Pore waters were collected anaerobically at 1–2 cm intervals from the sediment-water interface (SWI) to a depth of 50 cm using pore water diffusion equilibrators ("peepers"' see Ref. 8 for construction details). Prior to deployment in the field, peepers were washed with dilute HCl or HNO₃ and then kept in a Plexiglas box filled with deionized (DDI) water and degassed using N₂ for 3 to 4 days. The peepers were then placed in vinyl bags filled with N₂ and transported to the field, where they were inserted into the sediments and left to equilibrate with the surrounding sediment pore waters for 3–4 weeks. Peepers were emplaced during fall (Nov 2001), winter (Feb 2002), spring (May 2002) and summer (Aug 2002). Upon retrieval (in Dec 2001, Mar 2002, Jun 2002 and Aug 2002) peepers were placed into fresh vinyl bags flushed and filled with N₂, and were immediately returned to the lab. Pore waters were extracted from each bag using precleaned polypropylene syringes attached to stainless steel needles by piercing both the bag and the dialysis membrane covering each peeper chamber. Each pore water sample was subsequently filtered through a 0.2 μm pore size syringe filter. During the fluid extraction, N₂ was periodically flushed into the inflated vinyl bag to maintain positive internal pressure through needle holes and avoid leakage of atmospheric O₂ into the peeper.

All reagents were ACS reagent grade or purer, and all glassware and plasticware was precleaned with trace metal grade nitric acid. Solution pH was immediately measured for each filtered sample, which was then divided into portions for analysis of redox-sensitive species (alkalinity, dissolved Fe⁺²/Fe⁺³, ΣNH₄⁺, ΣPO₄^-3, ΣS^-2, and SO₄^-2). Analyses for ferrous and ferric iron, alkalinity, ammonium, sulfide and phosphate were immediately conducted via UV/Vis spectrophotometry. Samples to be analyzed for sulfate were preserved by acidification using concentrated nitric acid (see Ref. 8 for all spectrophotometric methods). Trace metal concentrations were analyzed using a ThermoElectron PQ ExCell inductively coupled plasma mass spectrometer (ICP-MS). Total dissolved inorganic carbon was analyzed using an OI Analytical 1020A TOC Analyzer.

Lake sediment cores were collected from within 1 m of peeper insertion points, within 1–2 days of peeper extraction, using a Russian peat borer (Aquatic Research Instruments Inc.). Core samples were taken in hemispheric sections 0.5 m in length, and were divided on site, under ambient atmospheric conditions, into subsections that were stored in plastic bags. The bags were squeezed shut to remove as much ambient air as possible, sealed, returned to the lab, frozen, and then freeze-dried in a LabConco sediment freeze dryer. Subsamples of freeze-dried sediments were combusted at 550°C for 2 hours for determination of loss on ignition 41.

Sequential sediment extractions were carried out on freeze-dried sediment samples using two different methods 424344. The two-step Kostka and Luther extraction scheme is intended to quantify the proportions of amorphous Fe(III) (hydr)oxide phases and iron monosulfides, respectively. Briefly, 10 mL of ascorbic acid reagent (10 g sodium citrate and 10 g sodium bicarbonate in 200 mL N₂-sparged ≥ 17.5 MΩ water with 4 g ascorbic acid added to give final pH of 8) is added to approximately 1 g of freeze-dried, homogenized sediment from each depth interval. This is mixed on a rotating wheel in a Coy anaerobic chamber (~5% H₂, 10% CO₂, 85% N₂) at 25 rpm for 24 hours, removed from the chamber and centrifuged at 6500 rpm for 30 minutes. The supernatant is filtered through a 0.2 μm syringe filter and analyzed using the ferrozine method for dissolved total Fe. This step releases ascorbate-extractable iron (AEF), which method calibration experiments suggest is primarily comprised of amorphous Fe(III) (hydr)oxides 42. The remaining sediment is thoroughly washed using ≥ 17.5 MΩ water, and is then extracted for 1 hour using 0.5 M HCl which will liberate acid volatile sulfides (AVS), together with any remaining amorphous Fe(III) oxides 42. Iron extracted in this step is referred to as HCl-extractable iron (HEF). Filtered supernatants are analyzed immediately using UV/Vis spectrophotometry for dissolved Fe⁺³ and Fe⁺².

