Relating the sediment phase speciation of arsenic, cadmium, and chromium with their bioavailability for the deposit-feeding polychaete Nereis succinea

Abstract

We studied the influence of sediment geochemistry on bioavailability of As, Cd, and Cr in deposit-feeding polychaetes. Metal phase speciation in sediments was determined with a sequential extraction scheme, and assimilation efficiencies (AEs) of ingested metals were determined by pulse-chase feeding experiments using γ-emitting isotopes. Worms were fed sediments collected from geochemically diverse estuaries that were labeled by sorbing dissolved radiotracers or mixing with radiolabeled algae. Uptake of sediment-bound metals was compared with that from labeled algae or goethite. Metal AEs showed a positive relationship with the exchangeable and carbonate sedimentary fractions, whereas metals in iron and manganese oxides and acid-volatile sulfides, or in pyrite and other refractory material, were inversely correlated with AEs. Arsenic was most bioavailable from algae (72%), less from sediments mixed with algae (24–70%) and least from sediments labeled directly (1–12%). Arsenic AEs in sediments labeled directly showed a positive correlation with sedimentary Mn and Al and negative correlation with Fe. Cadmium AEs were positively correlated with salinity and negatively correlated with sedimentary organic C. The AEs of Cr from sediments or algae were less than 5%, but they were 34% from pure goethite. By quantifying the relationship of metal speciation in sediments with their bioavailability for deposit-feeding polychaetes, the present study provides new insight into understanding metal bioaccumulation in benthic invertebrates. Environ. Toxicol. Chem. 2011; 30:747–756. © 2011 SETAC

INTRODUCTION

Elevated concentrations of trace metals are found associated with sediments on the floors of rivers and their estuaries, often accumulated as historical pulses several centimeters below the surface. Data on total sediment concentrations of specific metals are valuable but by themselves may be of limited use in evaluating the risks associated with this contamination. Concern over elevated metal concentrations stems primarily from the risks these contaminants can pose for living organisms, including people who might consume contaminated seafood. Because organisms must first accumulate metals before any toxic effect can be manifested, one must assess the extent to which the metals bound to sediments are accumulated in benthic animals. Contaminated sediments may be a dominant source of metals for benthic animals 1 if the contaminants are in a form that can be accumulated into biological tissue. The non-bioaccessible contaminants irreversibly bound to sediment 2 are not assimilated into biota and should have little ecological impact. The extent to which sediments can serve as sources of contaminants for marine organisms and the mechanisms responsible remain understudied for most sediment types and organisms. Complicating factors include metal partitioning into various sedimentary mineral and organic phases that may affect the uptake and assimilation of the metals by benthic organisms. Metals associate with these phases through surface adsorption, co-precipitation during diagenetic formation, or actual incorporation, such as covalent bonding in organic matter or precipitation into insoluble metal sulfides.

The chemical speciation of a metal dissolved in water and its solid phase speciation in sediment both can influence its bioavailability 3, 4. In this respect, the total concentration of a sedimentary metal could have little relevance to its bioavailability, just as is the case with dissolved metals. The bioavailability of different metals in contaminated sediments is likely to be a function of a metal's characteristics—for example, charge, ionic radius, and oxidation state, its phase speciation in the sediment, and the physiological and ecological characteristics of the organism inhabiting the sediment 5–8. Consistent with another study that related phase speciation with bioavailability, metals bound to more refractory fractions such as pyrite or part of the sediment matrix itself would be expected to be less available for animals than metals loosely bound to sediment particle surfaces in exchangeable and carbonate fractions 9.

Bioavailability of sedimentary metals has been assessed in light of abiotic factors such as ratios of simultaneously extracted metals to acid-volatile sulfides (AVS) 7, 10, metal partition coefficients (K_d)—often related to sediment grain size and organic matter content 9, pore-water metal concentrations 10, and geochemical heterogeneity in an animal's immediate microhabitat 11. Biological factors that influence the assimilation efficiency (AE) of ingested metal, a key parameter in kinetic or biodynamic bioaccumulation models 12, include ingestion rate 13, 14, gut passage time 9, and the gut environment 15–17, which determines the amount of metal that can be freed from particles into the gut fluid. For example, gut surfactants help extract organic matter coating mineral particles in the sediment 18, and anoxia can help reduce iron oxides, thereby releasing metals bound to them 15.

The assimilation of metals from contaminated sediment in deposit-feeding invertebrates such as polychaetes can effectively transfer the metal from abiotic substrate to living tissue, which can subsequently be transferred to their predators. To provide a better mechanistic understanding of metal transfer into benthic food chains from sediments, the present study involved a series of experiments that used gamma-emitting radioisotopes to evaluate the bioavailability to deposit-feeding polychaetes of sedimentary As, Cd, and Cr, all of which often display high concentrations in industrialized estuaries, sometimes reaching concentrations that are toxic to resident organisms 1. These three metals have contrasting biochemical associations in living organisms, contrasting geochemical associations in sediments, and divergent particle reactivities and residence times in aquatic systems 19. Although each of these metals is an environmental contaminant of concern in its own right, simultaneously studying their behaviors can reveal the behaviors of other metals with similar divergent characteristics. Furthermore, As and Cr exist as oxyanions in solution but also have multiple oxidation states that vary with the oxygen content of the water and sediments. In oxic sediments arsenate is known to bind with Fe and Mn oxides, and with pyrite in reduced sediments 20. Chromium III and VI both associate with organic matter, although Cr III is far more particle-reactive. In contrast to As and Cr, Cd is a nonredox cation in water that is strongly chloro-complexed 21. Each of these metals shows distinct patterns of mineral/organic association with particulate matter and distinct biological behavior.

