1690

Environmental Toxicology and Chemistry, Vol. 20, No. 8, pp. 1690–1697, 2001q 2001 SETAC

Printed in the USA0730-7268/01 $9.00 1 .00

EFFECTS AND RISK ASSESSMENT OF LINEAR ALKYLBENZENE SULFONATES INAGRICULTURAL SOIL. 5. PROBABILISTIC RISK ASSESSMENT OF LINEAR

ALKYLBENZENE SULFONATES IN SLUDGE-AMENDED SOILS

JOHN JENSEN,*† HANS LøKKE,† MARTIN HOLMSTRUP,† PAUL HENNING KROGH,† and LARS ELSGAARD‡†National Environmental Research Institute, Department of Terrestrial Ecology, P.O. Box 314, Vejlsøvej 25DK-8600 Silkeborg, Denmark

‡Danish Institute of Agricultural Sciences, Department of Crop Physiology and Soil Science, Research Centre Foulum, P.O. Box 50,DK-8830 Tjele, Denmark

(Received 25 May 2000; Accepted 28 September 2000)

Abstract—Linear alkylbenzene sulfonates (LAS) can be found in high concentrations in sewage sludge and, hence, may enter thesoil compartment as a result of sludge application. Here, LAS may pose a risk for soil-dwelling organisms. In the present probabilisticrisk assessment, statistical extrapolation has been used to assess the risk of LAS to soil ecosystems. By use of a log-normaldistribution model, the predicted no-effect concentration (PNEC) was estimated for soil fauna, plants, and a combination of these.Due to the heterogeneous endpoints for microorganisms, including functional as well as structural parameters, the use of sensitivitydistributions is not considered to be applicable to this group of organisms, and a direct, expert evaluation of toxicity data was usedinstead. The soil concentration after sludge application was predicted for a number of scenarios and used as the predicted envi-ronmental concentration (PEC) in the risk characterization and calculation of risk quotients (RQ 5 PEC/PNEC). A LAS concentrationof 4.6 mg/kg was used as the current best estimate of PNEC in all RQ calculations. Three levels of LAS contamination (530, 2,600,and 16,100 mg/kg), three half-lives (10, 25, and 40 d), and five different sludge loads (2, 4, 6, 8, and 10 t/ha) were included inthe risk scenarios. In Denmark, the initial risk ratio would reach 1.5 in a realistic worst-case consideration. For countries not havingsimilar sludge regulations, the estimated risk ratio may initially be considerably higher. However, even in the most extreme scenarios,the level of LAS is expected to be well beyond the estimated PNEC one year after application. The present risk assessment,therefore, concludes that LAS does not pose a significant risk to fauna, plants, and essential functions of agricultural soils as aresult of normal sewage sludge amendment. However, risks have been identified in worst-case scenarios.

Keywords—Ecotoxicology Soil fauna Plants Surfactants Detergent

INTRODUCTION

Linear alkylbenzene sulfonates (LAS) are one of the majoringredients of synthetic detergents and surfactants and are usedworldwide for both domestic and industrial applications. Theannual worldwide consumption of LAS is in the range of morethan one million tons per year. The LAS are rapidly degradedunder aerobic conditions but only very slowly, or not at all,degraded under anaerobic conditions. Therefore, LAS can befound in high concentrations in sewage sludge and, hence,may enter the soil compartment as a result of sludge appli-cation. A Danish ordinance from 1997 prescribes a maximumcontent of four groups of organic chemicals in waste productssuch as domestic sewage sludge or compost intended for useas fertilizers on agricultural land. The cut-off values includeda criterion for LAS of 2,600 mg/kg dry weight, which waslowered to 1,300 mg/kg in July 2000. The LAS have beenidentified as potentially hazardous substance and, therefore,are placed on the list of unwanted substances in Denmark. Thebasis for the current cut-off value of LAS in sludge is anecotoxicological soil-quality criterion of 5 mg/kg.

The fate and effects of LAS in terrestrial ecosystems wererecently reviewed by Jensen [1]. The basic structure of theLAS molecule is a benzene ring connected to an alkyl chainof different length and a sodium sulfate group. The LAS arecharacterized by having both a hydrophobic (alkyl chain) anda hydrophilic group (sodium sulfate group). The primary effect

* To whom correspondence may be addressed(john.jensen@dmu.dk).

of LAS is a general disruption of biomembranes and dena-turation of proteins. A previous paper by Kloepper-Sams etal. [2] dealt with the effect assessment of LAS in sludge-amended soils, and an initial worst-case risk assessment ofLAS in the terrestrial environment was performed by De Wolfand Feijtel [3]. However, new research has recently provideda substantial number of new effect data, which enable a morerealistic risk assessment of LAS in sewage sludge–amendedsoils [4–7]

MATERIALS AND METHODS

In ecological risk assessment, the risk characterization is,in essence, a simple comparison of predicted environmentalconcentration (PEC) and predicted no-effect concentration(PNEC), normally expressed as the risk quotient (RQ). Thispaper includes a preliminary assessment of PEC but generallyfocuses on the effect-assessment part of the risk characteriza-tion.

