View - Vrije Universiteit Amsterdam

bio.vu.nl

View - Vrije Universiteit Amsterdam

Environ. Sci. Technol. 2003, 37, 3399-3404

Elucidating the Routes of Exposure

for Organic Chemicals in the

Earthworm, Eisenia andrei

(Oligochaeta)

TJALLING JAGER,* ,†

ROEL H. L. J. FLEUREN, ‡

ELBERT A. HOGENDOORN, § AND

GERT DE KORTE §

Department of Theoretical Biology, Vrije Universiteit

Amsterdam, de Boelelaan 1085, NL-1081 HV Amsterdam,

The Netherlands, and Laboratory for Ecotoxicology and

Laboratory for Organic-Analytical Chemistry, National

Institute of Public Health and the Environment (RIVM),

P.O. Box 1, NL-3720 BA Bilthoven, The Netherlands

Earthworms take up organic compounds through their

skin as well as from their food, but the quantitative contribution

of each route is unclear. In this contribution, we

experimentally validate an accumulation model containing

a separate compartment for the gut. Uptake from the

gut is modeled as passive diffusion from the dissolved

phase in the gut contents. For the experiments, we exposed

Eisenia andrei in artificial soil spiked with tetrachlorobenzene,

hexachlorobenzene, and PCB 153. Apart from the standard

accumulation and elimination experiments, we ligatured the

worm (using tissue adhesive) to prevent feeding. Model

fits were good, thus supporting the validity of the model. The

contribution of the gut route increased with increasing

hydrophobicity of the chemical, and for PCB 153 the gut

route clearly dominated. Despite the importance of the gut

route, the final steady-state body residues did not

exceed equilibrium partitioning predictions by more than

25%. Rate constants for exchange across the skin and the

gut wall could be separately identified. The rate constant

across the skin decreases with K ow but was generally

higher than data derived from water-only exposure. The

relationship with hydrophobicity was less clear for the rate

constant across the gut wall.

Introduction

Earthworms are able to take up organic chemicals through

their skin (1) as well as from their food (2). However, the

quantitative contribution of each route remains unclear. As

earthworms regularly consume soil, it is difficult to study

both routes in isolation in a relevant experimental setup.

Earthworms can be exposed in water alone (3), but the

relevance of this setup for porewater uptake from soil is not

obvious. More work has been done in this area for sediment

organisms, showing that ingestion is an important pathway

for very hydrophobic chemicals such as pyrene and dioxins

(4, 5). For earthworms, Belfroid et al. (6) predicted, on the

* Corresponding author: e-mail: tjalling@bio.vu.nl; telephone:

+31 20 444 7134; fax: +31 20 444 7123.


Vrije Universiteit Amsterdam.


Laboratory for Ecotoxicology, RIVM.

§

Laboratory for Organic-Analytical Chemistry, RIVM.

basis of model extrapolations, that food uptake becomes an

important exposure route for very hydrophobic chemicals

(log K ow > 5). It seems to be a generally held opinion that

feeding on soil can lead to the invalidation of equilibrium

partitioning (EP), which is why additional safety factors were

prompted in European risk assessment guidelines (7).

In most case, uptake from food is modeled by simply

adding uptake routes (8), but in a previous contribution (9),

we proposed a more mechanistic accumulation model. Based

on the work of Gobas et al. (10, 11), the model includes a

separate compartment for the gut contents and a closed mass

balance. The mechanism for uptake from the gut is likely to

be the same as for uptake across the skin (i.e., passive

diffusion). This assumption is strongly supported by experimental

evidence as derived for intact goldfish (10), isolated

gut segments of catfish (12), and humans (13). For earthworms,

the validity of this assumption was indicated,

although a proper validation was impossible because the

physiological data regarding the feeding process were lacking

(9). Most routine studies with earthworms are carried out

with the compost worm (Eisenia andrei/fetida) in an artificial

soil medium (14). It is for this system that the essential feeding

parameters have recently been identified (15) including gut

loading, gut retention time, and digestion efficiency.

In this study, we set out to validate the accumulation

model with three organic compounds [tetrachlorobenzene

(TeCB), hexachlorobenzene (HeCB), and PCB 153] in artificial

soil. A series of experiments was performed with these

chemicals, starting with a straightforward accumulation and

elimination phase. Subsequently, soil from these experiments

was reused to see whether the bioavailable phase had

been altered, as indications of depletion have been observed

(1). Finally, an accumulation experiment was performed

with worms sealed with a tissue adhesive, thus

preventing feeding. This procedure was pioneered by Vijver

et al. (16) to demonstrate uptake routes for heavy metals.

