Disturbances to metal partitioning during toxicity testing of iron(II)‐rich estuarine pore waters and whole sediments

Abstract

Metal partitioning is altered when suboxic estuarine sediments containing Fe(II)‐rich pore waters are disturbed during collection, preparation, and toxicity testing. Experiments with model Fe(II)‐rich pore waters demonstrated the rates at which adsorptive losses of Cd, Cu, Ni, Mn, Pb, and Zn occur upon exposure to air. Experiments with Zn‐contaminated estuarine sediments demonstrated large and often unpredictable changes to metal partitioning during sediment storage, removal of organisms, and homogenization before testing. Small modifications to conditions, such as aeration of overlying waters, caused large changes to the metal partitioning. Disturbances caused by sediment collection required many weeks for reestablishment of equilibrium. Bioturbation by benthic organisms led to oxidation of pore‐water Fe(II) and lower Zn fluxes because of the formation of Fe hydroxide precipitates that adsorb pore‐water Zn. For five weeks after the addition of organisms to sediments, Zn fluxes increased slowly as the organisms established themselves in the sediments, indicating that the establishment of equilibrium was not rapid. The results are discussed in terms of the dynamic nature of suboxic, Fe(II)‐rich estuarine sediments, how organisms perturb their environment, and the importance of understanding chemistry in toxicity testing with whole sediments or pore water. Recommendations are provided for the handling of sediments for toxicity testing.

INTRODUCTION

Disturbances to estuarine sediments during collection and toxicity testing have the potential to bring about major changes to metal partitioning, thus altering exposure pathways for toxicants and affecting test outcomes. In undisturbed silty sediments, dissolved oxygen penetrates to only a few millimeters depth, below which suboxic and anoxic conditions follow in rapid succession [1,2]. In estuarine sediments, bioturbation (burrowing and irrigation activities of benthic organisms) and tidal currents increase the depth of oxygen penetration and create a mosaic of nonequilibrium microenvironments where both oxidized and reduced forms of Fe, Mn, and S coexist in a dynamic equilibrium under suboxic conditions [1,3–7]. Dissolved Fe(II) concentrations are often 1 to 10 mg/L in the pore waters of coastal and estuarine surface sediments (0–2 cm) [8–12], but have been measured at concentrations higher than 100 mg/L in some estuarine environments [3]. Depth distributions indicate that Fe(III)‐hydroxide phases and pore‐water Fe(II) coexist in most surface sediments and that an individual Fe atom is typically oxidized and reduced hundreds of times before ultimate burial [8,9,13].

The rate of oxidation of Fe(II) by dissolved oxygen (see Eqn. 1) increases with increasing pH (100‐fold per pH unit increase) and is rapid in seawater [14]

The oxidation of pore‐water Fe(II) and subsequent Fe hydrolysis reactions cause pore‐water pH to decrease, thus altering metal partitioning, whereas newly formed Fe(OH)₃ precipitates provide surfaces for the scavenging of dissolved metals [15,16]. Indeed, the pH of sediment pore waters often is lower than that of the overlying seawater.

In pore waters extracted from suboxic estuarine sediments, significant chemical changes will occur during storage and while undertaking pore‐water toxicity tests. For whole‐sediment toxicity tests, usually only the surface sediment layer (0–2 cm) is collected because most organisms are believed to live in the surface layer and this sediment is believed to be oxic (and contain the highest pore‐water metal concentrations). Once collected, sediment is commonly stored at 4°C, native organisms are removed, and the sediment is homogenized before tests commence. Sediments also are sieved, if required, to remove coarse particles [17,18]. In estuarine sediments, because of the importance of biologically mediated reactions involving Fe, Mn, and S species in redox changes, sample storage before testing may result in the release or loss of chemical constituents to or from the pore water. The process of removing native organisms and homogenization will also disturb sediment chemistry [19]. Furthermore, because sediment often is collected from depths where the dissolved oxygen concentration is low, the exposure to fully oxygenated waters during toxicity testing will alter the depth of the redox boundary and hence the sediment‐water partitioning of toxicants. For all of these processes, the rates at which these changes occur will depend on the sediment chemistry.