The five-step Tessier method is intended to assess trace metal partitioning among the following operationally-defined sediment fractions: readily exchangeable, carbonates, iron and manganese oxyhydroxides (FMOs), oxidizable (organics and sulfides) and residuals (primarily silicates). The targeted fractions, sequential extraction reagents and conditions are shown in Table 2. Three modifications are made compared to the original procedure proposed by Tessier et al. 43 to be consistent with methods used by Koretsky et al. 12 and to allow the use of small quantities of sediments. First, only 0.2 g of sediment is used for each extraction step, rather than the 1 g suggested in the original procedure. This should insure sufficient volume of each extracting reagent to remove all of each targeted fraction. Second, only 10 mL of hydroxylamine HCl is used to extract the FMO fraction, compared to 20 mL suggested volume in the original procedure. Due to the small sample size used in this study, this should not limit the extraction of the FMO phase. Lastly, in the residual extraction, microwave digestion is used, rather than the HF/HClO₄ reagent called for in Step 5 of the original Tessier method. In this study, this step is instead completed using a microwave digestion method modified from the "EPA 3051" method used for rapid soil acid digestion. In PFA-Teflon vessels, 25 mL of 50% HNO₃ (trace metal grade) was added to the sediment residue remaining after step 4 and heated under controlled pressure and temperature in a MARS5-DS 9015 Microwave Digester. Because HF was not used, a small quantity of refractory material remained after this step. After each extraction step, mixtures are centrifuged at 6500 rpm for 30 minutes and supernatants filtered through 0.2 μm syringe filters. As above, dissolved Fe is analyzed colorimetrically. Concentrated trace metal grade HNO₃ is added to preserve a second portion of each sample for trace metal analyses using ICP-MS. Between steps, sediment residues are washed repeatedly using ≥ 17.5 MΩ DI water which is carefully drained before addition of the next reagent.

Sediment pore waters collected using two sets of peepers during four seasons (total of eight peeper data sets) were analyzed for pH, alkalinity, redox-sensitive dissolved species (dissolved Fe⁺², Fe⁺³, ΣS^-2, SO₄^-2), dissolved nutrients (ΣNH₄⁺, ΣPO₄^-3) and total dissolved inorganic and organic carbon. Peepers were emplaced in areas separated by not more than ~1.5 m (see above), but were positioned in an attempt to capture maximum heterogeneity (e.g. by placing peepers in regions that were visually sparser or denser with respect to vegetation during fall, spring and summer).

During all seasons, pH values are circumneutral with little variation with depth below the sediment water interface (SWI; Fig. 1). Depth-profiles of pH are similar in the replicate peepers, except during summer, when the two profiles are offset by ~0.5 pH. Above the SWI, a decline in pH with depth is sometimes apparent.

In fall, alkalinity is similar in both peepers, increasing to ~10 mM by a depth of ~8 cm and then remaining essentially constant with depth (Fig. 2A). Replicate profiles are more variable in winter, spring and fall (Figs. 2B–D). In winter, more alkalinity accumulates at shallower depths than in fall (Fig. 2B). In spring, the peeper 1 alkalinity profile is essentially constant with depth (Fig. 2C). In peeper 2, there is an increase in alkalinity from ~11 mM at 10 cm above SWI to ~15 mM at 10 cm depth, followed by a decline to ~10 mM at ~25 cm depth (Fig. 2C). Alkalinities in summer are lower than during other seasons, and vary little with depth (Fig. 2D). Dissolved total inorganic carbon (TIC) profiles (not shown) measured during winter and spring using an OI Analytical TOC Analyzer are in excellent agreement with the total alkalinity values shown in Fig. 2B and 2C, suggesting that the total alkalinity is dominated by the contribution from dissolved carbonate.