Deposit-feeding polychaetes ingest large amounts of sediments 22, whose nutritional value is typically low in comparison with algae or bacteria but that can vary seasonally with productivity in overlying water. Metals associated with sinking biogenic debris can eventually associate with various geochemical fractions in the sediments. Thus, metals can loosely associate with the surfaces in the exchangeable or carbonate phases, bind to or precipitate with Fe/Mn oxides, and bind to less labile organic matter or iron sulfides, including AVS and pyrite, in anoxic sediments. Typically only a small fraction of sedimentary metals loosely associates with particles (e.g., exchangeable) and metals that are associated with oxygen- or pH-sensitive phases can be released into the solution when these conditions shift.

The assimilation efficiencies of ingested As, Cd, and Cr from surface sediments collected from two sites in the Chesapeake and one in San Francisco Bay in deposit-feeding polychaetes was compared with AEs from the same sediments amended with organic matter from algal debris, pure algal detritus, and from goethite. These AEs were regressed against the geochemical fraction patterns of these metals in the sediments. The three study sites were chosen based on differences in their sediment composition, such as organic carbon content, degree of metal contamination, and S content.

Pulse-chase feeding experiments using radiotracers have been used to determine metal AEs and efflux rates out of animals. These parameters are components of a metal bioaccumulation model that can evaluate the relative importance of dietary and aqueous sources of metals for aquatic animals and enable predictions to be made of steady-state metal concentrations in animal tissues in different geographic locations 12. Modeling has shown that many metals are found to be accumulated significantly from diet in diverse invertebrates 13. The application of gamma-emitting radioisotopes in bioaccumulation experiments provides the advantage that environmentally realistic metal concentrations can be used, and analyses are rapid, accurate, and precise. Previous studies have evaluated AEs of metals bound to different types of sediment in diverse marine invertebrates 8, 9, 15, 23, but most earlier studies did not relate AEs with metal fractionation patterns in sediments.

MATERIALS AND METHODS

Choice, collection, and handling of the test organism

The present study used the surface deposit-feeding polychaete Nereis succinea, ubiquitous in muddy sediments along the United States eastern coastline. This species, whose ecology and physiology are well described, was used previously in experiments that studied contaminant bioaccumulation 14. Worms of similar size (∼10 cm) were hand collected from a local salt marsh at Flax Pond from early spring to late fall. Individuals were placed with a small portion of sediment in separate containers to avoid cannibalism, because although N. succinea feeds on surface sediment, it is predatory when given the opportunity. Animals were transported to the laboratory, rinsed with Flax Pond water, and placed in clean containers with Flax Pond water (salinity 28) and approximately 0.5 g (wet wt) of Flax Pond sediment. Before each feeding experiment, approximately 20 individuals that were regularly producing fecal pellets were selected and further acclimated for 2 to 7 d (depending on water salinity) to the experimental conditions. The final selection of worms (n = 5–8 individuals) that were fed radiolabeled sediment was based on the presence of feces in their chambers.

Pulse-chase feeding experiments

Feeding experiments were conducted at 21°C by placing individual worms in feeding chambers that consisted of two small (ϕ´ = 5 cm) plastic Petri dishes connected by tygon tubing (length =15 cm; ϕ´ = 4 mm). Each chamber was filled with water, and approximately 0.5 g radiolabeled sediment was placed at the head end of the tube; fecal pellets were collected from the other end of the tube 22. After worms were placed in the chambers, water was periodically changed or filled to the top, and feces were removed.

Study sites

Sediments and water used for feeding experiments were collected from three locations, two sites in Chesapeake Bay and one in San Francisco Bay. They were Baltimore Harbor ([BH]; 39°12′ 25″N, 76°31′40″W) (Baltimore, MD, USA), Elizabeth River (ER; 76°20′ 09″ W, 36°52′ 32″ N) (Norfolk, VA, USA), and Mare Island ([MI]; 38°05′15″ N, 122°15′15″ W) (Vallejo, CA, in San Francisco Bay, USA). Sediments from Norfolk were collected in May, BH in June, and MI in October. After collection they were stored at 4°C in plastic containers for up to two years before experimentation, where deeper layers of the sediments stored in the bucket were anoxic. Sediment location choices were based on differences in their geochemical properties and extent of metal contamination. For example, BH sediments had the highest content of organic carbon and nitrogen (5 and 0.33%) compared with ER (2 and 0.12%) and MI (1.5 and 0.12%), but lowest salinity (8.5) compared with ER (19.5) and MI (23). The greatest degree of contamination with Cr and As was in BH (Cr: 322.6 and As: 47.19 µg/g dry wt; ER—Cr: 33.3 and As: 6.37 µg/g dry wt, and MI—Cr: 47.4 and As: 1.43 µg/g dry wt), and with Cd in MI (2.39 µg/g dry wt; BH—0.96 and ER—0.46 µg/g dry wt) (Z. Baumann et al., unpublished data).

Chemical analyses

Measurements of organic carbon, organic nitrogen, and sulfur were conducted using a Carlo Erba 1500 Elemental Analyzer on sediment samples that were dried (50°C), ground, and sieved through a 150-µm nylon mesh screen 24. Such prepared sediments were also used for metal concentration analyses. Sediment subsamples were digested in two steps with trace metal-grade concentrated nitric acid and then concentrated perchloric acid in a boiling bath. All of the sediment manipulations were conducted in a clean laboratory and under a high-efficiency particulate air laminar flow bench. Sediment digest solutions were analyzed using inductively coupled plasma mass spectrometry (Finnigan Element 2).

Food types and radiolabeling

Feeding experiments considered four different foods, each uniformly radiolabeled with gamma-emitting isotopes, either ⁷³As (V) (half-life=80.3d) or a combination of ¹⁰⁹Cd (half-life =461.4d) and ⁵¹Cr (III) (half-life = 27.7d): first, unamended sediment from the upper 10 cm at each site, mixed thoroughly immediately before radioisotope additions; second, sediment to which radiolabeled diatom debris was added and mixed in; third, pure diatom debris; and fourth, pure goethite purchased from Sigma-Aldrich (goethite ∼35% Fe; EC No. 2437464). Radioisotopes were added as arsenate for ⁷³As, cadmium chloride dissolved in 0.1 M HCl, and chromic chloride dissolved in 0.5 M HCl.