Predicted no-effect concentration

In the European Union, the procedures regarding how toassess the risks of new and existing substances are describedin a technical guidance document [8]. In this document, anassessment factor ranging from 1,000 to 1 is applied to lab-oratory or field-effects data to establish a PNEC. The choiceof assessment factor depends on the quantity and quality oftoxicity data. By the factorial method, the PNEC is calculatedbased on the lowest effect value measured, excluding the re-maining data in the assessment. Recently, alternative methods

Risk assessment of LAS Environ. Toxicol. Chem. 20, 2001 1691

based on statistical extrapolation have become available, andthese extrapolation methods apply sensitivity distributions forthe derivation of PNEC and, hence, use the full data set [9,10].According to the technical guidance document in support ofthe European Commission Directive 93/67/EEC, these meth-ods are not yet accepted in the risk-assessment procedures usedfor notification of new substances in the European Union.However, authorities deriving environmental-quality objec-tives or setting environmental standards use the methods. Forexample, in Denmark, The Netherlands, and Canada, sensitiv-ity frequency distribution methods have been useful tools inpredicting environmental effects based on laboratory data [11].

In the present study, the method of Wagner and Løkke [9]is applied. By extrapolation, this method determines a lowerstatistical tolerance limit, so that one can assert with a certainprobability that only a certain percentage of all species withina community are influenced by the substance in question. Theinput to the model is ecotoxicological effect data (no-observed-effect concentration [NOEC] or EC10/EC5) from laboratorystudies. These data were found in the existing literature andfrom recent studies in our laboratory [6,7]. For the presentassessment of PNEC, EC10 rather than NOEC values are pre-ferred for calculations. Many scientists [12–14] have suggestedthe use of a bound effect level (e.g., EC10) instead of theNOEC as a proper approach. Some of the advantages of theECx estimations as compared to NOEC values are that the ECxvalues are interpolated and less sensitive to the choice of testconcentrations. Additionally, the precision of the ECx valuescan be quantified with confidence intervals, thus making themcomparable, and because of large variations in a data set andthe need for statistical differences, NOEC values may corre-spond to large effects, in contrast to ECx values.

It is mandatory for the chosen sensitivity distribution meth-od that the data input has a log-normal distribution. In general,larger sets of toxicity data fulfill the ideal criteria regardinglog-normal distribution. However, a large variation of end-points and taxa, in combination with a limited number of testspecies, may cause serious errors in the extrapolation proce-dure because of incorrect assumptions regarding the distri-bution model or a lack of representativeness of the selectedtaxa [9]. The logarithms of the data were tested for normalityby use of Kolmogorov statistics [15]. The Kolmogorov statisticassesses the discrepancy between the empirical distributionand the estimated, hypothesized distribution. The p value ofKolmogorov statistic D (labeled Prob . D) determines whetherto reject the null hypothesis that the data are a random samplefrom the normal (or log-normal) distribution. In the presentstudy, all data were accepted as having a log-normal distri-bution.

By use of the log-normal distribution, a concentration (Kp)is found, for which the EC10 or NOEC values for 95% of allspecies in a community are greater. The value of Kp is usedas the estimate of the PNEC. For agricultural soils, the log-normal distribution is applied by using a confidence level of50%. This distribution equals the standard ta distribution, let-ting the logarithmic value be the random variable. For com-parison, Kp 5 5 is also estimated with a confidence of 95%. The50 and 95% confidences represent the normal procedure forderiving soil-quality criteria in The Netherlands and Denmark,respectively. The estimation of PNEC has been performed foranimals and plants separately and for a combination of bothgroups. A PNEC for microbial toxicity was not estimated bythe distribution method. A similar extrapolation method has

previously been used by van Beelen and Doelman [16] inassessing risk of pollutants to soil microorganisms. However,the inherent nature of and differences between the differentgroups of toxicity tests generally make the approach of sen-sitivity distributions less suitable for soil microorganisms andmicrobial processes [17]. In this study, therefore, we decidedthat it was unfeasible at this stage to use the approach ofsensitivity distributions for the impact on specific microor-ganisms in combination with microbial soil functions. Instead,we preferred to make an expert evaluation of recently obtainedtoxicity data, representing several microbial assays from dif-ferent functional tests (i.e., microbial biomass measurement,carbon and nitrogen transformations, and anaerobic and aer-obic enzymatic tests). Nonetheless, it is recommended thatwork continue regarding how and when to use microbial tox-icity tests for probabilistic risk assessment, and that the validityof such extrapolations continue to be studied.