Dissolved nutrient (ammonium and phosphate) profiles are much more spatially and temporally variable than pH or alkalinity (Figs. 3, 4). In fall, peeper 1 shows distinct peaks in dissolved phosphate and ammonium just above the SWI (Fig. 3A, 4A). These sharp peaks are absent in peeper 2, where nutrient concentrations accumulate gradually to much higher levels than in peeper 1 (Fig. 3A, 4A). Nutrients were not analyzed during winter. In spring and summer, dissolved phosphate concentrations are relatively low (~50 μM in spring; ~25 μM in summer) and vary little with depth below the SWI (Fig. 3B,C). Ammonium concentrations in spring, as in fall, are quite variable in the replicate peepers (Fig. 4B). Peeper 2 has concentrations approaching those observed at the high-ammonium site in fall, while peeper 1 has much lower concentrations, similar to levels measured in summer (typically <200 μM). In both summer peepers, dissolved ammonium concentrations are low and vary little with depth (Fig. 4C).

Dissolved Fe(II) profiles vary significantly with season (Fig. 5). In fall and spring, Fe(II) profiles are distinct in the two replicates, whereas in winter and summer, replicate dissolved Fe(II) profiles are very similar. In fall, Fe(II) accumulates in the pore waters from near detection limits at the SWI to 100–200 μM at 20 cm depth in peeper 1 and >400 μM at 45 cm depth in peeper 2 (Fig. 5A). In winter, dissolved Fe(II) begins to build up ~4 cm above the SWI, reaching concentrations of ~200 μM by 20 cm depth (Fig. 5B). In spring, Fe(II) concentrations are detectable in samples from 10 cm above the SWI and increase to a maximum of ~350 μM at ~12 cm depth in peeper 1. In peeper 2, Fe(II) concentrations also increase from the uppermost sample, but Fe(II) concentrations are lower (Fig. 5C). In summer, dissolved Fe(II) concentrations are very low (<30 μM) at all depths in both peepers (Fig. 5D).

Dissolved sulfide concentrations are also seasonally variable (Fig. 6). In fall, sulfide concentrations are always <5 μM (Fig. 6A). During spring, sulfide concentrations are low throughout peeper 2, but begin to increase at a depth of ~25 cm to ~25 μM by 40 cm in the peeper 1 (Fig. 6B). Sulfide concentrations are much higher in summer than in fall or spring. In summer, in peeper 2, sulfide begins to accumulate at least 6 cm above the SWI, steadily increasing to ~75 μM by 40 cm depth. In contrast, in peeper 1 during summer, a broad peak in sulfide occurs, with sulfide levels increasing from <5 μM at 8 cm above the SWI to ~100 μM from 6 cm above to 6 cm below the SWI, then declining to ~35 μM at 20 cm depth. Sulfide then accumulates further with depth, reaching ~90 μM by 40 cm depth (Fig. 6C).

Like dissolved Fe(II) and sulfide, pore water Fe(III) varies tremendously with season (Fig. 7). Fe(III) profiles are similar in fall and winter, with Fe(III) accumulating just above the SWI, and then declining with depth (Fig. 7A,B). There is more divergence between replicate Fe(III) depth profiles in spring and summer. In spring, Fe(III) concentrations above 10 cm are much lower in peeper 1. Below 10 cm depths, the profiles are virtually identical (Fig. 7C). In summer, Fe(III) levels are very low (<5 μM) at all analyzed depths in peeper 2. In contrast, in peeper 1, Fe(III) concentrations reach 100 μM at 7 cm above the SWI and decline to < 5 μM by ~10 cm depth (Fig. 7D).

Depth profiles of sulfate were measured in fall and spring only (Fig. 8). Trends of sulfate in the replicate peepers are similar with lower concentrations in peeper 1 during both seasons. In fall, sulfate concentrations in both peepers increase slightly from just above to just below the SWI, and then are relatively constant with depth (Fig. 8A). Sulfate levels are higher in spring than in fall, particularly in peeper 1. Sulfate concentrations in peeper 2 increase from 10 cm above the SWI to ~5 cm depth and then decline slowly with depth. In contrast, sulfate concentrations in peeper 1 vary little from 10 cm above the SWI to ~22 cm depth, and then decline to ~80 μM at 40–45 cm depth (Fig. 8B).