This study is an attempt to better understand the effects of changes to sediment chemistry during sediment and pore‐water sampling and toxicity testing, in order to provide better guidance on acceptable procedures. Although many areas of uncertainty and concern often are acknowledged with current practices, research is required to support or reject such practices, rather than continuing to rely on best professional judgment [20,21]. This study investigates the changes occurring to metal partitioning between sediment and pore water, when suboxic estuarine sediments containing Fe(II)‐rich pore waters are disturbed in the process of collection, preparation, and use in toxicity testing. Experiments with model waters were used to investigate the rate at which Cd, Cu, Ni, Mn, Pb, and Zn concentrations change as Fe(II)‐rich pore waters are exposed to air. Anthropogenically contaminated estuarine sediments with significant pore‐water Zn and Fe(II) concentrations were used to investigate changes to metal partitioning that result from sediment storage, removal of organisms, and homogenization before testing. Measurements of metal release from sediment cores were used to determine the effect of changing overlying water conditions, choosing a 2‐ or 15‐cm depth of sediment for toxicity tests, sediment bioturbation, and the rate at which sediments recover from disturbances to pore‐water chemistry. The results are discussed in terms of the dynamic nature of suboxic, Fe(II)‐rich estuarine sediments, how organisms perturb their environment, and the importance of understanding chemistry when designing toxicity test procedures that use whole sediments or pore water.

MATERIALS AND METHODS

General methods and reagents

All glassware and plasticware were cleaned by soaking in 10% (v/v) HNO₃ (Trace Pur, Merck, Darmstadt, Germany) for more than 24 h followed by three rinses with deionized water (Milli‐Q®, Millipore, Sydney, New South Wales, Australia). All laboratory‐ware used for sampling and analysis of dissolved metals was cleaned in a Class‐100 laminar flow cabinet (metal‐free HWS, Clyde‐Apac, Sydney, New South Wales, Australia). High‐purity Milli‐Q deionized water (18 MΩ.cm) was used to prepare all solutions. Clean seawater was collected from Cronulla (New South Wales, Australia). All chemicals were analytical reagent‐grade or equivalent analytical purity. Deoxygenated waters were prepared by bubbling solutions with high‐purity oxygen‐free nitrogen gas for more than 8 h to give dissolved oxygen concentrations less than 0.1 mg/L. The pH of deoxygenated seawater was adjusted in the range pH 7 to 8.2 by bubbling solutions with a mixture of highpurity oxygen‐free nitrogen and carbon dioxide gases using a mass flow controller (Model 5876–2, Brooks Instrument B.V., Veenendaal, The Netherlands) to adjust gas ratios.

Measurements of pH (calibrated against pH 4 and 7 pH buffers, Orion Pacific, Sydney, New South Wales, Australia) and redox potential (vs the standard hydrogen electrode) used a pH meter (pH 320, Wissenschaftlich‐Technische Werkstättan [WTW], Weilheim, Germany) equipped with combination pH (Sure‐flow 9165BN, Thermo Orion, Beverley, MA, USA) or redox (Model 6.0412, Metrohm, Herisau, Switzerland) probes. Salinity and temperature measurements used a conductivity meter (LF 320, WTW) with a probe (TetraCon 325, WTW). Dissolved oxygen measurements were undertaken by using a meter (Oxi 196, WTW) with an oxygen electrode (EO96, WTW) calibrated according to manufacturer's instructions.

Dissolved metal concentrations were determined by inductively coupled plasma atomic emission spectrometry (Spectroflame EOP, Spectro Analytical Instruments, Kleve, Germany) calibrated with matrix‐matched standards according to Simpson et al. [12]. Sediment pore waters were extracted using the procedure described by Simpson et al. [12]. In this procedure, sediment samples were centrifuged under a nitrogen atmosphere followed by membrane filtration (0.45 μm) of the displaced pore water.

Experiments on model pore waters

Model pore waters were prepared in 98% N_2(g)/2% H_2(g) in a glove bag fitted with a Pd/H catalyst for removing residual oxygen to subparts per million levels (Coy Laboratory, Grass Lake, MI, USA). Acidified (0.2% HNO₃) stock solutions of Cd, Cu, Ni, Mn(II), Pb, and Zn were prepared in deoxygenated Milli‐Q water. Iron(II) solutions were prepared in deoxygenated seawater (salinity 30‰) adjusted to the desired initial pH, immediately (<1 h) before experiments. Test solutions were prepared in low‐density polyethylene bottles (Nalgene, Interpath Services, Sydney, Australia) by spiking the desired metal ion into deoxygenated seawater in the absence or presence of Fe(II).