Unamended sediments were radiolabeled by a direct addition via pipette of radioisotope dissolved in dilute HCl; microliter quantities of dilute NaOH were immediately added to the unamended sediment so that the pH was not affected by isotope addition. Amended sediments received radioisotopes by mixing 22.3 ± 4.5 mg (dry wt) previously radiolabeled diatoms (Thalassiosira pseudonana, clone 3H) with approximately 2 g (wet wt) sediment. The ratio of dry to wet weight for all the sediments ranged between 0.7 and 0.5 as determined by Baumann et al. (unpublished data). Both types of sediments were aged at 21°C before the experiment for 2 or 30 d to assess the influence of sediment aging on metal partitioning and bioavailability over this period. Given the radioactive half-lives of the isotopes, conducting experiments involving bioavailability of all three isotopes for longer time periods was not practical.

Radiolabeled T. pseudonana was produced as described previously 12. Because the assimilation efficiencies of ingested metals in marine invertebrate herbivores from phytoplankton diets reflects the cytoplasmic distribution of metals in the algal cells 19, 25, the cellular distribution of each metal in aliquots of the diatom cells was determined by differential centrifugation of algal components after cells were broken 25. Radioactive diatom cells were harvested and mixed with sediment before feeding to worms. Radioactive diatoms were first harvested by filtration on 1-µm polycarbonate membranes, then resuspended in 10 ml seawater and centrifuged at 840 g for 5 min; the resulting radioactive pellet containing the cells was mixed thoroughly with the sediments or fed without sediment to the worms. Goethite (2.5 g dry wt) powder was radiolabeled after suspension in 2 ml seawater. Radioactive goethite after thorough mixing was briefly centrifuged to remove excess seawater, and then goethite pellets were fed to N. succinea. For sediments that received direct addition of dissolved radioisotopes, a small amount of dilute sodium hydroxide was added to neutralize the acid associated with radioisotope additions (dissolved in dilute HCl). The radioactivity and the added metal concentration in each food, determined by using the specific activity of each added radioisotope, are given in Table 1. The radioisotope additions contributed only a small fraction of the measured background concentrations of these metals in surface sediment (<<1%) (Z. Baumann et al., unpublished data). Carbon additions in the form of added algal debris were 8% of background organic matter in the BH sediment, 19% in ER sediment, and 27% in MI sediment.

Table 1. Radioactivity added to natural sediments, pure algae, and goethite used for the pulse-chase feeding experimenta

Location	Label	Age (days)	kBq/g (wet wt)			pmoles/g (wet wt)
			^73As	^109Cd	^51Cr	^73As	^109Cd	^51Cr
a ND = not determined.
Elizabeth River (ER), VA, USA	direct	2	9	102	15	0.15	9.76	0.09
		30	ND	9	28	ND	0.86	0.16
	algae	2	3	11	6	0.05	1.05	0.03
		30	10	55	31	0.17	5.26	0.18
Baltimore Harbor (BH), MD, USA	direct	2	7	44	17	0.12	4.21	0.10
		30	1	13	2	0.02	1.24	0.01
	algae	2	4	8	6	0.07	0.77	0.03
		30	5	4	2	0.08	0.38	0.01
Mare Island (MI), CA, USA	direct	2	11	169	4	0.18	16.17	0.02
		30	9	10	3	0.15	0.96	0.02
	algae	2	12	72	8	0.20	6.89	0.05
		30	3	62	19	0.05	5.93	0.11
Pure algae			3	12	87	0.05	1.15	0.50
Goethite			32	13	16	0.53	1.24	0.09

Measurement of radioactivity

Radioactivity of worms was measured after 1 to 6 h feeding, depending on the presence of feces in the chamber. Worms were removed from their feeding chambers and rinsed three times with filtered seawater and twice with an ethylenediaminetetra-acetic acid solution in seawater (10⁻⁴ M) to remove adsorbed metal and adhering particles. To assay their radioactivity, individual worms were placed into 50-ml plastic containers, filled with a small amount of water to assure their position on the bottom of the container, and inserted into a well-type NaI(Tl) gamma detector. Counting times were typically 5 min, yielding propagated counting errors that were typically less than 5%. Radioactivity of ⁷³As was detected at 53 keV, of ¹⁰⁹Cd at 88 keV, and of ⁵¹Cr 256 keV. This kind of radioactive counting is nondestructive, so individual worms could be counted repeatedly over different periods. For all radiolabeled worms, after counting, individuals were returned to their feeding chambers, which were filled with new water and nonradioactive sediment that they could ingest to purge their guts of unassimilated radioactive material. After determining the radioactivity of worms after their radioactive feeding, the retention of the radioisotopes postfeeding on nonradioactive food over time in the individual worms was determined to quantify the AEs of the ingested metals for each treatment, as described in Wang and Fisher 13.