Other forms of ligaturing have been applied to demonstrate

that pesticides are mainly taken up through the skin in wateronly

exposure (1) and that calcium is mainly taken up from

the diet (17). However, this study is to our knowledge the

first to separate exposure routes for organic chemicals in a

soil situation.

The data from all these experiments are used together to

fit the accumulation model and to identify the rate constants

for uptake through the skin and from the gut. Furthermore,

the model can shed light on the central questions: which

exposure route dominates and does feeding lead to body

residues exceeding the predictions made by equilibrium

partitioning.

Experimental Section

Exposure Media and Spiking Procedure. Artificial soil was

used for the experiments (14). The water content was brought

to 40% (weight basis, water/dry medium) with a lutetium

solution (Lu, hydrated chloride salt, purity 99.9%, Alfa Aesar,

Karlsruhe, Germany) to obtain a nominal concentration of

15 mg/kg dwt. The Lu is used as a nonassimilated tracer to

compare the feeding activity to earlier experiments (15). After

the soil was wetted, it was thoroughly mixed and stored in

closed plastic containers at 5 °C for 1 week prior to spiking

with organic chemicals. After storage of the containers, the

pH (KCl) of the soil was 5.0. 1,2,3,4-Tetrachlorobenzene and

hexachlorobenzene were obtained from Riedel de Haën,

Seelze, Germany (99% purity, Pestanal); PCB 153 was

synthesized at the IRAS, Utrecht, The Netherlands (99%

purity).

10.1021/es0340578 CCC: $25.00 © 2003 American Chemical Society VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3399

Published on Web 07/02/2003


The spiking procedure for these compounds was adapted

from Northcott and Jones (18). Because we needed to spike

4kg dwt of medium, we had to do the procedure in steps

(dilution spike). First, the chemicals needed to achieve a

nominal concentration of 10 mg/kg dwt for each chemical were

dissolved in 100 mL of acetone (pro analysis). Wet soil (1 kg)

was placed in a kitchen blender, and the acetone solution

was slowly added while mixing. Mixing continued for several

minutes (stopping a few times to crush aggregates with a

spatula). The spiked soil was left in a fume cabinet overnight

after which the acetone and also most of the water had

evaporated. Next, one-fifth of the spiked soil was put in the

blender with 900 g wwt of uncontaminated medium and the

water needed to restore the 40% water content. This was

mixed for several minutes, stopping a few times to prevent

the medium from overheating and crushing the aggregates.

This entire procedure was followed five times until the entire

medium was spiked. The moisture content was checked by

oven drying at 80 °C and was 41%. The fraction organic matter

(F om) in the soil was 10.5% (loss on ignition). To allow

equilibration, the medium was stored at 10 °C in glass jars

for 1 week before animals were introduced. One day before

animals were introduced, soils were transferred to the test

temperature of 20 °C.

Test Animals and Experimental Setup. Sub-adult earthworms

(E. andrei), were taken from mass cultures at RIVM

(Bilthoven, The Netherlands). The animals weighed between

200 and 300 mg wwt. First, the animals were allowed to evacuate

their gut contents by keeping them on moist filter paper for

24hat20°C. Next, the animals were transferred to plastic

containers with 175 g wwt of uncontaminated medium (four

worms per container), and the containers were placed at 20

°C, covered by a black plastic pot to minimize disturbance.

The animals were left to acclimatize for 1 week under these

conditions. After this, they were exposed to the chemicals in

1-L glass jars using 250 g wwt of spiked medium and four

animals per jar. Several jars were used for the determination

of the feeding activity (see Supporting Information).

For the accumulation experiment, the worms were

recaptured following exposure (0, 1, 2, 3, 5, 7, 14, and 21 d)

and placed in a Petri dish on moist filter paper for 24 h at

20 °C. After this period, worms were packed in aluminum

foil and frozen at -20 °C. Soil samples were taken and frozen

in glass jars at -20 °C (four samples at t ) 0 and two

samples at days 14 and 21). At t ) 14, three additional jars

were emptied, and the worms were recaptured. The worms

were transferred to 250 g wwt of uncontaminated medium

and allowed to eliminate for 2, 5, and 11 d. The bioavailability

of the chemicals may change during the experiment, which

is why the spiked soil from the four jars emptied at t ) 14

was reused. Fresh worms were taken from the culture, placed

on wet filter paper for 24 h and subsequently on uncontaminated

medium for another 24 h, and then introduced

in the reused soil. These worms were recaptured after 1, 3,

7, and 11 d.

