which is the angle between adsorption and desorption isotherms, became bigger as Cs increased. The increase of adsorption irreversVOL. 36, NO. 16, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3523
ibility in Figure 3 corresponded to the increase of adsorption rates as Cs increased (Figure 1). P Limitation in the Eastern Mediterranean. When SD was put into E. Med seawater, the net effect was for SD to release 3.3 µmol/g P into the water. There was no obvious readsorption of P back to the dust sample after the leaching (Figure 1a, open symbols). Herut et al. (5) estimated that the dry deposition at Tel Shikmona (Israel) released 12.5 µmol/g P into the water and that the readsorbed amount of P counted for 15% of the leached P. Herut et al. (26) also estimated that Israel loess released 1.3 µmol/g P into the surface seawater. Among the data available in the literature, dust samples in the region are all sources of P to the E. Med seawater. Saharan dust is therefore unlikely to be the reason for P limitation in the region. The Role of Nile PM during pre-1964 Floods. Because the river Nile no longer floods into the Mediterranean basin after the completion of the Aswan dam, it is not possible to measure directly the past flux of P associated with particulates from the river into the basin. However it is possible to use the model here to reconstruct the behavior of Nile river PM as it reached the coastal waters during the past floods. The P level of the Nile water was typically 6.4 µmolP/L (39). The estimated particulate flux was 25 × 106 ton/year. Using a water flux of 43 × 109 m3/year, the solid concentration was 580 mg/L during the floods. Suppose the EPC0 value of Nile water was 0.4 µM, a Psor of 3.37 µmolP/g was calculated by using Psor ) 43 × C-0.58 × (C0.7 - EPC0.7 s 0 ). Suppose the Nile floodwater met an Egyptian coastal water of P ) 0.1 µmol/L, Cs ) 15 mg/L, where particles held the same NAP value as that in Nile water (EPC0 ) 0.4 µM), a Psor of -2.92 µmol/g was calculated using the same equation. Negative Psor indicated a desorption process. This indicated that Nile PM could adsorb P in river water (P sink), where the P level was higher than its EPC0, and release it into the coastal water (source) where P level was lower than its EPC0. The total P released from river PM into the Egyptian coastal water during the Nile floods in this example was 4877 tonsP/year (an upper limit). This number is greater than the dissolved P flux of 3200 tons/yr during the floods (S.W. Nixon, personal communication). Our calculation confirmed what was previously only a qualitative observation that much of the nutrient supply to the continental shelf during the floods was supplied by phosphate attached to the particulate matter (39, 40). The input of nutrients triggered a thick and continuous diatom blooms in the surrounding seawater, which supported large shoals of sardines. The sardine fishery collapsed in 1970s after the closure of the Aswan Dam. Thus, the damming of the Nile could reduce ∼8000 ton/year P to the coastal water, which might play an important role in the P limitation in the region.
Acknowledgments The work was supported by Chinese NNSF Grant No. 20073060, UK NERC Grant No. GR3/10016, and Chinese CAS Grant RCEES-KIP-9901. Samples of Nile Delta sediment were provided by Professor D. J. Stanley (Smithsonian Institute, Washington, DC). We appreciate Dr. H. Ji’s (Ocean University of Qingdao, China) help for the experiment. We thank three reviewers for their valuable comments and suggestions.
Literature Cited (1) Krom, M. D.; Brenner, S.; Kress, N.; Gordon, L. I. Limnol. Oceanogr. 1991, 36(3), 424.