For pore‐water oxidation experiments, the dissolved metal concentrations in the model pore waters were measured after filtering the spiked samples through a 0.45‐μm cellulose acetate membrane filter (Minisart, Sartorius, Göttingen, Germany). Filtration was undertaken either inside the glove bag, or outside the glove bag immediately after swirling solutions for 5 s to expose to air, or at times of 2 min, 30 min, 2 h, 6 h, and 24 h after exposure to air. Experiments were performed in triplicate. Test solutions comprised either individual metal ions (Cd, Cu, Fe(II), Ni, Mn(II), Pb, or Zn), or Fe(II) and a secondary metal ion (Cd, Cu, Ni, Mn(II), Pb, or Zn). Dissolved metal concentrations measured over the 24‐h period were used to quantify the amount of Fe(II) that was oxidized and precipitated as Fe(III) hydroxide and the adsorption of metals onto these precipitates.

Effect of sediment storage on pore‐water metals

The effect of storage time on the pore‐water chemistry of whole‐sediment samples was investigated with sediments collected from Five Dock Bay (Sydney, New South Wales, Australia) on two occasions. Surface sediment samples were collected at low tide with a square‐mouthed shovel to retrieve a large volume from depths of 20 to 40 cm of water, without disturbing the surface layer of sediment. A clean plastic spoon was used to transfer the top 0 to 2 cm of the collected sample to a clean 50‐ml centrifuge tube (Cellstar, Greiner Labortechnik, Germany). Centrifuge tubes were completely filled with sediment (no air bubbles or head space), capped, and placed on ice until returned to the laboratory, where they were stored at 4°C until the time of analysis. Twenty sediment samples were collected for the first set of experiments and 40 samples were collected for the second set. For each experiment, pore waters were extracted from five to seven of the collected sediment samples (randomly selected) within 4 h of collection and at times 1 to 22 d after collection. Dissolved metals were determined on each of the samples.

Measurement of metal release rates from sediment cores

Sediment cores were collected at low tide (1‐ m tides) from Five Dock Bay nearshore sediments, by inserting acrylic corer‐reactor tubes (40 cm long, 15 cm diameter) to a depth of 15 cm into water‐covered sediment. Valve closure at the top of the tube created a sufficient pressure seal to allow the corer‐reactor to be removed intact with overlying water. Full tubes were capped at both ends and transported in an upright position to the laboratory. Two sediment cores were collected from sediments under approximately 80 cm of overlying water and four cores were collected from sediments with approximately 30 cm of overlying water.

In the laboratory, sediment corer‐reactors were stored upright in a controlled‐temperature environment (20 ± 2°C). The overlying waters of the sediments were stirred (30 rpm) and bubbled with air, maintaining a stable pH (7.8–8.0) and dissolved oxygen concentration (6–8 mg/L).

The renewal of overlying water was achieved by using a peristaltic pump to carefully drain and replace the 4 L of water with clean seawater (avoiding disturbance to the surficial layer). Subsamples were taken from the waters overlying the sediment cores at times of 4, 24, 48, 72, and 96 h after changing the overlying waters. All samples (filtered at 0.45 μm) were taken in duplicate and during the course of experiments removed less then 5% of the original overlying water. The pH and dissolved oxygen concentrations of the overlying waters were monitored daily.

To remove organisms from the corer‐reactor sediments, anoxic conditions were created by bubbling the overlying waters with nitrogen gas for two weeks (dissolved oxygen < 0.1 mg/L), followed by bubbling with air for two weeks for return to normal operation conditions. During the anoxic period, surfacing dead organisms were removed with minor disturbance and deoxygenated seawater was replaced regularly.

The effect of the sediment depth (chosen for whole‐sediment toxicity tests) on pore‐water chemistry was investigated by carefully transferring the top 2 cm of sediment from the sediment cores to a new container in which metal release could be measured.

The disturbances caused by bioturbating organisms on sediment pore‐water chemistry were investigated using benthic organisms, including the polychaete worm Australoneries ehlersi and the bivalve Soletellina alba, which were collected from the estuarine Woronora River (New South Wales, Australia).