For the ligaturing experiment, worms were taken from

the culture and allowed to empty their gut for 24 h on moist

filter paper. Their anterior end was ligatured using a tissue

adhesive (Indermil, from Loctite Ireland Ltd.), conforming

to the procedure by Vijver et al. (16). Gluing turned out to

be difficult for this species because the worms were irritated

by the procedure and excreted coelomic fluid, which

interfered with the setting of the glue. As exposure time, 0,

1, 2, 3, 5, 7, and 14 d were employed using five worms per

jar with 250 g wwt of soil. After exposure, the worms were

recaptured and placed individually in a Petri dish with moist

filter paper overnight. Only the worms that did not evacuate

any solids were used for chemical analysis. For t ) 0, five

worms were taken; on days 1 and 5, three worms had been

successfully exposed; on day 2, only one worm was exposed

(after longer periods, all worms excreted solid materials and

were discarded).

Analysis of Organochlorine Compounds. For the sample

pretreatment of soil, 10 g wwt was mechanically shaken during

10 min in a glass tube with 25 mL of acetone. Next, 50 mL

of light petroleum (a mixture of saturated alkanes with a

boiling point of 40-60 °C) were added, and the contents

were mechanically shaken for 20 min. After centrifugation

(5 min. at 3000 rpm), the liquid phase was transferred into

a shaking funnel. The remaining part was extracted again

following the same procedure, and the liquid was transferred

to the funnel. After the addition of 500 mL of water, the funnel

was manually shaken for 1 min. The aqueous phase was

discharged, and the upper layer was extracted once more for

1 min with 500 mL of water. The light petroleum phase was

passed through a funnel with about 10 g of anhydrous sodium

sulfate and concentrated to a volume of 10 mL.

For analysis of the worms, a sample of approximately 1

g was placed into a glass tube and weighed. After addition

of 100 µL of the internal standard (approximately 100 ng of

[ 13 C 12]-PCB 153), 9 mL of isopropyl alcohol and 10 mL of

cyclohexane were added, and the mixture was macerated

with an ultra-speed homogenizer for 2 min. Next, 10 mL of

water was added, and the mixture was macerated again for

1 min. The phases were separated by centrifugation for 10

min at 3000 rpm. The upper organic layer was transferred

through a funnel with sodium sulfate to a Kuderna-Danish

evaporation apparatus by means of a pasteur pipet. The

remaining part of the sample was macerated again for 1 min

with 10 mL of a mixture of 2-propanol-cyclohexane (13:87,

v/v). After centrifugation, the upper layer was added to the

first extract, and sodium sulfate was rinsed with 5 mL of

cyclohexane; the organic layer was concentrated to 1 mL.

For cleanup (fat destruction), the extract was brought onto

a chromatography column filled with 0.5 g of silica gel

impregnated with sulfuric acid (100 g of silica heated for 4

hat200°C and 43.5 mL of concentrated sulfuric acid mixed

by rotating for 12 h). Next, 5 mL of hexane was passed through

the column, and the organic solvent was collected in a

calibrated glass tube and brought to a volume of 5 mL.

For instrumental analysis with GC/MS operating in the

electron impact (EI) mode, 1 mL of extract was transferred

into an autosampler vial, and 10 µL of pentachlorobenzene

(PeCB, 1 µg) was added as internal standard. A total of 1.5

µL was on-column injected into a fused silica DB-5MS

capillary column coated with 5% cross-linked 5% phenyl

methyl siloxane with a length of 30 m × 0.25 mm i.d. and a

film thickness of 0.25 µm. Helium was applied as carrier gas

at a flow of 1 mL/min. Quantification was done using the

internal standard PeCB for calibration and, in case of worm

samples, the isotope PCB 153 to correct for losses during

sample pretreatment. The average recoveries performed at

levels between 2 and 10 µg/g ranged between 83 and 103%

with SD below 9% (n ) 4 for each analyte-matrix combination).

The Model. The model is set up as a three-compartment

model with a closed mass balance (Figure 1). Diffusion (the

two-way arrows) and advection (one-way arrows) are the

basic transport processes. The model has been described

earlier (9), and the full model formulation is available in the

Supporting Information. Each compartment is assumed to

be well-mixed and of constant volume. Although the gut is

probably better reflected by a plug-flow reactor (19), the

simpler mixed compartment serves as an approximation (11).

The chemicals will be taken up into the tissue of the worm

from the outside soil as well as from the gut contents; both

processes are modeled as passive diffusion from the dissolved

phase.