Sequential extraction procedure

Sedimentary metal phase speciation was evaluated using a modified scheme originally described by Tessier et al. 26 and Huerta-Diaz and Morse 20. We determined seven sedimentary fractions, nominally identified as exchangeable, carbonate, acid-volatile sulfides, Fe/Mn oxides, two organic pools, and pyrite. These fractions were determined chemically, and names assigned to each of them are strictly operational; details are given elsewhere (Z. Baumann et al., unpublished data). Briefly, metals in the exchangeable pool were extracted for 1 h by 1 M MgCl₂, in carbonate for 1 h by sodium acetate solution at pH = 5, for 0.5 h in AVS by 0.5 M HCl. Sediments were incubated for 6 h in a hot reducing solution of hydroxylamine to dissolve the Fe/Mn oxides and metals associated with them. Metals in the organic fractions were extracted first for 8 h by a hot 1 N NaOH solution and later concentrated H₂SO₄ for 6 h. Metals remaining in sediments, thought to associate with pyrite, were extracted for 2 h with 11 M HNO₃. After each extraction step, the extractant was separated from the sediment by centrifugation at 834 g for 10 min and transferred to a separate container. This extract was then analyzed for the amount of radioactivity attributable to each of the elements. After the final extraction, a small residual fraction of radioisotopes remained in sediment.

The geochemical fractionation of As, Cd, and Cr in operationally defined sedimentary fractions (Z. Baumann et al., unpublished data) was related to AE values for all metals, sediment locations, label types, and ages.

RESULTS

Assimilation efficiencies

Individual worms feeding on radiolabeled sediments ingested at least 1 mg wet wt of sediment, corresponding to 1 to 32 Bq ⁷³As (or 0.02–0.53 fmol of As), 4 to 169 Bq ¹⁰⁹Cd (or 0.38–16.17 fmol of Cd) and 2 to 87 Bq Cr (or 0.01–0.50 fmol Cr). In nearly all cases, sufficient radioactivity was found in each worm to measure the depuration rates of radioisotopes from individual worms over time. Assimilation efficiencies of ingested metals, determined by analyzing the loss patterns of metal during depuration, were determined for all experimental treatments; Figure 1 shows representative results of pulse-chase feeding experiments in which the retention of initially ingested radioactivity in N. succinea over time after feeding on radioactive food is given for ⁷³As, ¹⁰⁹Cd, and ⁵¹Cr. Typically a two-phase loss pattern is observed, a sharp decline within 24 h reflecting loss of unassimilated metal via defecation, followed by a much slower loss, representing physiological turnover of assimilated metal 13. It is evident that N. succinea generally retained more As and Cd than Cr, resulting in correspondingly higher AEs for As and Cd (Table 2). Assimilation efficiencies of Cr in most cases were very low (<5%), except when worms were fed pure radiolabeled goethite, yielding AEs of approximately 34% (Table 2). Significant differences were seen in AEs for metals that were labeled via direct addition of radioisotopes to the sediment or via addition of radioactive algal debris to the sediment. The AEs of As were much higher for all sites when worms fed on sediments with added algal debris than on unamended sediment (Table 2). These differences were between sediments from BH and MI that were labeled with ¹⁰⁹Cd aged for 2 d (one-way analysis of variance [ANOVA]: p < 0.01), and between 2- and 30-d-old MI sediments labeled with Cr (one-way ANOVA: p < 0.05). Significant differences in metal AEs were noted for As between all sample sites and for Cd and Cr between some sites (one-way ANOVA: p < 0.01; Cd: BH vs ER and BH vs MI; Cr: BH vs ER and ER vs MI). Generally, AEs of As and Cr decreased (one-way ANOVA: p < 0.01) from sediment with aging, regardless of whether the sediments were amended with diatom debris.

Figure 1. Metals retained in Nereis succinea after feeding on a pulse of 2 and 30 d old radiolabeled sediment from Baltimore Harbor, MD, USA. Data points denote mean values (n = 5–8 individuals) of percent metal retained over time, and error bars represent 1 standard deviation. Open circles indicate data for sediment mixed with radiolabeled algal detritus, and solid circles indicate data for sediments labeled via direct injection of isotope.

Table 2. Assimilation efficiencies (AE% ± standard deviation) of As, Cd, and Cr in Nereis succinea (n = 5–8), when diet containing those elements consisted of natural sediments that were radiolabeled by a direct addition of radioisotopes or by mixing the natural sediment with previously radiolabeled algal detritus, pure radiolabeled algal detritus, or goethite labeled by a direct addition of radioisotope.a Means without error indicate all worms for a given treatment were pooled for radioanalysis.

		Unamended sediment		Sediment mixed with algae
		2 d	30 d	2 d	30 d
a ND = not determined.
Elizabeth River, USA	^73As	1.2 ± 1.05	ND	69.7 ± 9.7	16.8 ± 4.1
Baltimore Harbor, USA		7.8 ± 6.2	6.59	51.6 ± 11.6	30.1 ± 1.4
Mare Island, USA		10.2 ± 6.8	12.1 ± 12.5	50.7 ± 9.0	24.0 ± 12.2
Pure algae (not mixed with sediment)				72.41 ± 3.30
Goethite		2.48 ± 0.68
Elizabeth River	^109Cd	30.8	43.6 ± 16.4	9.9 ± 3.5	21.5 ± 6.1
Baltimore Harbor		1.5 ± 1.29	2.4	68.7	21.6 ± 14.9
Mare Island		46.1 ± 18.7	58.9 ± 6.6	9.4 ± 4.0	7.6 ± 4.6
Pure algae (not mixed with sediment)				22.91 ± 19.73
Goethite		24.15 ± 1.78
Elizabeth River	^51Cr	4.5	1.0 ± 0.5	0.9 ± 0.2	3.6 ± 1.7
Baltimore Harbor		4.2 ± 3.3	4.6	1.2	0.8 ± 1.0
Mare Island		4.0 ± 3.1	0.7 ± 1.3	5.0 ± 2.7	0.3 ± 0.6
Pure algae (not mixed with sediment)				2.8 ± 1.6
Goethite		34.2 ± 6.5

Subcellular distribution of metals in Thalassiosira pseudonana

The cellular distribution of As, Cd, and Cr in the radiolabeled diatoms indicated that 83% of Cr, 55% of As, and 43% of Cd were bound to cell surfaces (Table 3). Assimilation efficiencies of As in N. succinea that fed on pure algal debris were much higher than when fed goethite (72 ± 3 vs 2.5% ± 0.7; Table 2). The opposite trend was apparent for Cr (2.8% ± 1.6 vs 34 ± 6; Table 2). For Cd, the method of labeling the sediment had no significant effect on AEs (23% ± 20 vs 24 ± 1.8).