RESULTS AND DISCUSSION

Experiments on model pore waters

The oxidation of Fe(II) in seawater, which had an initial pH of 7.7 and less than 0.1% dissolved oxygen (a simulated saline pore water), was examined at 2 and 8 mg Fe(II)/L (concentrations that could be considered typical of suboxic pore waters) at two temperatures of 4 and 20°C (to simulate the most common conditions of pore‐water storage). Samples were opened to the atmosphere and swirled for 5 s to allow air to intrude, then dissolved pore‐water metal concentrations were measured after 0.45‐μm filtration. Figure 1 shows that, as expected, the rate of Fe(II) oxidation is slower at the lower temperature. At 20°C, more than 90% of Fe(II) was lost from solution, probably through precipitation as Fe(III) hydroxide, with a half‐life, for 8 mg Fe(II)/L, of approximately 15 min at 20°C and approximately 35 min at 4°C.

The effect of the initial pore‐water pH, in the range 7.20 to 8.25, on Fe(II) oxidation in samples initially containing Fe(II) at 5 mg/L is shown in Figure 2. The rate of oxidation of Fe(II) was greater for waters of higher pH, and, as expected [14], the pH decreased as Fe(II) oxidation proceeded. When the oxidation was complete, the pH gradually increased as waters equilibrated with atmospheric carbon dioxide. The final equilibrium pH was lower than the initial pH.

The losses of Zn from simulated pore waters (pH 7.7) during Fe(II) oxidation are illustrated in Figure 3. In the absence of Fe(II), losses of Zn due to adsorption or precipitation from an initial dissolved concentration of 250 μg/L were not significant. For the same solutions containing Fe(II)/L at 2 or 8 mg, Zn concentrations decreased as Fe(III) hydroxide phases formed. At 20°C, the removal reached a plateau value after approximately 6 h, corresponding to a removal of 70 and 160 μg/L of Zn, respectively, for Fe(II) at 2 and 8 mg/L. The different losses reflect in part a greater number of binding sites, which are not linearly related to concentration because of possible aggregation. At 4°C, the loss of Zn was appreciably slower.

In simulated pore waters (pH 7.7) initially containing 5 mg Fe(II)/L and 200 μg/L of either dissolved Cd, Cu, Mn, Ni, Pb, or Zn, the losses of dissolved metals, upon exposure to air, were metal‐specific (Fig. 4). Again, in the absence of Fe(II), no decreases in metal concentrations were observed. In the presence of Fe(II), the concentrations of Cd, Ni, and Mn did not change as Fe(II) oxidation proceeded. The behavior of these metals was consistent with the known strength of adsorption of their ions to hydrous Fe hydroxide phases (Pb > Cu > Zn > Mn > Ni > Cd) [15,16].

The behavior of Cu was found to be strongly pH dependent. The losses of Cu from a simulated pore water initially containing Fe(II) at 5 mg/L and Cu at 200 μg/L, over the pH range from 7.20 to 8.25, are shown in Figure 5. As shown earlier (Fig. 2), Fe(II) oxidation also is dependent on pH. For waters where the pH is high (≥8), Fe(II) oxidation occurs rapidly and Cu is rapidly adsorbed to the precipitating Fe(III) hydroxide phases. However, for waters with a lower initial pH, both Fe(II) oxidation and Cu adsorption are slow. The pH dependence of Pb and Zn adsorption was not investigated; however, the position of the adsorption edge for these metals (near pH 7) would suggest that their adsorption also should be affected.

Effect of sediment storage and homogenization on pore‐water metals

The effect of sediment storage initially was investigated by using the top 2 cm of metal‐contaminated sediment samples collected from a nearshore, shallow (10‐ to 30‐cm water depth at low tide), intertidal site (1‐m tides). These sediments contained high concentrations of anthropogenic Zn contamination (500–7,000 μg/g) resulting from historical industrial discharges. Acid‐soluble Zn concentrations exceeded reactive sulfide concentrations (acid‐volatile sulfide) for most of the surface sediments (unpublished results). Waters overlying the sediments were near saturation with dissolved oxygen, and the surface sediments had pore‐water pH values in the range 7.2 to 7.5 and redox potentials in the range ‐150 to 50 mV (consistent with conditions for Fe and Mn (hydr)oxide reduction, but not for sulfate reduction).