The diffusion gradient between soil or gut contents and

worm tissue is defined by the concentrations in both

3400 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003


FIGURE 1. Schematic representation of the accumulation model.

compartments and the partition coefficient between organic

matter (OM) and earthworm tissue (K ws in kg OM/kg wwt). The

K ws can be viewed as the ratio of the bioconcentration factor

with water (BCF in L/kg wwt) and the organic matter specific

sorption coefficient (K om in L/kg OM). When multiplied by the

soil concentration on OM basis, K ws would reflect the

equilibrium body residue of a nonfeeding worm. The OMworm

partition coefficient is the same for soil to worm as

from gut contents to worm. However, the magnitude of the

gradient differs between these two uptake routes because

the chemical concentration as well as the F om in the gut

contents differs from that in soil (due to selective feeding,

compaction, and OM digestion) (10, 11). The net result of

the two uptake routes depends on the kinetics of the various

transport processes, and the steady-state body residue in a

feeding earthworm will thus end up somewhere between

equilibrium with the soil and equilibrium with the gut

contents.

In this study, we chose to ignore degradation in the gut

and biotransformation as removal processes. This is acceptable,

given the very short gut retention time (Table 1) and

lack of indications for biotransformation of chlorobenzenes

and PCBs in earthworms (20, 21). However, the analysis

results prompted us to include a first-order loss term for the

soil (k d). Supported by pilot calculations (9), instantaneous

chemical equilibrium between solid and water phases is

assumed. Several adaptations to the previous model (9) are

made. First, compaction of the gut contents is included,

meaning that the gut volume decreases as food is absorbed

(15). Also, a slightly different formulation for the F om in the

gut is taken as average of F om determined in ingesta and

egesta instead of egesta only.

Model Fitting. The model equations are implemented in

matrix form in MatLab Version 6.1 (The Mathworks, Inc.)

and solved with the matrix exponential function. For each

chemical, we have five data sets (soil concentrations, two

accumulation phases, one elimination phase, and accumulation

in glued worms) that must be described by the

same model and with the same parameter values. Therefore,

all data sets must be fitted simultaneously. This is accomplished

by defining a likelihood function on the basis of

the sums of squares (SSQ) from the model fits on each data

set assuming normally distributed data (22):

5

L(θ|data) ∝ ∏ SSQ(θ; data i ) -n i/2

i)1

where θ is the entire set of parameters, and n i is the number

of points in data set i. Different likelihood functions may be

multiplied, so in this way we end up with one expression for

the overall likelihood of the model parameters given all five

data sets. For the gluing experiment, the number of worms

that were successfully glued was taken as a weight coefficient

(1)

in the SSQ. The overall log-likelihood function is maximized

by a Nelder-Mead simplex search in MatLab, yielding

maximum-likelihood (ML) estimates of the parameters. The

likelihood function is also used to construct confidence

intervals by calculating the profile likelihood (23), which is

more realistic for small data sets than standard asymptotic

procedures based on large-sample theory.

Even though the gut retention time is only 2.9 h in feeding

worms, 24 h depuration on filter paper is insufficient to

remove all of the gut contents. However, a longer depuration

could lead to bias as also chemicals will be lost from the

tissues. We corrected the modeled concentration for the

remaining gut contents using an estimate of the remaining

fraction (F rem, Table 1, see Supporting Information). The ML

estimates are used to estimate the net chemical assimilation

efficiency (AE), the biota-soil accumulation factor (BSAF),

and the deviation from EP. The AE can be calculated from

the uptake flux from the gut contents into the worm tissues

and the chemical flux with feeding (see Supporting Information).

BSAF (in kg OM/kg wwt) is calculated from the modeled

concentration in the worm (C w in mg/kg wwt) att ) 21, the

concentration in the soil (C s in mg/kg dwt), and the F om in soil:

BSAF ) C w (21)F om (soil)

C s (21)

The magnitude of the BSAF cannot be directly interpreted

in relation to EP as we did not measure porewater concentrations

and lipid content of the worms. However, the

deviation from EP can be assessed in an indirect manner by

comparing C w(21) to equilibrium estimates based on the

concentration in soil and the modeled concentration in the

gut (C g):

C w (EP, soil) ) K ws C s (21)

F om (soil)

C w (EP, gut) ) K ws C g (21)

F om (gut)

In these equations, we use the parameter estimate for K ws

(resulting from the model fit) for the EP estimates. The worm

can come to equilibrium with the soil (eq 3) or the gut

contents (eq 4) or end up somewhere between (in a

nonequilibrium steady state).

To obtain a confidence interval on these derived results,

we applied a random parameter search. Parameters were

randomly drawn from log-uniform distributions. When the

likelihood of these parameters (eq 1) was not significantly

lower than the ML estimate, the parameter combination was

stored. Random parameters were drawn until 200 acceptable

parameter combinations were obtained. For each parameter

combination in this set, the BSAF, AE, and deviation from

EP were calculated. The maximum and minimum values serve

as confidence intervals.