Table 3. Cellular distribution (% ± standard deviation) of As, Cd, and Cr in Thalassiosira pseudonanaa

Subcellular fraction	As	Cd	Cr
a 1st pellet—nuclei, plasma membranes, cell walls; 2nd pellet—mitochondria, lysosomes, peroxisomes; 3rd pellet—endoplasmic reticulum, Golgi bodies, ribosomes, polysomes; supernatant-soluble enzymes, lipid, small molecules.
1st pellet	54.6 ± 1.0	43.3 ± 0.5	82.9 ± 5.4
2nd pellet	12.1 ± 2.7	8.7 ± 3.8	10.9 ± 4.1
3rd pellet	30.2 ± 1.3	44.7 ± 3.2	5.1 ± 2.0
Supernatant	3.0 ± 1.8	3.2 ± 3.3	1.1 ± 1.4

Distribution of ⁷³As, ¹⁰⁹Cd, and ⁵¹Cr in sedimentary fractions

The geochemical fractionation patterns of the metals in the sediments that were fed to the worms are shown in Table 4. Arsenic was primarily in organic fractions, with smaller amounts in other fractions among which oxides dominated. Cadmium was primarily in the exchangeable fraction, but carbonate, oxide and, to a lesser extent, AVS fractions also contained this metal. In the sediments radiolabeled directly with Cr, this metal was predominantly in the AVS and oxide phases after 2 d incubation, but, unlike the other metals, the distribution of Cr changed significantly over time so that the oxide and AVS fractions were largely replaced by pyrite after 30 d.

Table 4. Percent distribution of arsenic, cadmium and chromium in geochemical fractions of sediments collected from Chesapeake Bay (Baltimore Harbor and Elizabeth River, USA) and Mare Island in San Francisco Bay, USAa

	Fraction	Baltimore Harbor			Elizabeth River			Mare Island
		2 d Direct	30 d Direct	30 d Via algae	2 d Direct	30 d Direct	30 d Via algae	2 d Direct	30 d Direct	30 d Via algae
a EX = exchangeable fraction; CARB = carbonate fraction; AVS = acid volatile sulfide fraction; OX = iron and manganese oxide fraction; ORG = organic fractions combining humic and fulvic acids; PYR = pyrite fraction. Radioisotopes were added either directly from aqueous solution or via previously radiolabeled algal biomass. Sediments were aged for 2 or 30 d when labeled directly or for 30 d when sediments were radiolabeled via algae. Data indicate means of three replicate sediment batches.
^73As	EX	0	2	4	0	2	21	1	3	3
	CARB	1	1	5	2	1	9	3	3	9
	AVS	9	13	6	29	10	11	25	37	18
	OX	4	4	20	13	14	21	11	6	19
	ORG	55	46	50	36	41	20	40	34	38
	PYR	14	7	0	7	7	0	8	4	0
^109Cd	EX	45	32	39	59	90	62	24	37	21
	CARB	14	17	9	8	4	10	24	21	17
	AVS	12	21	7	11	1	5	33	21	5
	OX	29	31	33	22	5	13	17	21	34
	ORG	1	0	7	1	0	6	1	0	13
	PYR	1	0	3	0	0	3	0	0	8
^51Cr	EX	1	1	1	1	1	2	1	1	1
	CARB	3	2	1	4	2	3	3	2	2
	AVS	54	14	10	40	7	11	52	8	14
	OX	37	13	66	35	10	55	29	15	41
	ORG	6	10	13	19	22	18	14	30	23
	PYR	0	57	8	0	58	10	0	43	16

Relationship of AEs with metal fractionation

The relationships of all metal AEs with their geochemical fractionation patterns in all sediments are given in Table 5, which presents regression slopes between AEs of all metals and the percentage of metal in individual or combined geochemical fractions of all sediments. These relationships are also presented for all metals as a function of sediment with or without added algal debris, by each individual metal, or as a function of sediment age (Table 5). Statistical analyses in Table 5, performed on arcsine transformed data to normalize the distribution of the data (AEs and percentages in each geochemical fraction), using PASW 18 statistical software, indicate that the association of As with the exchangeable pool alone or combined with carbonate had a significant positive slope (Table 5). Relationships between metal AEs and metal association with the exchangeable pool or when combined with the carbonate pool as a whole are positive for sediments from Norfolk and Mare Island but not for Baltimore Harbor, whereas metal AEs show a negative relationship with their association with the Fe/Mn oxide fraction for all sites (Table 5). For both 2- and 30-d aged sediments, metal AEs in the exchangeable + carbonate fractions had a positive relationship, and metals in oxides (Fe/Mn oxides + AVS) had a negative one (Table 5).