A number of samples were collected in plastic centrifuge tubes with no head space as discussed earlier. Pore waters isolated immediately after collection from 7 of the 42 samples had dissolved Fe at 10 to 1,300 μg/L, Mn at 60 to 170 μg/L, and Zn at 160 to 1,100 μg/L. The variability in data is considerable, and was assumed to reflect the spatial heterogeneity of the sediments. This was surprising because the extraction process for each tube was already integrating pore waters from a significant surface volume. Mean pore‐water Fe, Mn, and Zn data for individual sediment samples collected from the same field site at the same time, but stored at 4°C (with no air exposure) for different times before pore‐water extraction, are shown in Figure 6. During storage, dissolved Fe and Mn concentrations in the pore waters increased, consistent with the reductive dissolution of Fe and Mn hydroxide phases. The increases in dissolved Zn concentrations were not significant when the variability was taken into account.

Plotting the pore‐water Zn concentrations against pore‐water Fe (Fig. 7) shows that the average pore‐water Fe concentrations increased, and the variability (standard deviation between sample concentrations extracted on the same day) increased for both metals. Examination of data for Mn showed similar variability (not shown). The duplicate experiment (20 samples collected two months earlier, with data not shown) showed similar trends in dissolved Fe, Mn, and Zn data and increasing variability with increasing storage time before pore‐water isolation.

In the two data sets comprising four and seven samples extracted per day, respectively, although the variability in pore‐water Fe data increased with increasing storage time, the percent variability (relative to the mean) decreased from more than 200% on day 1 to approximately 30% on day 22. In contrast, the percentage variability in dissolved Zn concentrations increased with increasing storage time from 15 to 85%, whereas the variability of dissolved Mn was relatively constant, and varied from 26 to 62% over the storage period. The variability in dissolved Fe and Mn may be related to different biogeochemical microenvironments existing in each of the individual samples. These microenvironments will exist as a result of bioturbation, which disturbs the vertical stratification of Fe and Mn predicted for undisturbed sediments [2,3,22,23]. The absence of Fe hydroxide precipitation at container walls indicated that diffusion of oxygen through the plastic centrifuge tubes was not sufficient to cause Fe(II) oxidation. The reason for the greater, and increasing, variability in dissolved Zn concentrations is not clear, but is likely to be due to the variability in Fe and Mn speciation on storage. The spatial variability in metals such as Zn in sediments has been observed previously, with the existence of discrete high metal‐containing microenvironments being revealed by diffusive gel samplers (H. Zhang, Lancaster, UK, personal communication).

When the pore waters isolated from the stored sediments were stored for 8 h under conditions used for the model pore‐water experiments (20°C, air exposure), losses of dissolved Fe and Zn occurred. Because of the higher Fe(II) concentrations in sediments that had been stored for longer periods (Fig. 6), greater losses of Zn were observed for pore waters isolated from sediments that had been stored for longer time periods (data not shown). This was consistent with the results from the model pore‐water experiments (Fig. 3). These experiments highlight the dynamic nature of sediment pore waters and indicate that oxidation and reduction processes do not necessarily cease while sediments are stored cold and that storage for long periods will result in large changes of pore‐water chemistry and the availability of metal toxicants.

To mimic the effect of sediment homogenization (in air) on pore‐water chemistry, the top 0 to 2 cm of sediment from two large‐diameter cores collected from the Five Dock Bay site were split into halves. On one half, the sediments were transferred to a centrifuge tube, the headspace was filled with nitrogen, and the pore waters were isolated under anoxic conditions. The other half of the sediment was homogenized for 2 min (with air exposure), allowed to sit for 3 min, and then the pore waters were isolated with the same procedure. In the homogenized sediments, dissolved Fe and Zn concentrations were found to be significantly lower (by 64 ± 24% for Fe and 41 ± 22% for Zn) and dissolved Mn concentrations were higher (by 335 ± 198%) than in the nonhomogenized samples. Pore‐water Fe and Zn losses (occurring within 5 min) were greater than those observed for pore waters isolated from un‐homogenized, stored sediments (18 ± 13% for Fe and 10 ± 8% for Zn), indicating that other constituents in the sediments also may contribute to oxidative precipitation of Fe(II) and adsorptive losses of Zn.