Results and Discussion

General Observations. We want to apply the parameter

values for the feeding activity, as observed previously (Table

1), to our current experiments. For this reason, we first

confirmed that these values are indeed representative for

this study (see Supporting Information). The worms increased

approximately 10% in weight during the 21-d exposure; the

potential effects on accumulation kinetics are expected to

be small and were ignored.

The initial concentration in the soil was 5.8 mg/kg dwt for

TeCB, 7.2 for HeCB, and 6.8 for PCB 153. These were lower

than the nominal 10 mg/kg dwt, presumably due to losses when

evaporating the acetone. However, the homogeneity of the

(2)

(3)

(4)

VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3401


TABLE 1. Fixed Parameters for the Feeding Process (15)

symbol description unit value SE

F ege egested feces as fraction of body weight kg dwt/kg wwt 0.12 0.012

F rem fraction of gut contents remaining after 24 h depuration kg dwt/kg dwt 0.056 0.021

F sel selectivity for OM in diet (-) 2.1 0.053

F dig digestion efficiency of OM (-) 0.35 0.033

F com factor by which gut contents are compacted (-) 1.09 0.011

T ret gut retention time h 2.9 2.5-3.3 a

F solids fraction solids in worm kg dwt/kg wwt 0.14 0.0013

a

90% probability interval.

TABLE 2. Parameter Estimates and Derived Measures, Resulting from the Model Fits a

parameter unit TeCB HeCB PCB 153

Chemical Properties and QSAR Estimations

log K ow 4.64 (25) 5.73 (25) 6.92 (26)

K ws estimated from QSARs (20, 27) kg OM/kg wwt 0.12 0.20 0.33

Estimated Model Parameters

K ws (first accumulation exp) kg OM/kg wwt 0.11 [0.077, 0.13] 0.20 [0.19, 0.22] 0.24 [0.23, 0.27]

K ws (second accumulation exp) kg OM/kg wwt [-] b 0.16 [0.15, 0.18] 0.16 [0.13, 0.21]

rate constant skin (k s) d -1 0.74 [0.42, 2.1] 0.30 [0.22, 0.43] 0.027 [0.020, 0.037]

rate constant gut wall (k g) d -1 0.27 [0, +∞] 0.43 [0.18, 0.72] 0.16 [0.13, 0.20]

degradation and/or volatilization (k d) d -1 0.0065 [0.0051, 0.0077] 0.0055 [0.0051, 0.0058] 0.0062 [0.0058, 0.0066]

Derived Results

BSAF at t ) 21 kg OM/kg wwt 0.12 [0.072, 0.13] 0.23 [0.21, 0.26] 0.30 [0.25, 0.36]

kg OC/kg

c lip 7.1 14 18

assimilation efficiency (maximum) % 10 [2.6 × 10 -4 , 50] 26 [0.19, 42] 16 [12, 22]

deviation from EP with soil % 7.3 [0,18] 13 [1.0, 23] 24 [18, 27]

deviation from EP with gut contents % -14 [-21, 0] -7.1 [-21, -1.9] -3.3 [-6.4, -0.3]

a

Maximum-likelihood estimates with 95% likelihood-based confidence intervals. K ws is the OM-worm partition coefficient. b Assumed the

same value as in the first accumulation phase. c Calculated, assuming a lipid content of 1% (20) and a factor of 1.7 between organic carbon and

organic matter.

FIGURE 2. Model fits for the different accumulation and elimination experiments (top) and the modeled uptake fluxes from soil and gut

contents, with 95% confidence intervals (bottom). The EP prediction marks the estimated body residue for a worm in equilibrium with

the soil.

spiking was acceptable as the standard deviations were

4-5% (n ) 4) of the average value. The soil concentrations

appeared to decrease some 10% in the course of the exposure

experiment. It is striking that the rate constants for this

disappearance (k d, Table 2) are practically identical for all

three chemicals. The reasons for this disappearance have

not been investigated but could include the formation of

3402 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

resistant fractions (24), degradation, or volatilization.

The accumulation model can adequately describe the data

from the different experiments simultaneously (Figure 2) with

only five free parameters (see Table 2), an average of one

parameter per data set. Only the data for TeCB are rather

scattered (especially in the first accumulation phase). In most

cases, the parameters are accurately identified by the data,

evidenced from the tight 95% likelihood-based confidence

intervals (Table 2). However, it should be noted that these


confidence intervals do not include the uncertainty in the

feeding parameters (Table 1). Only for TeCB, the poor fit

results in ill-defined estimates. The good fit for the other

compounds is consistent with the assumption that gut uptake

is mediated through passive diffusion from the dissolved

phase in the gut (10, 11).

