Nitrate Controls on Iron and Arsenic in an Urban Lake

See allHide authors and affiliations

Science  28 Jun 2002:
Vol. 296, Issue 5577, pp. 2373-2376
DOI: 10.1126/science.1072402


Aquatic ecosystems are often contaminated by multiple substances. Nitrate, a common aquatic pollutant, strongly influenced the cycling of arsenic (As) under anoxic conditions in urban Upper Mystic Lake (Massachusetts, USA) by oxidizing ferrous iron [Fe(II)] to produce As-sorbing particulate hydrous ferric oxides and causing the more oxidized As(V), which is more particle-reactive than As(III) under these conditions, to dominate. This process is likely to be important in many natural waters.

Arsenic cycling in aquatic systems is strongly influenced by redox processes. For example, oxidized [arsenate, H2As(V)O4 ] and reduced [arsenite, H3As(III)O3] forms of inorganic As can differ both in their tendency to form soluble or insoluble complexes (1–3) and in their toxicity to humans and aquatic communities (1). Moreover, because surface complexation of As by solid hydrous ferric oxide (HFO; Fe in the +III oxidation state) often plays a dominant role in immobilizing As (2, 3), redox processes that affect Fe speciation can also have a strong indirect effect on As (4). Thus, redox-active pollutants (oxidants or reductants) could have the potential to affect As mobility. Nitrogen (N) pollution, arising from activities such as agricultural fertilization and fuel combustion, is generally of concern for its roles in eutrophication and acidification (5); however, nitrate (NO3 ), one of the major forms of fixed N, is also a powerful oxidant. Laboratory studies have recently identified bacteria that can mediate both Fe(II) (6–8) and As(III) (9) oxidation by NO3 , and field evidence suggests that NO3 may influence Fe cycling in natural systems (10–13). Here we show that NO3 is dominant in the control of both As and Fe cycling during anoxia in urban, seasonally stratified, and eutrophic Upper Mystic Lake (UML; maximum depth ∼ 24 m, surface area ∼ 50 ha, volume ∼ 7 × 106m3). The results have implications for element cycles in many N-polluted aquatic systems.

UML's sediments contain 200 to 2100 parts per million (ppm) of As, derived primarily from industrial activity (14). These sediments are a seasonal source of As to the water column (15–17). Conventionally, the presence or absence of O2 is considered the main determinant of Fe (10) and As (18) chemistry in nonsulfidic lake waters (19), and anoxia does initiate release of Fe and As from UML sediments into the hypolimnion (depths ∼ 10 to 24 m). However, the expected reduced chemical forms, As(III) and dissolved Fe(II) (19), have only materialized occasionally in UML (15). Instead, oxidized As and particulate Fe(III) have accumulated in the anoxic water column during most years (16, 17, 20). For example, by late July 1997, O2 had fallen below detection limits (∼5 μM) in waters deeper than 15 m, and Fe and As began to be released from the sediments (Fig. 1, A and B) (21). Fe(II) and As(III), however, represented only small percentages of total As and Fe throughout the following 4 months of anoxia. Because the Fe must have been remobilized from the sediments in soluble form as Fe(II), most of this Fe was oxidized subsequent to release, in the absence of O2. Similarly, any As remobilized as As(III) must have been anaerobically oxidized to As(V).

Figure 1

Observed Fe and As profiles in UML during 1997. Primarily (A) HFO and (B) As(V) accumulated in the hypolimnion under anoxic, NO3 -rich conditions, as opposed to the expected Fe(II) and As(III). Elevated Fe(II) was measured within 25 to 50 cm of the sediment-water interface on 8/13/97 (not shown); however, Fe(II) concentrations remained much lower than HFO concentrations, until mid-November in the deepest samples. O2concentrations were below detection limits (less than ∼5 μM) in waters deeper than those denoted by the dashed line. HFO was calculated as the difference between total Fe and Fe(II). As(V) was calculated as the difference between total As and As(III).

We hypothesized that NO3 was responsible for the dominance of oxidized Fe and As. In UML, as in many eutrophic lakes, NO3 levels (21) frequently exceed 100 μM (Fig. 2). Persisting for several months after seasonal thermal stratification (thermocline at depth of 7 m) and subsequent O2 depletion in UML, a large fraction of the NO3 pool was produced in situ through microbial oxidation of NH4 + (i.e., nitrification).