The observed increase in pore‐water Mn for the homogenized sediments when compared to the nonhomogenized sediments is consistent with Mn (hydr)oxide phases in the sediments being reductively dissolved when they are brought into contact with pore‐water Fe(II). This reaction, shown in Equation 2, is widely recognized for its importance in the cycling of Fe and Mn in sediments [9]. This reaction is important because it applies even when sediments are homogenized under anoxic conditions

The findings of these experiments clearly demonstrate that disturbing sediments by homogenizing or even during the practice of sieving to remove larger particulates has the potential to significantly alter pore‐water concentrations of metal contaminants from the concentrations to which organisms might be exposed in the undisturbed sediments.

Metal release from sediment cores

Changes to sediment chemistry are likely to occur when moving sediments from the field to the laboratory as a result of alterations to the depth (pressure) and dissolved oxygen concentration of the overlying waters. These chemical changes were simulated by using sediment cores (15 × 15 cm) collected from Five Dock Bay at low tide from a water depth of 80 cm (1.8 m at high tide). The sediment cores were maintained in stirred reactors overlain by 22 cm of seawater (4 L), which was continuously bubbled with air, slowly stirred, and replaced with clean seawater at the start of each week. The Zn concentration of the overlying water measured over the first 96 h of each week, for five weeks after collection, increased rapidly initially and began to plateau as the overlying water concentration approached that of the released pore water (Fig. 8). The rate of Zn release and the plateau Zn concentration increased in successive weeks, consistent with significant changes to Zn speciation occurring in the surface sediments. This suggests that changes to dissolved oxygen concentrations in overlying waters, caused by different laboratory conditions to those in the field, are sufficient to have marked effects on the chemistry of surface sediments.

The pore‐water Fe(II) concentrations ranged from 300 to 1,200 μg/L in the surface sediments at this site. The observation that Zn release rates increased, rather than decreased, upon exposure to more oxygenated waters indicated that oxygen penetration into the sediments did not increase sufficiently to cause oxidation of pore‐water Fe(II) and adsorptive losses of Zn from the pore waters. The increased rate of Zn release from the sediments most likely is associated with the slow oxidation of Zn sulfide phases present in the surface sediments (0–2 cm). Thermodynamics predict that sulfide oxidation should occur in preference to Fe(II) oxidation. Measurements of acid‐volatile sulfide (20–30 μmol/g) and simultaneously extractable Zn concentrations (45–50 μmol/g) in the surface sediments indicated an excess of Zn over acid‐volatile sulfide (unpublished data), supporting the observed Zn release, although the chemical form of Zn was not known. The exposure of the sediments to fully oxygenated waters would be likely to cause oxidation of reactive sulfide phases (acid‐volatile sulfide) in the sediments and shift the redox boundary position for sulfate reduction to deeper into the sediments. Previous studies have shown that Zn sulfide oxidation occurs slowly over days [24].

The implications of these findings are that sediment oxidation can result in high amounts of metal release if a simultaneous increase in binding sites (e.g., Fe(III) hydroxide) does not occur. For whole‐sediment toxicity testing, the recommendation is made that metal concentrations in the waters overlying test sediments should be monitored during tests and, if necessary, overlying waters should be changed regularly to achieve concentrations similar to those expected in an environment where overlying waters are regularly flushed.

Effect of sediment depth (2 or 15 cm) used in toxicity tests

The changes to sediment chemistry associated with the common procedure of using only the top 2 cm for toxicity tests were investigated by using sediment cores collected from water depths of 30 cm at low tide. The measured Zn release rate from these sediment corer‐reactors had been relatively constant for three weeks. The sediment surface layers were removed and placed in a second reactor, the overlying seawater was replaced, and Zn release measurements were commenced (Fig. 9). Even with considerable care not to disturb the surface sediments when transferring to the new container, a much lower Zn release rate was measured for the week immediately after transferring. For weeks 2 to 5 after transferr, the Zn release rates remained lower than those measured from the original sediment core. This indicates that for Fe(II)‐rich sediments, even minor disturbances will cause major changes to metal partitioning. Equilibria are established slowly and the redox conditions observed in the 15‐cm core are not established in the thin 2‐cm core section. Sediments in the 15‐cm core become increasingly reduced at greater depths. In the 2‐cm core, oxygen penetration from the overlying water, and also from slow diffusion through the plastic below, will result in the redox potential being higher than that of the top 2 cm of the 15‐cm core. The 15‐cm core effectively has a greater buffering capacity for maintaining reduced conditions. The increased redox potential in the 2‐cm core causes lower Fe(II) concentrations and a greater amount of Zn adsorbed to freshly precipitated Fe hydroxide phases.