Ligaturing the earthworm allows to isolate exchange across

the skin in a soil situation. Even though the worms survive

for several days without problems, the treatment is stressful.

The stress is indicated by the coelomic fluid, expelled by all

the worms in reaction to the glue. Expelling fluid is a normal

response of E. andrei to rough handling and has no lasting

effects on their health. Other species (e.g., the genus

Lumbricus) do not have this response, but E. andrei was the

species for which the feeding parameters were available

(Table 1). The ligatured worms may show deviating behavior

in soil, which can also bias chemical uptake. Nevertheless,

the difference between glued and intact worms is remarkably

small for both chlorobenzenes, providing some reassurance

that the earthworms are sufficiently active to take up the

chemicals through the skin.

Exposure Routes and Assimilation Efficiency. The estimated

uptake fluxes clearly show an increase of the

importance of the gut in the total uptake with increasing

hydrophobicity (Figure 2). Furthermore, the net fluxes

decrease in time as the animal is approaching steady state.

Figure 2 shows that for TeCB the skin is probably the most

important route, although the large confidence intervals

preclude firm conclusions. For HeCB, both routes are

approximately equally important, but for PCB 153, the gut

route is truly the dominant exposure route. Although we do

not agree with the approach taken by Belfroid et al. (6), our

conclusion is comparable: the gut begins to become an

important route for chemicals with a log K ow above approximately

5 and dominates above 6. This is also reflected

in the deviation from EP with soil, which increases with K ow

from 7 to 24%. However, the confidence intervals are quite

large, and only the deviation for PCB is clearly higher than

0%. Similarly, the deviation from EP with the gut contents

decreases with increasing K ow so that for PCB 153 the tissue

residues are nearly in equilibrium with the gut contents.

The parameter estimates also allow calculation of the net

chemical AE from the estimated fluxes in the model. As shown

previously (9), the net AE depends on time; therefore, only

the maximum is given here. No general trend with K ow is

observed, although a trend may be obscured by the large

confidence intervals.

Partition Coefficients. The OM-worm partition coefficient

(K ws) was accurately predicted by the QSARs for wormwater

(20) and organic carbon-water partitioning (27),

assuming a factor of 1.7 between organic carbon and organic

matter (Table 2). Note that K ws in this case is a model

parameter; the final body residue in steady state depends on

the kinetics and lies between the extremes: equilibrium with

the soil or the gut contents (eqs 3 and 4). The actual BSAF

is thus a secondary result (eq 2) and is slightly higher than

the K ws, indicating that the concentrations in the worms

exceed equilibrium with the soil. The BSAFs after lipid and

carbon normalization are larger than unity, reflecting that

sorption increases less with K ow (27) than does bioconcentration

(20).

The K ws in the reused soil was generally somewhat lower

than the value in the initial accumulation phase. Only for

TeCB, a higher K ws was estimated in the reused soil. As we

judged this behavior to be an artifact (because of the scatter

in the initial accumulation phase), we chose to use only one

K ws for both accumulation experiments with this chemical.

For PCB 153, we saw a clear decrease in K ws for the reused

soil, showing that bioavailability had declined over 2 weeks.

This difference could not be modeled as a depletion of the

FIGURE 3. Rate constants for exchange across the skin and across

the gut wall vs log K ow. Error bars represent 95% likelihood-based

confidence intervals. The broken line indicates elimination rates

for earthworms in water only (3).

bioavailable phase, as mentioned in the Introduction.

Unfortunately, we cannot offer a convincing explanation for

this phenomenon, but some form of sequestration (either

autonomous or caused by the worms or their feces) remains

possible.

In the model, we assume the same partition coefficient

(K ws) for exchange from soil to worm (via porewater) as from

gut contents to worm (via the gut fluid). However, gut fluids

differ from porewater in that they include secretions from

the worm to aid digestion. Mayer et al. (28) have shown that,

in marine deposit feeders, these secretions include surfactants

that also act to solubilize organic contaminants above levels

expected in seawater. It may appear that the action of gut

fluids invalidates the hypothesis of passive diffusion via a

water phase. However, we do not believe this to be the case

as there is strong support for a diffusion-driven uptake (see

Experimental Section), even though surfactants facilitate this

transport (13). Furthermore, although secretions with surfaceactive

properties will increase the dissolved chemical concentration,

they cannot increase the fugacity gradient.

Because the gut fluid becomes less polar than water due to

these secretions, the chemical has less urge to flee to the

earthworm’s tissues. The gut fluid-worm partition coefficient

is decreased by the same factor as the solubility is increased,

leading to the same net uptake. The same conclusion was

reached by Lu et al. (29). We therefore believe that gut

secretions act mainly on the gut rate constant (k g) and not

on the OM-worm partition coefficient (K ws).