Three replicate cores at a distance of <1.5 m from each other were collected in fall. Two replicate cores were collected during spring and summer. These cores, like the peepers, were visually located in areas of variable density of vegetation. Because of the difficulty associated with coring through the frozen lake in winter, only one core was taken in winter.

Sediment redox zonation can be recognized and delineated from measured pore water Fe(II), Fe(III), sulfate and sulfide depth-concentration profiles. Specifically, the presence of dissolved Fe(II) can be used to define a suboxic zone located just below the zone of oxygen penetration (oxic zone) and just above the sulfidic zone, which is defined by accumulation of dissolved sulfide 9. In saltmarsh sediments, pore water redox stratification is strongly dependent on season, the presence or absence of vegetation and the activity of bioturbating organisms [e.g., 845]. More compressed redox stratification (thinner zones, with the onset of the sulfidic zone nearer to the sediment surface) is favored by warmer temperatures, more dense vegetation and less bioturbation activity. Similarly, sulfur cycling and pore water composition in lake 5111316 and freshwater marsh 12 sediments and overlying waters have been shown to vary with season. This study demonstrates that redox stratification in the littoral sediments of a freshwater kettle lake also varies strongly with season.

Dissolved Fe(II) and sulfide pore water profiles suggest a gradual compression of redox stratification (i.e. suboxic and sulfidic zones occurring at shallower depth) from fall to summer. In fall, the suboxic zone stretches from ~1 cm above the SWI, where Fe(II) begins to accumulate in the pore water to at least 50 cm depth (Fig. 5A). Throughout this suboxic zone, sulfide levels are below detection limits (Fig. 6A). Upward migration of the suboxic zone in winter is apparent from the accumulation of Fe(II) at ~3–4 cm above the SWI and by declining Fe(II) concentrations below 20 cm depth (Fig. 5B). In spring, redox compression increases further, with Fe(II) accumulating ~8–10 cm above the SWI. Although peeper 2 remains suboxic throughout the measured depth profile, in peeper 1 a sulfidic zone appears below 20 cm depth (Fig. 5C, 6B). By summer, the sulfidic zone stretches from above the SWI to a depth of at least 50 cm in both peepers (Fig. 5D, 6C).

The decline in dissolved Fe(II) within the sulfidic zone is likely due to precipitation of amorphous and crystalline Fe sulfides, as has been observed in other freshwater sediments 34647484950. Calculations using the speciation code JCHESS 51 are used with pH and measured alkalinity and total dissolved Fe(II), SO₄^-2 and ΣS^- to assess the saturation indices of several phases, including pyrite (FeS₂), siderite (FeCO₃) and pyrrhotite (FeS), as a function of sediment depth and season. Alkalinity is assumed to be equal to total dissolved bicarbonate as indicated by total inorganic carbon measurements completed during two seasons. The default JCHESS database is used with reactions and stability constants shown in Table 3. Pyrite is always strongly supersaturated, with the saturation index typically varying between 6.0 and 8.0 (Fig. 21A). Wersin et al. 52 also report high saturation indices for pyrite in lake sediments. Previous studies have shown that Fe monosulfides are close to saturation in lake sediments [e.g., 2852]. As seen in Fig. 21B, pyrrhotite is undersaturated above the SWI, but during fall, spring and summer is close to saturation in the sediments, consistent with removal of Fe(II) from the porewaters via precipitation of Fe monosulfides. It is also possible that siderite precipitates and removes Fe(II) from the porewaters. Except during summer, siderite is undersaturated above the SWI and is close to saturation in the sediments (Fig. 21C).