Table 5. Results of regression between assimilation efficiencies (AEs) and % metal in geochemical fraction of sediments collected from different locationsa

Fraction	By location						By metal
	BH		ER		MI		As		Cd		Cr
	n = 60		n = 51		n = 67		n = 58		n = 64–73		n = 62–71
	t	p	t	p	t	p	t	p	t	p	t	p
1	−0.71	0.483	6.16	0.000	6.76	0.000	2.64	0.011	0.85	0.399	−0.46	0.647
2	−1.87	0.066	3.37	0.001	6.48	0.000	6.11	0.000	−0.04	0.969	−0.28	0.781
3	−1.93	0.058	−3.86	0.000	−0.32	0.747	0.00	0.997	2.93	0.005	2.92	0.005
4	−2.49	0.016	−4.77	0.000	−2.47	0.016	0.18	0.857	−3.81	0.000	0.14	0.887
5	3.61	0.001	−1.66	0.103	−0.83	0.409	1.16	0.250	−2.44	0.018	−1.85	0.069
6	2.20	0.032	−3.83	0.000	−7.31	0.000	0.99	0.325	−2.72	0.009	−1.70	0.095
7	0.69	0.493	−2.69	0.010	−3.01	0.004	1.62	0.112	−2.84	0.006	−1.87	0.066
8	2.61	0.012	0.59	0.560	0.06	0.952	0.65	0.519	ND	ND	−1.02	0.311
9	−0.98	0.332	5.95	0.000	6.91	0.000	5.61	0.000	1.09	0.281	−0.32	0.748
10	−2.36	0.022	4.97	0.000	5.78	0.000	3.75	0.000	2.52	0.014	1.51	0.136
11	−2.80	0.007	−5.39	0.000	−1.55	0.127	−1.60	0.115	−0.60	0.553	2.98	0.004
12	3.44	0.001	−3.05	0.004	−2.74	0.008	1.34	0.186	−2.68	0.009	−3.06	0.003
13	1.87	0.067	−1.70	0.096	−2.92	0.005	1.32	0.192	−2.61	0.011	−1.89	0.064

Fraction	By age						By labeling
	Pure algae		2d		30d		Direct label		Mixed with algae
	n = 24		n = 80		n = 88		n = 124		n = 54
	t	p	t	p	t	p	t	p	t	p
a Regression was applied to arcsine transformed data; data set was divided by location, metal, age, and method of radiolabeling; characters in italics indicate regressions that are statistically significant (p < 0.05), t indicates a result of a t statistic for n samples (Predictive Analytics SoftWare PASW 18); [Single pools: 1 = exchangeable, 2 = carbonate, 3 = AVS, 4 = Fe/Mn oxides, 5 = humic acids, 6 = fulvic acids, 7 = pyrite, 8 = residual; pooled fractions; 9 =  exchangeable + carbonate, 10 = exchangeable + carbonate + AVS, 11 = AVS + Fe/Mn oxides, 12 = humic + fulvic acids, 13 = pyrite + residual]; ND = not determined; AVS = acid-volatile sulfides. See Table 1 for explanation of locations.
1	−1.57	0.131	1.65	0.103	6.31	0.000	3.53	0.001	2.72	0.008
2	−3.16	0.005	2.92	0.005	6.03	0.000	4.99	0.000	1.62	0.110
3	5.01	0.000	−3.05	0.003	0.10	0.924	−2.71	0.008	3.45	0.001
4	ND	ND	−4.22	0.000	−3.03	0.003	−3.08	0.003	−6.68	0.000
5	ND	ND	1.73	0.088	−1.77	0.080	−0.09	0.929	1.50	0.140
6	ND	ND	−1.58	0.118	−5.93	0.000	−2.50	0.014	−6.24	0.000
7	ND	ND	1.86	0.067	−4.04	0.000	−2.29	0.024	−6.99	0.000
8	ND	ND	1.72	0.090	−0.80	0.425	0.87	0.385	1.01	0.318
9	−2.06	0.051	2.15	0.035	7.21	0.000	4.21	0.000	4.06	0.000
10	−0.02	0.987	0.45	0.652	8.54	0.000	3.11	0.002	2.88	0.005
11	ND	ND	−3.86	0.000	−2.70	0.008	−3.23	0.002	−7.00	0.000
12	ND	ND	1.11	0.272	−3.74	0.000	−0.70	0.484	−1.15	0.254
13	ND	ND	1.77	0.081	−4.47	0.000	−1.44	0.154	−3.95	0.000

The significant (one-way ANOVA: p < 0.05) positive relationship of metal AEs with their fractionation in the exchangeable + carbonate sediment fractions (exchangeable and carbonates) for all metals and sample sites with unamended and amended sediments (both 2 and 30 d) is shown in Figure 2. The significant (one-way ANOVA: p < 0.05) inverse relationship of metal AEs with their fractionation in oxides and AVS fractions in sediment labeled via algal detritus and aged for a month is shown in Figure 3. Figure 4 summarizes the general pattern of slopes from regressions relating metal AEs and sediment fractionation for all metals, sampling sites, sediment labeling methods, and sediment age. This figure shows the slopes of significant (linear regression: p < 0.05) positive relationships for exchangeable (fraction 1) and carbonate (fraction 2) fractions alone or combined (fraction 9) and the slopes of significant (linear regression: p < 0.05) negative relationships for Fe/Mn oxides (fraction 4) alone or when combined with AVS (11). Pyrite (fraction 7) and fulvic acid (fraction 6) phases also show a negative slope.

Figure 2. Significant regression (p < 0.05) between assimilation efficiency (AE) and concentration of metal (As, circle; Cd, triangle; and Cr, square) in exchangeable + carbonate fractions in sediments collected from Baltimore Harbor, USA (gray), Elizabeth River, USA (black), and Mare Island, USA (white) and labeled via mixing with radiolabeled algal detritus (symbols with thicker borders). Regressions were performed on all individual replicates; error bars represent one standard deviation for AEs (n = 5–8 individuals).

Figure 3. Significant regression (p < 0.05) between assimilation efficiency (AE) (y axis) and concentration of metal (x axis; As, Cd, and Cr) in acid-volatile sulfides (AVS) + Fe/Mn oxides fractions in sediments collected from Baltimore Harbor, USA (gray), Elizabeth River, USA (black), and Mare Island, USA (white) and labeled via mixing with radiolabeled algal detritus. Regressions were performed on all individual replicates; vertical error bars represent 1 standard deviation for AEs (n = 5–8 individuals), and the horizontal error bars represent 1 standard deviation for the percentage of metal in combined AVS + Fe/Mn oxides fractions (n = 3).