Rate Constants. The rate constants for exchange across

the skin (k s) and the gut wall (k g) could be separately identified

(Table 2). Only for TeCB, a confidence interval for the gut

rate constant cannot be made. There is a value that has the

highest likelihood, but very high and very low values are not

significantly worse (apparently exchange across the skin is

so rapid that the gut route cannot compete). The rate

constants for chemical exchange are shown in Figure 3, which

also gives a fit on data for the elimination from worms in a

water-only situation (3). Comparison to literature data for

elimination rates in soil is not useful as reported constants

will always reflect the total elimination flux through all routes.

The skin rate constant (k s) for TeCB is quite comparable to

the water-only data, but our value for HeCB is much higher.

This is striking given the fact that, in a water-only situation,

the contact between worm and water is likely to be more

intensive. Furthermore, rate constants for passive diffusive

exchange are expected to decrease with K ow with a slope

close to 1 (on log scale) in this K ow range because for these

compounds the diffusion across a stagnant water layer is

rate-limiting (30). However, for the rate constant across the

VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3403


skin, we find a slope that is less steep than for water-only

exposure (-0.63 vs -1.3). An explanation may be derived

from the earthworm’s physiology. In a soil situation, earthworms

lose 10-20% of their body weight in moisture each

day due to their respiratory system, which requires the

maintenance of a moist outer surface (31). In water-only

exposure, the water loss will likely be much less. The water

losses need to be replenished, requiring water transport

across the skin (also advectively transporting the chemical).

This process may explain a higher exchange rate in soil than

in water and the different relationship with K ow.

The relationship between the rate constant across the gut

wall (k g) and K ow cannot be properly assessed because of the

large confidence interval for TeCB (Figure 3). However, given

the tight confidence intervals for the other chemicals and

judging from the maximum likelihood estimate of TeCB, it

appears that k g is actually quite constant over the studied

K ow range (Figure 3). Again, what we expected was a decrease

with a slope around unity. It is possible that surfactants in

the gut facilitate the transport across the stagnant water layer,

thus influencing the relationship with K ow. This was also

proposed for the absorption of lipids in the gut, a process

following passive diffusion but with bile salts enhancing the

transport (32).

Consequences for EP in Risk Assessment. To our

knowledge, this is the first time that, in earthworms, the

uptake of organic chemicals from soil through the skin has

been separated from uptake resulting from feeding on soil

particles. The importance of the gut route increases with

increasing hydrophobicity, and very hydrophobic chemicals

(log K ow > 6) will mainly be absorbed from the gut contents.

There is some additional uptake as a result of feeding on soil,

but the deviation from EP with the soil is less than a factor

of 1.3, which is well within the accuracy of risk assessment

applications. The general fear that feeding leads to the

invalidation of EP is thus unwarranted.

The model presented here can adequately describe the

experimental data, and the results are consistent with the

diffusion mechanism for gut uptake (10, 11). However, in

view of the small deviations, risk assessment can rely on EP,

and specific modeling of the gut compartment is usually not

necessary. Nevertheless, this model may be useful for specific

cases, especially when the worms are not feeding on soil

alone but on a diet that is specifically contaminated (e.g.,

manure from farm animals treated with pharmaceuticals or

pesticide residues in leaf litter). In these situations, soil

concentrations along with EP are insufficient to predict body

residues.

Acknowledgments

We thank the Laboratory for Inorganic Chemistry at RIVM

for performing the lutetium measurements in soil and worms

and the Institute for Risk Assessment Sciences (IRAS, Utrecht)

for kindly providing the organic chemicals. Furthermore, we

thank Martina Vijver for demonstrating the ligaturing procedure,

Rob Baerselman for support in the experiments, and

Willie Peijnenburg and Joop Hermens for reviewing drafts of

this manuscript.

Supporting Information Available

Model equations and data on checking the validity of the

feeding parameters of Table 1. This material is available free

of charge via the Internet at http://pubs.acs.org.

Literature Cited

(1) Lord, K. A.; Briggs, G. G.; Neale, M. C.; Manlove, R. Pestic. Sci.

1980, 11, 401-408.

(2) Belfroid, A.; Meiling, J.; Sijm, D.; Hermens, J.; Seinen, W.; Van

Gestel, K. Arch. Environ. Contam. Toxicol. 1994, 27, 260-265.

(3) Belfroid, A.; Van Wezel, A.; Sikkenk, M.; Van Gestel, K.; Seinen,

W.; Hermens, J. Ecotoxicol. Environ. Saf. 1993, 25, 154-165.