In most of the peepers, elevated Fe(III) concentrations are apparent in the pore waters just above the depths of maximum Fe(II) accumulation (Fig. 7). In agreement with seasonal changes in vertical redox stratification described above, the zone of Fe(III) accumulation migrates upwards from fall to summer. The presence of Fe(III) in the pore waters is consistent with oxidation of Fe(II) diffusing upward from the suboxic zone by oxygen (or other oxidants, such as nitrate or manganese) diffusing downward from the oxic zone [e.g. 67] High levels of Fe(III) are most likely due to chelation with natural organic matter. In winter, cold, oxygenated waters may increase Fe(II) oxidation, increasing Fe(III) levels in the shallow surface waters, however, ice cover can also act as a barrier to oxygen 3.

Sulfate profiles were measured only during fall and spring (Fig. 8). Sulfate concentrations vary little with depth during fall, when the sediment pore waters are mostly suboxic. In contrast, during spring, as redox zones migrate upwards, sulfate concentrations decrease in the lower portion of the sediments. Diminishing sulfate levels are particularly apparent in peeper 1, the same peeper with decreased Fe(II) and accumulation of sulfide in the lowermost samples.

Seasonal changes in AEF are consistent with expectations from the inferred pore water redox stratification. AEF is thought to be composed primarily of readily-reducible amorphous Fe(III) (hydr)oxides 5354. During fall, when the pore waters point to suboxic conditions, maximum levels of AEF are observed. In summer, when the pore waters are sulfidic, AEF concentrations are at their lowest, presumably due to either chemically- or microbially-mediated reductive dissolution of Fe(III). During spring and winter, both redox compression and AEF levels are intermediate between the fall and summer extremes.

The oxidation of organic matter by oxygen releases protons and bicarbonate, as illustrated by the general reactions

"CH₂O"_(aq) + O_2(aq) = CO_2(aq) + H₂O_(l)     (1)

and

H₂CO_3(aq) = H⁺_(aq) + HCO₃^-_(aq)     (2)

where "CH₂O" represents labile organic matter. Protons are also released by O₂-promoted reoxidation of reduced solutes, such as Mn(II), Fe(II) or H₂S, for example,

4Fe⁺²_(aq) + O₂ + 8HCO₃^-_(aq) + 2H₂O_(l) = 4 Fe(OH)_3(s) + 8CO_2(aq)     (3)

55. For this reason, pH profiles have been suggested as indicators of O₂-penetration depths in saltmarsh sediments 45 and freshwater marsh sediments 212. In fall and spring, when sediment redox stratification is least compressed, there is a decrease in pH near the top of the profile, consistent with oxidation of organic matter or reduced solutes by oxygen. In winter, the shallow water overlying the sediments lies beneath a frozen lake surface. The low pH values measured in both profiles during winter may reflect increased O₂-promoted oxidation of organic matter and reduced solutes near the surface of these cold waters.

Measurements of total dissolved inorganic carbon in winter and spring demonstrate that the alkalinity profiles represent predominantly bicarbonate alkalinity. Bicarbonate is produced by anaerobic respiration of organic matter by dissimilatory manganese reduction, dissimilatory iron reduction and microbial sulfate reduction [e.g. 55]. Bicarbonate can also be removed from the porewaters via precipitation of carbonate minerals. Therefore, the alkalinity profiles represent the net effect of organic matter degradation and carbonate precipitation. Throughout most of the profiles, alkalinity is nearly constant with depth (Fig. 2). This probably reflects saturation with respect to a carbonate phase, such as siderite (Fig. 21C), or more likely calcite (CaCO₃) or rhodochrosite (MnCO₃), but without dissolved Ca and Mn data this question cannot be satisfactorily resolved. The lower alkalinity values observed at the top of the fall and winter profiles (Fig. 2A,B) likely reflect the less compressed redox stratification during these seasons and are consistent with organic matter degradation coupled to oxygen as the terminal electron acceptor at the top of these profiles. The rapid increase in alkalinity with depth indicates the onset of anaerobic organic matter degradation well above the SWI.