Figure 4. Regression slopes (y axis) for assimilation efficiencies (AEs) and % metal in geochemical fractions (x axis); regressions were applied to arcsine transformed data. Only slopes of statistically significant (p < 0.05) regressions are shown. Single pools: 1 = exchangeable; 2 = carbonate; 3 = acid-volatile sulfides (AVS); 4 = Fe/Mn oxides; 5 = humic acids; 6 = fulvic acids; 7 = pyrite; 8 = residue. Pooled fractions: 9 = exchangeable + carbonate; 10 = exchangeable, carbonate + AVS; 11 = AVS + Fe/Mn oxides; 12 = humic + fulvic acids; 13 = pyrite + residual.

DISCUSSION

The positive relationship of metal assimilation efficiencies with exchangeable + carbonate sediment fractions for As(V), Cd, and Cr(III) is consistent with reports that suggested that metals bound to these two sediment fractions can be more bioavailable for benthic invertebrates than other fractions 27. Ligands in the gut fluid of deposit-feeders provide a site for metal ion exchange 17; hence, the choice of combining the two metal ion exchanging pools (e.g., exchangeable + carbonate fractions) was made. Generally, oxic sediments have more metal associated with these phases, and therefore metals in oxic sediments would be expected to be more bioavailable than metals in anoxic sediment; exceptions have been noted, however, such as for Cd, Cr, and Zn in the suspension-feeding mussel Mytilus edulis9. Metals bound to operationally identified phases of iron and manganese oxides, organic phases, and pyrite in sediments showed an inverse relation with assimilation efficiencies in N. succinea (Fig. 4), indicating that metals associated with these phases have low bioavailability for N. succinea. This polychaete is a surface deposit-feeder and hence feeds primarily on oxic sediment.

In comparing the behavior of metals in sediments from different sample sites, assimilation of As showed a negative relationship with total concentrations of Fe (r² = 0.97) and Mn (r² = 0.99) in sediment but a positive relationship with Al (r²= 0.99) in sediments from all three study sites (Fig. 5). It is evident that in sediments arsenic shows a strong association with iron- and manganese-rich phases such as Fe/Mn oxides, AVS, and pyrite 28. Lower As AEs occurred for sediments with higher Fe and Mn concentrations (Fig. 5), further supporting the observations that Fe/Mn and AVS (Fig. 4) phases control the bioavailability of ingested As. In sediments, Al can be found in the mineral structure of aluminosilicates. Aluminum's positive correlation to As AEs may be explained by a weak metal binding with the surfaces of aluminosilicates, which can more easily release metal ions into the gut fluid than Fe and Mn oxides 29. The particle reactivity of Cd is inversely related to chloride concentration 30, and as salinity increases Cd's retention by ingested particles would be expected to decrease, releasing more Cd into the gut, where it could be eventually assimilated, consistent with observations shown in Figure 5. The organic carbon content in sediments from our study sites showed an inverse relation to Cd AEs (Fig. 5). Degraded organic matter can bind metals and thereby limit their bioavailability. The negative relationship of Cd AEs with its association with extracted organic fractions was also significant (Table 5), further supporting this relationship. AE values for Cr were very low and variable, and relationships with sediment characteristics are more tentative.

Figure 5. Relationships between As assimilation efficiencies (AEs) (y axis) and the total concentrations of aluminum (triangles), manganese (squares), iron (circles) in sediments (x axis), and Cd AEs (y axis) and percent C in sediments (black circles) and salinity (x axis) of overlying water at sediment collection sites (open circles). The AEs of As and Cd are for sediments labeled directly and aged for 2 d; AE errors are ignored for clarity of the figure and are given in Table 2. In situ metal (Al, Mn, and Fe) concentrations are from Baumann et al. (unpublished data). No other relationships were found for metal AEs and total elemental concentration in sediment.

Generally, these findings suggest that metals must be released from ingested particles into the gut fluid before they can be transported across the gut lining and become assimilated into tissues. As such, these findings support the contention of Mayer and colleagues that metal and organic contaminants must be released into gut fluid before they can be assimilated. Most of these earlier studies focused on release from particulates to the dissolved phase in gut fluid and did not assess their AEs, although some studies determined both the solubilization and absorption of some organic contaminants 14, 31.

Assimilation efficiencies of As in N. succinea were highest when diet consisted of fresh algal debris or sediments amended with algal debris, suggesting that As, which is bound to some labile sugars in algal cells 32, is the bioavailable form, and arsenic bound to sedimentary organic matter such as humic or fulvic acids is not assimilable in worms. Furthermore, arsenobetaine, a common form of As in marine invertebrates 33, is far more assimilable (42%) by crustacean predators, Crangon crangon, than inorganic As species—arsenate (1.2%; 34)—consistent with our observations that inorganic As bound to goethite was much less assimilable (2.5%) by N. succinea than organic As in algal debris (72%). Because sugars that contain As in cells are likely labile and degrade over time, the decreased assimilation of As over 30 d in sediment mixed with algal debris was expected as these sugars decompose over time. Few other studies determined AEs of As in marine invertebrates. Assimilation efficiencies of As in another nereid worm Nereis diversicolor reached 62% 35, comparable to our results but much higher than AEs reported for the deposit-feeding polychaete Arenicola marina (4–11%; 36).

Studies by Griscom et al. 9 and Baumann et al. (unpublished data) illustrate the shift of metals from exchangeable + carbonate phases to more refractory phases over time, which can possibly result from microbial degradation of labile organic matter or an increase in Fe/Mn oxide–metal associations after a month of sediment incubation (Z. Baumann et al., unpublished data). Labile organic compounds represent the most biologically available source of C, in contrast to more refractory sources, with 55 to 95% of C assimilated by N. succinea feeding on fresh algae compared with only 5 to 18% of reworked organic matter in sediment 18. This difference between labile and refractory organic matter coincides with decreased AEs of As in N. succinea (the present study); Cd and Ag in the clam Macoma balthica9; Cd, Cr, and Zn in the clam Ruditapes philippinarum; and Cd in the mussel Perna viridis37, although Wang and Fisher 13 found that AEs of physiologically regulated Co, Se, and Zn were unaffected by metal–sediment exposure time. Other studies that examined bioavailability and toxicity of chemicals for worms in soils similarly showed a decrease in bioavailability with increased exposure time 38.