(4) Leppänen, M. T.; Kukkonen, J. V. K. Environ. Sci. Technol. 1998,

32, 1503-1508.

(5) Loonen, H.; Muir, D. C. G.; Parsons, J. R.; Govers, H. A. J. Environ.

Toxicol. Chem. 1997, 16, 1518-1525.

(6) Belfroid, A. C.; Seinen, W.; Van Gestel, K. C. A. M.; Hermens, J.

L. M.; Van Leeuwen, K. J. Environ. Sci. Pollut. Res. 1995, 2,5-15.

(7) EC. Technical Guidance Documents in Support of Directive 93/

67/EEC on Risk Assessment of New Notified Substances and

Regulation (EC) No. 1488/94 on Risk Assessment of Existing

Substances (Parts I, II, III and IV); Office for Official Publications

of the European Community: Luxembourg, 1996.

(8) Landrum, P. F.; Lee, H.; Lydy, M. J. Environ. Toxicol. Chem.

1992, 11, 1709-1725.

(9) Jager, T. Ecotoxicol. Environ. Saf. (in press).

(10) Gobas, F. A. P. C.; McCorquodale, J. R.; Haffner, G. D. Environ.

Toxicol. Chem. 1993, 12, 567-576.

(11) Gobas, F. A. P. C.; Zhang, X.; Wells, R. Environ. Sci. Technol.

1993, 27, 2855-2863.

(12) Weber, L. P.; Lanno, R. P. Environ. Toxicol. Chem. 2001, 20,

1117-1124.

(13) Moser, G. A.; McLachlan, M. S. Environ. Sci. Technol. 2002, 36,

3318-3325.

(14) OECD. Guideline for Testing of Chemicals No. 207. Earthworm,

Acute Toxicity Tests; Organization for Economic Cooperation

and Development: Paris, France, 1984.

(15) Jager, T.; Fleuren, R. H. L. J.; Roelofs, W.; De Groot, A. C. Soil

Biol. Biochem. 2003, 35, 313-322.

(16) Vijver, M. G.; Vink, J. P. M.; Miermans, C. J. H.; Van Gestel, C.

A. M. Soil Biol. Biochem. 2003, 35, 125-132.

(17) Piearce, T. G. J. Anim. Ecol. 1972, 41, 167-188.

(18) Northcott, G. L.; Jones, K. C. Environ. Toxicol. Chem. 2000, 19,

2409-2417.

(19) Penry, D. L.; Jumars, P. A. Am. Nat. 1987, 129, 69-96.

(20) Jager, T. Environ. Toxicol. Chem. 1998, 17, 2080-2090.

(21) Belfroid, A.; Van den Berg, M.; Seinen, W.; Hermens, J.; Van

Gestel, K. Environ. Toxicol. Chem. 1995, 14, 605-612.

(22) Box, G. E. P.; Tiao, G. C. Bayesian Inference in Statistical Analysis;

Wiley-Interscience: New York, 1992.

(23) Meeker, W. Q.; Escobar, L. A. Am. Stat. 1995, 49, 48-53.

(24) Pignatello, J. J.; Xing, B. Environ. Sci. Technol. 1996, 30, 1-11.

(25) De Bruijn, J.; Busser, F.; Seinen, W.; Hermens, J. Environ. Toxicol.

Chem. 1989, 8, 499-512.

(26) Hawker, D. W.; Connell, D. W. Environ. Sci. Technol. 1988, 22,

382-387.

(27) Sabljić, A.; Güsten, H.; Verhaar, H.; Hermens, J. Chemosphere

1995, 31, 4489-4514.

(28) Mayer, L. M.; Chen, Z.; Findlay, R. H.; Fang, J.; Sampson, S.; Self,

R. F. L.; Jumars, P. A.; Quetel, C.; Donard, O. F. X. Environ. Sci.

Technol. 1996, 30, 2641-2645.

(29) Lu, X.; Reible, D. D.; Fleeger, J. W.; Chai, Y. Environ. Toxicol.

Chem. 2003, 22, 153-160.

(30) Gobas, F. A. P. C.; Opperhuizen, A.; Hutzinger, O. Environ.

Toxicol. Chem. 1986, 5, 637-646.

(31) Lee, K. E. Earthworms. Their ecology and Relationships with

Soils and Land Use; Academic Press: Sydney, Australia, 1985.

(32) Friedman, H. I.; Nylund, B. Am. J. Clin. Nutr. 1980, 33, 1108-

1139.

Received for review January 20, 2003. Revised manuscript

received May 7, 2003. Accepted May 12, 2003.

ES0340578

3404 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

More magazines by this user
Similar magazines