Bioturbation

In organism‐free homogeneous sediments, bacterially catalyzed redox reactions cause the vertical stratification of redox‐active species (e.g., Fe, Mn, and S) [2]. The actions of bioturbation by organisms and tidal currents intermix sediments and result in a mosaic of biogeochemical microenvironments [2,5,6,22]. These processes cause oxidized conditions to fluctuate with dissolved oxygen pulses [6,7]. Sulfide concentrations remain negligible and suboxic, Fe(II)‐rich conditions extend to depths of 2 to 3 cm in surface sediments. In bioturbated sediments, chemical species considered incompatible according to thermodynamic calculations will appear to coexist in this three‐dimensional environment. Because the concentrations and distributions of many metals in sediment pore waters are influenced by adsorption and precipitation with Fe hydroxides, disturbances caused to these equilibria through sample collection, manipulation in the laboratory, or while conducting toxicity tests, will affect test outcomes.

The effect of bioturbation on sediment chemistry was investigated by adding sediment‐dwelling biota (five bivalves [S. alba] and five polychaete worms [A. ehlersi]) to sediment corer‐reactors from which native organisms previously had been removed and the Zn release rate had remained relatively constant for the previous three weeks (Fig. 10). For the two weeks after the addition of the organisms, Zn release rates were much lower than observed before organism addition. The lower release rates were attributed to sediment disturbances by the organisms causing Fe(II) oxidation and adsorption of pore‐water Zn to precipitated Fe(III) hydroxide phases, analogous to the experiments investigating sediment homogenization effects on pore‐water metals. When the organisms appeared to have established themselves in the sediments, as indicated by their moving about less, the Zn release increased again and became greater than the release rates measured in the absence of organisms. An important observation from these experiments was the long time required for sediment chemistry to recover from initial disturbances caused by adding organisms to sediments (as may occur at the start of whole‐sediment toxicity tests). Week by week, Zn release from sediments was more variable in the presence of organisms, when compared to the relatively constant Zn release for the weeks before organisms were added. For all of the experiments described, duplicate experiments (data not shown) gave similar trends. The observations are consistent with past studies that showed significant changes to pore‐water Fe(II) concentrations due to sediment‐dwelling biota [3,22,23].

CONCLUSIONS

Implications for toxicity testing

The experiments discussed above indicate how sensitive the chemical behavior of metals in sediments and pore waters is to the operations of sampling, sample storage, and sample handling of sediments as part of the process of setting up whole‐sediment or pore‐water toxicity tests (Table 1). In conducting laboratory sediment bioassays, the dominant concern will be for the achievement, in the laboratory, of conditions that are as close as possible to those in the natural field sediments.

Isolated pore waters are likely to have quite different metal contents in the sediments in situ, whereas the manipulation of sediments also can disturb metal partitioning. The suitability of isolated pore waters for toxicity testing when using non‐benthic biota is therefore questionable. Whole‐sediment tests are clearly more appropriate.

In laboratory test tanks, microbiological activity in the sediments is likely to be sustained, and after allowing sufficient time for the sample to reequilibrate (with either continuous or regular changes of natural and aerated overlying water), it is likely that conditions as close as possible to a field environment can be reestablished. Note that overlying water metal concentrations may become quite unrealistically high if waters are not changed.

The above protocol applies only to shallow subtidal environments. Experiments on shoreline sediments are more difficult, although tidal machines have been used to simulate the natural tidal covering and uncovering of sediments with water [25]. The difficulty in achieving near‐natural conditions is also apparent for sediments collected from significant depths, where consolidation of the hydrous upper sediment layers will occur during sampling or in the laboratory.

In their natural environment, benthic organisms will alter sediment and pore‐water chemistry. In this dynamic environment, what an organism is exposed to will depend on its behavior, the nature of the biogeochemical microenvironment it is in contact with, and the external forces acting on both of these components, for example, the presence of other organisms, the availability of food, tidal currents and overlying water conditions. The feeding strategies of sediment‐dwelling organisms, the types of habitats that they construct (if any), and their tolerance to conditions of low dissolved oxygen are probably the most important in determining the contaminant concentrations that organisms ultimately are exposed to [2,3,22,23,26].