Figure 2

Spring and early-summer profiles of NO3 and NH4 + in UML. After spring stratification, much of the hypolimnetic ammonia was nitrified, i.e., NH4 + + 2O2→ NO3 + 2H+ + H2O. This increased the hypolimnetic NO3 pool by 150%, making NO3 the most abundant oxidant. Mass balance estimates indicate that ∼45% of hypolimnetic O2 was consumed by this process during 1998. Similar profiles were observed in 1999.

Field observations confirm nitrate's control over Fe chemistry in UML. NO3 did not become fully depleted in 1997, and as predicted, Fe(II) levels remained low. By contrast, in 1999, NO3 depletion occurred in the deepest waters by late October, and resulting Fe(II) accumulation was highly coincident, spatially and temporally, with the NO3 -depleted region (Fig. 3A). An HFO peak developed at the NO3 -rich/ NO3 -depleted interface, analogous to the peak observed at the oxic-anoxic interface in systems in which O2 controls Fe cycling (19).

Figure 3

Impact of NO3 on Fe and As redox chemistry. (A) During 1999, particulate HFO initially accumulated in anoxic, NO3 -rich waters, but was replaced by dissolved Fe(II) at NO3 -depleted depths. (B) Fe(II) oxidation by NO3 in anaerobic sediment-slurry. In live bottles, slurry color changed from black to orange-brown as Fe(II) levels decreased, suggesting that Fe(II), likely as Fe(II)Ss, was oxidized to HFOs. The concomitant accumulation of sulfate (data not shown) is consistent with this interpretation. Nitrite (NO2 ) did occur transiently in live bottles (maximum of 0.2 mM) but decreased to less than 0.01 mM before most Fe(II) oxidation occurred. In addition, NO2 was observed in the formaldehyde controls (up to 0.13 mM by day 32) in which no Fe(II) oxidation was observed, suggesting that abiotic Fe(II) oxidation by NO2 was not important under these conditions. (C) As(V) accumulated in the water column during summer and early fall of 1999 when NO3 was present, but As(III) replaced As(V) at NO3 -depleted depths. O2concentrations were below detection limits (less than ∼5 μM) in waters deeper than those denoted by the dashed line. HFO was calculated as the difference between total Fe and Fe(II). As(V) was calculated as the difference between total As and As(III).

Mass balance at depths of 21 to 24 m indicated that NO3 , rather than other oxidants, was responsible for oxidizing most Fe(II) during anoxic, NO3 -rich periods (21). In 1997, O2 and Mn(IV)O2 accounted for at most 25% of Fe(II) oxidation [total Fe(II) oxidation = 60,000 ± 18,000 mol = 60,000 e equivalents), whereas only 40% of the NO3 consumption was required (total NO3 consumption = 29,000 mol = 145,000 e equivalents, assuming N2to be the product). The remaining NO3 consumption likely occurred through conventional denitrification. Mass balances for Fe(II) oxidation in 1998 and 1999 yielded similar results (20).

Although abiotic Fe(II) oxidation by NO3 proceeds at insignificant rates under conditions typical of UML (22), microcosm experiments using anoxic surface-sediment slurries spiked with NO3 (21) demonstrated the feasibility of biologically mediated Fe(II) oxidation by NO3 in this system (Fig. 3B). NO3 consumption (∼1 mM, or 5 meq e liter−1 when NO3 reduction to N2 is considered) was sufficient to explain the observed oxidation of both Fe(II) (1.2 meq eliter−1) and S(−II) [2.1 meq eliter−1 (20)], whereas negligible Fe (or sulfide) oxidation was observed in controls killed with azide or formaldehyde. These observations are consistent with culture-based studies demonstrating the existence of bacteria that can mediate Fe(II) oxidation through reduction of NO3 , primarily to N2 (and some N2O) (6–8).