Dissolved phosphate is another potential tracer of organic matter degradation, because it is released during mineralization of natural organic matter. However, because phosphate associates strongly with Fe(III) oxides [e.g. 14], dissolved phosphate can also be released to solution by reductive dissolution of Fe oxides [e.g. 4849565758]. Dissolved phosphate, frequently a limiting nutrient in freshwaters, can also be efficiently removed from solution by macrophytes and other organisms 5960. The dissolved phosphate profiles thus reflect the sum of these processes. Low levels of dissolved phosphate during summer (Fig. 3C) most likely indicate efficient uptake of phosphate by aquatic organisms and not a lack of organic matter degradation. The large peak in phosphate observed in one of the peepers during fall (Fig. 3A) probably does reflect organic matter mineralization. The high phosphate levels above the SWI in one of the summer peepers (Fig. 3C) can be attributed to a combination of organic matter mineralization and possibly release of phosphate from the reductive dissolution of Fe(III) oxides (see discussion of solid phase Fe speciation below).

Like dissolved phosphate, dissolved ammonium is produced by mineralization of organic matter and can be consumed by aquatic organisms. In oxidized sediments, dissolved ammonium can also be removed via nitrification. In fall, a peak in ammonium occurs just above the SWI, coincident with a peak in dissolved phosphate (Fig. 4A). This likely reflects organic matter mineralization. The much higher levels of ammonium observed in the other peeper may be due to spatial heterogeneity with respect to vegetation and uptake of ammonium by macrophytes. Similarly, differences in levels of ammonium observed in the two sets of peeper data collected in spring (Fig. 4B) could be caused by patchiness of macrophyte distribution and ammonium uptake. Declining levels of dissolved ammonium with depth during spring, and low levels of ammonium beneath the SWI in summer, again point to efficient uptake of this nutrient by macrophytes. Although nitrification is unlikely to occur under the relatively reducing conditions of the "bulk" sediments, it is possible that nitrification plays an important role in nitrogen transformations near macrophyte roots or irrigated burrows, where oxygen may be directly introduced into the sediments at depth 61626364.

Without measured reaction rates for nitrogen transformations, microbial sulfate reduction and dissimilatory or chemical iron or manganese reduction, it is not possible to make a quantitative assessment of natural organic matter degradation pathways in these sediments, as has been done for other lakes 7. Furthermore, no data regarding transport of solutes by advection, bioirrigation or root leakage or of solids by bioirrigation 65 are available for these sediments. Nonetheless, the data do allow some important qualitative observations.

Accumulation of Fe(II) or sulfide above the SWI suggests that organic matter degradation coupled to oxygen or nitrate as terminal electron acceptor will be limited to a zone well above the SWI, or to patches in the sediment containing oxygen introduced by macrophyte roots [e.g. 666768] or bioirrigating macrofauna [e.g 645]. Tessier extractions suggest that Mn is primarily associated with carbonates (Fig. 16). Thus, organic matter mineralization coupled to Mn(IV) reduction is unlikely to be a dominant pathway, unless manganese is cycled so quickly that Mn(IV) cannot accumulate in the sediments 3. The large quantity of AEF, especially in the upper portion of the sediment column during fall, winter and spring, (Figs. 10, 11, 12), together with the accumulation of dissolved Fe(II), implies that dissimilatory iron reduction is an important pathway for organic matter mineralization. This is consistent with quantitative estimates of organic matter oxidation in the upper 18 cm of deep lacustrine sediments from Lake Michigan, which point to the importance of microbial iron reduction (44%) with lesser sulfate reduction (19%) or oxidation by oxygen (37%) and negligible coupling to manganese reduction 7. In summer, organic matter degradation via microbial sulfate reduction probably becomes much more significant in these shallow sediments. Very high rates of microbial sulfate reduction have been measured in other freshwater sediments, in spite of relatively low concentrations of sulfate 69707172. Thus, the small quantity of dissolved sulfate does not preclude microbial sulfate reduction as a significant organic matter degradation pathway. AEF, while not entirely absent, is present in much lower quantities in the summer cores than during other seasons. If the remaining AEF is comprised of the most refractory Fe, microbial sulfate reduction will more easily outcompete dissimilatory iron reduction 373. Furthermore, studies of saltmarsh sediments have demonstrated that microbial sulfate reduction can produce quantities of sulfide sufficient to allow sulfate reducers to outcompete iron reducers 8. The mechanism of competition is hypothesized to be removal of labile Fe(III) oxides via chemical reduction (by sulfide), amplified by macrofaunally-mediate solute transport. Similar processes could occur in these sediments, but without measurements of microbial sulfate reduction rates, iron-reducing bacteria populations or activities or bioirrigation intensities, this cannot be substantiated.