3524
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 16, 2002
(2) Zohary, T.; Robarts, R. D. Limnol. Oceanogr. 1998, 43, 387. (3) Chester, R.; Baxter, G. G.; Behairy, A. K. A.; Connor, K.; Cross, D.; Elderfield, H.; Padgham, R. C. Mar. Geol. 1977, 24, 201. (4) Martin, J. M.; Elbax-Poulichet, F.; Guieu, C.; Loye-Pilot, M. D.; Han, G. Mar. Chem. 1989, 28, 159. (5) Herut, B.; Zohary, T.; Robarts, R. D.; Kress, N. Mar. Chem. 1999, 64, 253. (6) Praska, K.; Sanders, R.; Jickells, T. Mar. Pollut. Bull. 1998, 36, 718. (7) Fox, L. E.; Sager, S. L.; Wofsy, S. C. Ceochim. Cosmochim. Acta 1986, 50, 783. (8) Froelich, P. N. Limnol. Oceanogr. 1988, 33, 649. (9) Lebo, M. E.; Sharp, J. H. Est. Coastal Shelf Sci. 1992, 35, 235. (10) Rendell, A. R.; Horrobin, T. M.; Jickells, T. D.; Edmunds, H. M.; Brown, J.; Malcolm, S. J. Est. Coastal Shelf Sci. 1997, 45, 653. (11) Lebo, M. E. Mar. Chem. 1991, 34, 225. (12) Fox, L. E.; Sager, S. L.; Wofsy, S. C. Limnol. Oceanogr. 1985, 30, 826. (13) Barrow, N. J. J. Soil Sci. 1983, 34, 733. (14) Slomp, C. P.; Malschaert, J. F. P.; Van Raaphorst, W. Limnol. Oceanogr. 1998, 43, 832. (15) Lyons, J. B.; Gorres, J. H.; Amador, J. A. J. Environ. Qual. 1998, 27, 895. (16) Sundareshwar, P. V.; Morris, J. T. Limnol. Oceanogr. 1999, 44, 1693. (17) O’Connor, D. J.; Connolly, J. P. Water Res. 1980, 14, 1517. (18) Honeyman, B. D.; Santschi, P. H. Environ. Sci. Technol. 1988, 22, 862. (19) Gschwend, P. M.; Wu, S. C. Environ. Sci. Technol. 1985, 19, 90. (20) Nyffeler, U. P.; Li, Y. H.; Santschi, P. H. Geochim. Cosmochim. Acta 1984, 48, 1513. (21) Cohen Stuart, M. A.; Scheutjens, J. M.; Fleer, G. J. J. Polym. Sci.: Polym. Physics Ed. 1980, 18, 559. (22) Di Toro, D. M.; Mahony, J. D.; Kirchgraber, P. R.; O’Byrne, A. I.; Pasquale, L. R.; Piccirilli, Environ. Sci. Technol. 1986, 20, 55. (23) MacKinley, J. P.; Jenne E. A. Environ. Sci. Technol. 1991, 25, 2082. (24) Pan, G.; Liss, P. S. J. Colloid Interface Sci. 1998, 201, 71. (25) Pan, G.; Liss, P. S. J. Colloid Interface Sci. 1998, 201, 77. (26) Herut, B.; Krom, M. D.; Pan, G.; Mortimer, R. Limnol. Oceanogr. 1999, 44, 1683. (27) Siegel, F. R.; Slaboda, M. L.; Stanley, D. J. Environ. Geol. 1994, 23, 89. (28) Loring, D. H.; Rantala, R. T. T. Earth Sci. Rev. 1992, 32, 235. (29) Krom, M. D.; Brenner, S.; Isrialov, L.; Krumgalz, B. Oceanological Acta 1991, 14, 189. (30) Ruttenberg, K. C. Limnol. Oceanogr. 1992, 37, 1460. (31) Eijsink, L. M.; Krom, M. D.; de Lange, G. Mar. Geol. 1997, 139, 147. (32) Pan. G.; Liss, P. S.; Krom, M. D. Colloid Surf. A-Physicochem. Eng. Asp. 1999, 151, 127. (33) Trivedi, P.; Axe, L. Environ. Sci. Technol. 2000, 34, 2215. (34) Liu, C. X.; Kota, S.; Zachara, J. M.; Fredrickson, J. K.; Brinkman, C. K. Environ. Sci. Technol. 2001, 35, 2482. (35) Meszaros. R.; Nagy, M.; Varga, I. Langmuir 1999, 15, 1307. (36) Raposo, M.; Oliveira, O. N. Langmuir 2000, 16, 2839. (37) Schlupen, J.; Haegel, F. H.; Kuhlmann, J.; Geisler, H.; Schwuger, M. J. Colloid Surf. A-Physicochem. Eng. Asp. 1999, 156, 335. (38) Trivedi, P.; Axe, L.; Dyer, J. Colloid Surf. A-Physicochem. Eng. Asp. 2001, 191, 107. (39) Halim, Y. Ocean Margin Processes in Global Change: Dahlem Conference; Wiley: Chichester, 1991; p 301. (40) Eijsink, L. M.; Krom, M. D. Herut, B. Am. J. Sci. 2000, 300(6), 483.
Received for review January 9, 2002. Revised manuscript received May 21, 2002. Accepted May 22, 2002. ES020516D