The presence or absence of NO3 also dictated As redox chemistry (Figs. 1B and 3C). Contrary to conventional expectation (19), As(V), rather than As(III), accumulated during anoxic but NO3 -rich periods. In that As is primarily released from anoxic sediments as As(III) (19), the dominance of As(V) at these times is best explained through As(III) oxidation by NO3 [at a pseudo–first order rate of ∼0.2 day−1, estimated by mass balance, conservatively assuming that 50% of the As was remobilized as As(III)]. This is consistent with the recent discovery of bacteria that couple As(III) oxidation with NO3 reduction (9). Because As occurred at submicromolar concentrations, electron balance arguments alone cannot uniquely identify the oxidant [i.e., levels of Mn(IV)O2(s) or O2 slightly below detection limits cannot be eliminated on strict mass balance considerations]. However, the presence of NO3 and dominance of As(V) were spatially and temporally coincident throughout summer and fall 1997, summer to mid-fall 1998 (20), and summer to early fall 1999.

By contrast, immediately upon NO3 depletion during fall 1999, As(III) concentrations increased rapidly at NO3 -depleted depths (Fig. 3C), further supporting the argument that NO3 had been the primary As(III) oxidant. An alternative explanation, that NO3 had been oxidizing As(III) indirectly through NO3 -produced Fe(III) (9), is countered by the fact that much As(III) accumulated during 1999 (Fig. 3, A and C) before all Fe(III) was reduced. Following NO3 depletion (October to December 1999), mass balance suggests that both continued release of As(III) from the sediments and reduction of As(V) already in the water column occurred. The latter process could have been inhibited earlier by the presence of NO3 or simply masked by concurrent oxidation of As(III) (9).

An important consequence of NO3 leading to the dominance of HFO and As(V) is that most As should form particulate HFO-complexes. To confirm this hypothesis, we designed a N2-purged in situ serial filtration system to measure size distributions of Fe and As in anoxic water when NO3 was present (20). This device rigorously excludes O2 and filters at low velocities (∼ 0.02 cm min−1), minimizing possible artefacts such as Fe(II) oxidation and accelerated coagulation. During 1997, ∼95% of As was associated with particles (including colloids) at 22 m (representative hypolimnetic depth) throughout the period of anoxia (fig. 1S). Similar observations were made at 20 and 22 m during fall 1996. Using surface complexation modeling (21,23) of As sorption by HFO, we predicted (typically within ±15%) the measured distribution of As between particulate and dissolved (i.e., smaller than 0.05 μm) phases. Although the presence of other sorbing surfaces cannot be ruled out, none is needed to explain the observations. We calculate that >90% of As was complexed by HFO in late November 1997 in the bottom 4 m of the lake. Had As been present entirely as As(III), less [only ∼60%, calculated using sorption constants of (3)] would have been complexed by HFO. Over the entire season, mass balance estimates indicate that settling resulted in a 40% decrease in net As remobilization to the water column.

N pollution may thus have direct effects on the cycling of As in numerous other systems and may indirectly alter the cycling of other particle-reactive substances (e.g., PO4 3−, Pb, Hg, Cd) through the Fe cycle. These effects may not necessarily be adverse; in some instances, lowered metal toxicity could result owing to sorption by HFO. NO3 levels greater than 50 μM are common in lakes across the United States and Europe (table S1). Elevated NO3 is at least partly responsible for many coastal eutrophic “dead zones” (24) (table S1) and may influence Fe and trace metal geochemistry there. Many groundwaters in the United States contain elevated NO3 (25) (table S1). Postmaet al. (26) provide evidence for oxidation of reduced Fe (as pyrite) in a NO3 -contaminated aquifer, and biologically mediated Fe(II) oxidation by NO3 has recently been demonstrated in laboratory studies of anoxic paddy soils (27). Elevated NO3 has also been measured in some Bangladesh groundwaters (28, 29) (table S1), which commonly contain elevated As. Although these As concentrations are often much greater than those in UML, and multiple solid phases are present, the few available data (i.e., in which both NO3 and As have been measured) suggest that As tends to decrease at high NO3 levels (28, 29).

  • * Present address: Department of Environmental Science and Engineering, Harvard School of Public Health, Building 1 Room G21, 665 Huntington Avenue, Boston, MA 02115, USA.

  • To whom correspondence should be addressed. E-mail: dbsenn{at}


View Abstract

Navigate This Article