The Kostka and Luther extractions suggest that Fe occurs mostly as amorphous Fe(III) (hydr)oxides (AEF), with a much smaller pool of Fe in acid volatile sulfides (HEF) present during all seasons and at nearly all sediment depths (Figs. 10, 11, 12, 13). AEF varies considerably with season and with depth in these sediments, with this variability typically greater than what is observed for replicate cores. Much more AEF is present, especially in the uppermost portion of the cores, during fall and winter than in spring and summer. This is in agreement with seasonal variations in redox geochemistry inferred from the pore waters, as described above, and suggests that AEF is a pool of Fe that undergoes a seasonal cycle of reductive dissolution and oxidative precipitation. In all seasons, AEF decreases significantly with depth. This is consistent with the expected spatial sequence of redox reactions, sometimes termed a redox loop, which concentrates Fe and Mn oxides in near-surface sediments [e.g. 36].

Operationally-defined Fe extractions based on the Tessier method (Fig. 14) give a very different picture of Fe distribution in these sediments compared to the Kostka and Luther extractions (Fig. 10, 11, 12, 13). The Tessier extraction produces more Fe total (sum of the five extraction steps) then the two steps of Kosta and Luther. This is not surprising, as the Kostka and Luther method should only target amorphous Fe(III) oxides and AVS, whereas the 5-step Tessier extraction should remove all Fe from the sediments. However, in contrast to the Kostka and Luther results, the Tessier extraction indicates that most of the Fe is associated not with FMO, but rather with the oxidizable organics/sulfides fraction. Much more organics/sulfides associated Fe is extracted then by either, or indeed by the sum, of the two Kostka and Luther steps. The depth distribution of the organics/sulfides fraction is unlike that of AEF or HEF, instead closely resembling the depth distribution of LOI (Fig. 9). This suggests that much of the Fe extracted in this fraction is associated with organic matter, rather than sulfides.

The Tessier FMO fraction, which should correspond to the AEF fraction of Kostka and Luther, accounts for the second largest pool of Fe extracted with the Tessier method. Interestingly, there is a relatively good correlation between the Tessier FMO fraction and the Kostka and Luther HEF fraction (R² of 0.88 in winter and R² of 0.55 in spring). Furthermore, in winter, nearly identical total quantities of Fe are extracted in the FMO and HEF steps. During spring, the quantities of iron extracted by the two methods are similar in the lower portion of the core, but much more Fe is extracted as FMO than as HEF in the upper 30 cm. The correlation between FMO and HEF is unexpected, as these two extraction steps are supposed to target quite different sediment fractions. Possible explanations include: (1) Kostka and Luther HEF step actually extracts Fe(III) oxides, (2) Tessier FMO step actually extracts acid volatile sulfides, or (3) HEF and FMO both extract Fe from carbonates or another non-target phase.

The first explanation is unlikely, because the HEF extractions are dominated by Fe(II), with almost no Fe(III). It could perhaps be argued that HEF actually represents Fe(III) originating from Fe(III) oxides reductively dissolved in the ascorbate extraction, which is reduced, released and adsorbed onto remaining sediment particles and then released as Fe(II) in the HCl extraction step. If this is the case, it is not clear why only the fraction targeted by the Tessier FMO would be adsorbed and released as HEF, while a much larger quantity of Fe is released in the Kostka and Luther ascorbate extraction. In fact, a similar correlation between Tessier FMO and Kostka and Luther HEF does not occur in freshwater marsh sediments 12, so this explanation seems unlikely.