Important factors will be where organisms live in the sediment (surface or depth), and how they burrow and how well the burrows are irrigated. These factors also will be dependent on how favorable the sediment grain size is, as a habitat in which organisms can burrow easily and seek refuge from which they can obtain food. The source of food will depend on the organism and could include combinations of plant material or organisms obtained from pore waters or sediments.

The chemistry of contaminants in the sediments and pore waters will be altered as a function of the many physical parameters. The present study indicates that active bioturbation (burrowing and irrigation activities) of sediments, which increases oxygen penetration into Fe(II)‐rich pore waters, will decrease organism exposure to pore‐water Pb, Cu, and Zn but not Cd, Ni, or Mn. During toxicity tests, the pore‐water concentrations of these metals will vary to different degrees depending on what organisms are present. The effect of changes to pore‐water pH on metal toxicity, as a result of storage of isolated pore waters, sediment handling, or organism activity, also will vary with the test procedure and organism used [27].

Bioturbation activities are natural disturbances in a field situation that might affect contaminant bioavailability and toxicity to a benthic organism. Situations where a test organism behaves unnaturally in a laboratory situation such that the chemical changes are more extreme would be a concern. Examples of this are given earlier in this paper, where the introduction of bivalves and worms into a sediment initially resulted in high bioturbation activity that oxidized sediments and lowered Zn release compared to that in the absence of organisms. After several weeks, when the habitats were established, the release of Zn increased again to amounts similar to that in the absence of organisms. Other examples of stressed behavior could include those resulting from a confining test vessel where an organism in the act of burrowing continuously contacted the container walls. Table 2 lists some possible effects on sediment chemistry that may occur through the natural activities of test organisms.

Metal speciation changes are not isolated to disturbances to sediments with Fe(II)‐rich pore waters. The mixing of sediments containing microenvironments of Fe and Mn oxides in predominantly sulfidic sediments will cause oxidation of sulfide phases [13,28]. Bioturbation may also achieve this in anoxic sediments [26].

Recommendations for sampling pore waters and sediments for toxicity testing

Although we have questioned the use of pore waters for toxicity testing, if these are to be collected, we recommend their isolation by centrifugation [29]. Sample containers (ideally those that can be transferred immediately to a centrifuge) should be completely filled with sediment (no headspace) and centrifuged under nitrogen. Squeezing sediment under nitrogen is an option; however, centrifugation techniques cause less disturbances (in transferring), are often faster, and oxidation artifacts are more easily avoided. Note that bulk pore waters isolated in this way will represent an average of the field composition. All information of pore‐water gradients in the sediments is lost. Iron(II) present in extracted pore waters will precipitate and adsorb many metal (and organic) contaminants.

Care should be taken with the use of equilibrium‐based pore‐water samplers such as peepers in estuarine environments where fluctuating redox conditions in surrounding sediments may result in Fe(II) oxidation and Fe(III) hydroxide precipitation after Fe(II) diffuses into the peeper chamber. Iron(III) hydroxide phases often are slower to reduce, and accumulate as fine solids that are not always obvious upon sampling. Small Fe precipitates within peeper chambers adsorb large amounts of metals that will be desorbed upon sample acidification for preservation.

When collecting surface sediments from depth (> 1 m) for toxicity test purposes, we recommend first the collection of a bulk sediment sample with an undisturbed surface layer. This can be achieved by using a box corer (or other devices that do not disturb surface sediments). A cylindrical corer that can be converted into a test chamber, the so‐called corer‐reactor, is a useful alternative.

For whole‐sediment toxicity tests, sediments should be stored for a minimum amount of time before undertaking the tests. Where sediment sieving or homogenization is necessary, some equilibration period may be necessary before starting tests, but this can only be evaluated from a knowledge of the chemistry of the sediments in their natural environment. Separate sediments should be collected for pore‐water isolation and analyses of dissolved constituents and pH. Measurements of the pH and the redox potential of the sediments (usually surface) should be made in the field. Waters directly overlying the sediments should be collected from the site for analyses of dissolved constituents and measurements should be made of temperature, salinity, pH, dissolved oxygen, and redox potential in the field. All of these measurements should be made on overlying waters and sediment samples immediately before commencing, during, and at the end of toxicity testing procedures. All measurements should be reported so that the similarity between laboratory conditions and those in the field can be clearly judged.

Acknowledgements

We thank Karl Bowles for constructive comments on this article. We thank Ian Pryor and Mylissa Dowse for assistance with experiments.

REFERENCES