With further sediment aging, ferric iron associated with iron oxides would be expected to be ultimately reduced and dissolve as ferrous iron. Ferrous iron on reaction with dissolved sulfide species would precipitate as iron sulfides, contributing to the AVS phase. In anoxic sediments, AVS can ultimately be transformed to pyrite. Acid-volatile sulfides and pyrite, when present in sediments, can bind metals and organic matter that are dissolved in pore water 11. Metals bound to these phases are more tightly bound to particles and can only be released by strong acids (e.g., HCl and HNO₃), which are harsher than the digestive fluids of marine invertebrates.

Unlike As, the AEs of Cd and Cr did not consistently decline with sediment aging (Table 2), suggesting that bioavailability of these metals may not be tied directly to their association with labile organic C in sediments. This is further supported by the observation that AEs of Cd bound to pure goethite were comparable to those bound to pure algal debris. Presumably, Cd is released equally well into gut fluid from goethite and algal debris. Cadmium showed a wide range of AEs (from <1% to nearly 69%), whereas others reported a narrower AE range for N. succinea, from 29 to 39% 22. Bivalves appeared to assimilate similar amounts of Cd in comparison with worms. Filter-feeding mussels assimilated from 9.5 to 44.5% 13, 37, and clams assimilated 31 to 51% of Cd 15, 37. The estuarine crustacean Palaemonetes pugio39 assimilated 57% of Cd from an oligochaete diet. The AE of Cd in mussels from pure mineral phases was lower than that from algal cells 12.

As seen in many previous studies with diverse invertebrates, Cr displayed lower AEs than the other metals 13. Commonly, its AE from ingested food is less than 10%, particularly for deposit-feeding animals 40. In San Francisco Bay, the clam Potamocorbula amurensis, feeding on particles rich in labile organic compounds collected after a spring bloom, assimilated more than 5% of Cr, compared with AEs of 1.7% during a nonbloom period 40. It is noteworthy that mussels feeding on algae assimilate up to 10% of Cr(VI) but less than 2% of Cr(III), the difference being explainable by Cr(VI)'s greater ability to penetrate into the cytoplasm of the algal cells 41.

Analogous to their fractionation in sediments, metal fractionation in algal cells can significantly influence their assimilation in animals. Previous work has shown that AEs of ingested metals in marine herbivores are strongly correlated with the cytological distribution of the metals in algal cells, with AEs showing a nearly 1:1 relationship with cytoplasmic distribution of the metals in the algal cells that constituted the diet 25. The low AE of ingested Cr from pure algal debris (2.8%) coincides with its predominant association with diatom cell walls and membranes, in contrast with Cd and As, consistent with earlier findings. The AEs of algal As (69–76%) and Cd (up to 43%) are comparable (p < 0.05) to their extraction in the exchangeable fraction (Cd: 43%) alone or in the exchangeable + carbonate pool (As: 65%) within the algae. Nevertheless, although 18% of algal Cr was extracted in the exchangeable pool (Z. Baumann et al., unpublished data), its assimilation in N. succinea was low (2.8%). Reasons for this discrepancy are not known, but trivalent metals cannot readily penetrate the gut lining. Chromium also displayed low AEs (<5%) from sediments (Table 2), and its association with exchangeable fractions in these sediments was correspondingly low (Z. Baumann et al., unpublished data).

Thus, the bioavailability of As, Cd, and Cr to the surface deposit-feeder N. succinea is positively related to their exchangeable + carbonate fraction in sediment and negatively related to their fractions in the AVS, Fe/Mn oxides and pyrite and nonextractable phases. We therefore suggest combining the exchangeable and carbonate pools into a carbonex pool, recognizing that geochemical and physiological processes can positively impact sedimentary metal assimilation in deposit-feeding polychaetes through an ion-exchange process (ion exchange in both MgCl₂ and NaOAc extractions; presence of ligands in the gut fluid that can serve as ion exchange sites).

The present study did not accurately evaluate ingestion rates in N. succinea in these experiments. Ingestion rates can vary in deposit-feeding polychaetes, depending on the particle being ingested, such as natural sediment vs goethite. Gut retention times can influence the AEs of ingested metals 13, and although this may possibly explain differences noted in AEs between metals bound to goethite and algal-supplemented sediment, both of these radiolabeled food sources were purged with the identical unlabeled sediment. The gut passage times of the various radioactive foods were unlikely to account for their differing significantly, and accounted for AE differences.

Worms assimilate more As when fed pure algae and less when algae are mixed with sediment. AEs of As from directly labeled sediment were lower than As AEs from ingested sediments mixed with algae, and were positively related to total Mn and Al concentrations in sediments but negatively related to sedimentary Fe. Assimilation efficiencies of ingested sedimentary Cd increased with salinity and decreased with sedimentary organic carbon. The present study confirms that Cr has generally low bioavailability for deposit-feeding polychaetes. Further appreciation of metal assimilation in deposit-feeders will result from physiological and biochemical studies that also consider the sediment geochemistry.

Acknowledgements

We thank G. Cutter and J. Pan for assistance in the field, G. Cutter for providing the protocol for the sequential extraction procedure, G. Lopez and L. Mayer for constructive comments, and three anonymous reviewers for helpful comments. This research was supported by SERDP W912HQ06C0014 and NSF CHE0221934.