The second possibility is that the Tessier FMO technique does not release Fe from oxides, but rather releases Fe associated with acid volatile sulfides. This is plausible, especially given that freeze-drying in air may oxidize and redistribute acid volatile sulfides 7475. Other workers have demonstrated that the hydroxylamine HCl used to extract FMO can partially release amorphous sulfides 7677. If this is the case, then another question arises: given the large quantity of AEF, why isn't more FMO extracted? Specifically, why is it not equal to the quantity of AVS plus the AEF? In fact, AEF profiles from the Kostka and Luther technique are similar in magnitude and depth-distribution to the oxidizable Fe released by the Tessier technique. Thus, AEF must either be composed largely of organic-bound Fe, or Fe oxides must be released in the organics/sulfides of the Tessier method. Given the excellent agreement between seasonal changes in AEF and redox conditions inferred from the pore waters, it does not seem plausible that the AEF is comprised primarily of organic-bound Fe. A more likely explanation is that Fe oxides are at least partially released in the organics/sulfide step of the Tessier method. This could reflect an insufficient quantity of hydroxylamine HCl to extract all of the Fe in these sediments, as has been demonstrated for more Fe-rich sediments 787980.

A third possibility is that neither Tessier FMO, nor Kostka and Luther HEF extract the targeted phases. For example, both might actually remove Fe(II) from carbonates. In anoxic lake sediments, freeze-drying has been shown to increase the quantity of Fe extracted in a reducible phase, with most of the "excess" reducible Fe mobilized from a combined exchangeables/carbonates fraction 75. This was attributed to oxidation of poorly crystalline siderite 75. This explanation, however, would require carbonates to escape extraction in both the Tessier carbonate extraction phase and the ascorbate extraction of Kostka and Luther. This discussion illustrates the inherent difficulties in interpretation of data from operationally-defined extraction techniques (see also reviews of Refs. 81 and 82).

In summary, the most likely scenario appears to be that the Kostka and Luther AEF removes a mixture of Fe associated with readily reducible oxides and organics, while the Tessier organics/oxidizable fraction removes a mixture of Fe associated with readily reducible oxides, organics and disulfides. Thus, the general depth-distribution of organic/sulfide associated Fe is similar to LOI profiles, except that Fe is enriched at the top of the organic/sulfide cores compared to the LOI profiles. This is not surprising, as redox reactions should serve to concentrate Fe(III) oxides into the uppermost portion of the sediments. At the bottom of the cores, larger Fe concentrations extracted in the Tessier organics/sulfides compared to Fe extracted by Kostka and Luther may indicate the presence of Fe(II) disulfides (e.g. pyrite) at depth. The Tessier organics/sulfides step is likely to remove a significant quantity of Fe disulfides 76, whereas the Kostka and Luther technique only removes Fe monosulfides. Other studies have shown progressive aging of Fe monosulfides to Fe disulfides over time 83, so it is not surprising that larger quantities of Fe disulfides would be present deeper in the sediment.

The significant seasonal variations in redox stratification of these sediments clearly influences Fe cycling. Therefore, significant seasonal variations in trace metal speciation should be expected. Reductive dissolution of FMO should mobilize associated metals, which, like Fe(II) or Mn(II) may diffuse upwards into a more oxidizing zone, where they could be reincorporated into fresh FMO. Mobilized chalcophiles diffusing downwards to more reduced sediments will likely become associated with sulfides, while lithophiles such as Mn or Cr may be incorporated into carbonate phases. Similarly, when redox stratification becomes less compressed and reducing sediments become more oxidized, degradation of sulfide phases may release and mobilize associated metals. These metals could diffuse downwards and become incorporated in other sulfides, or might diffuse upwards and associated with oxides. Depending on the geochemistry of a particular element, seasonal precipitation and dissolution of FMO and AVS could lead to significant changes in partitioning among the pore waters and various sediment constituents. In this study, trace element partitioning among operationally defined solid phases was assessed during winter and spring only. Kostka and Luther extractions and pore water sampling conducted during all seasons suggests that comparisons of winter and spring only will underestimate the potential magnitude of change throughout the entire year. Nonetheless, comparisons between these two seasons and possible implications for the measured trace elements are discussed below.