Predicted environmental concentration

The LAS can be found at elevated concentrations in soilimmediately after sludge amendment. The LAS concentrationin soils that have not received sewage sludge within one ortwo months is generally less than 1 mg/kg, and very seldommore than 5 mg/kg [1]. By assuming a uniform distributionin the plough layer, LAS concentrations can be estimated foragricultural soils after receiving sewage sludge. The concen-tration level of LAS in sludge may vary considerably. TheLAS concentrations in sludge from 19 Danish wastewatertreatment plants ranged, for example, between 11 and 16,100mg/kg dry weight, with 10th, 50th, and 90th percentiles of13,530, and 13,200 mg/kg [18]. With a sludge load of 10 t/ha, the soil concentration may reach more than 50 mg/kg. InDenmark, a realistic maximum load of sewage sludge or com-post on agricultural fields is 6 t/ha every three years, or anequivalent amount containing phosphorus at 30 kg/ha/year asan average over three years. The cut-off value for LAS insludge at 2,600 mg/kg dry weight would, in a realistic worst-case situation (i.e., sludge load of 6 t/ha) lead to a LAS loadof 15.6 kg/ha, or approximately 7 mg/kg dry weight in soil.This is under the assumption of a uniform distribution of LAS-containing sludge in the top 15 cm of a soil with a density of1.5 kg/L. A normal or typical concentration can be estimatedto 1 mg/kg dry weight using the median LAS concentrationof 530 mg/kg and an average sludge load of 4 t/ha.

These values are in accordance with other findings, but theydo not necessarily represent worst-case scenarios in othercountries or specific situations with extremely high loads ofsludge to the soil system. Prats et al. [19] reported from Spainan initial LAS concentration in soil immediately after amend-ment of sludge at 22.4 mg/kg dry weight. Both anaerobic,aerobic, and composted products in total amounts of as muchas 16 t/ha dry weight were used twice a year. Marcomini etal. [20] observed a disappearance of LAS from an initial soilconcentration of 45 to 5 mg/kg in 320 d. For more than 10years, the soil received a total of 142 t/ha (average, 13.5 t/year dry wt) of anaerobically digested sludge. Holt et al. [21]and Waters et al. [22] reviewed the results from a monitoringprogram to ascertain the level and fate of LAS in agriculturalsoils in the United Kingdom. Following sludge applications,five soils from three locations had initial LAS concentrationsfrom 2.6 to 66.4 mg/kg. De Wolf and Feijtel [3] calculateddifferent LAS scenarios for agricultural soils and grassland

1692 Environ. Toxicol. Chem. 20, 2001 J. Jensen et al.

Table 1. Effect concentrations for soil invertebrates showing no-observed-effect concentration (NOEC), EC10 and EC50 values (mg/kg) and theEC10:EC50 ratioa

Species Endpoint NOEC EC10 EC50EC50:EC10

ratio Reference

Eisenia foetidaLumbricus terrestrisAporrectodea caliginosaAporrectodea longaFolsomia fimetariaFolsomia candidaHypoaspis aculeiferEnchytraeus albidusPlatynothrus peltiferIsotoma viridisHypogastrura assimilis

ReproductionWeightReproductionReproductionReproductionReproductionReproductionReproductionReproductionGrowthReproduction

667

320

383

1427961881.76.2

4199.8

558

129137442

91236

40.5467

421

1.5—9.35.14.05.12.96.51.5—4.2

[45][46]

[6][6]

[6,7][45]

[6][6]

[45][45]

[6]

a For the springtail Folsomia fimetaria, the mean of 7 individual experiments from Holmstrup et al. [6,7] is used for the extrapolation. The dataused for derivation of predicted no-effect concentration are italicized.

Table 2. Effect concentrations for terrestrial plants (mg/kg)a

Species Endpoint EC10 EC50Extrapolated

NOECb Reference

Brassica rapaBrassica rapaMalvia pusilaSolanum nigrumChenopodium albumAmaranthus retroflexusNigella arvensisGalinsoga parvifloraSorghum bicolorHelianthus annuusPhaseolus aureusAvena sativaSinapis alba

GrowthGrowthGrowthGrowthGrowthGrowthGrowthGrowthGrowthGrowthGrowthGrowthGrowth

90

50200

134, 7200204, 2169, 2164, 3142, 5132, 9

90, 1137289316300300

20.416.916.414.313.39.0

13.728.931.6

[47][48][47][47][47][47][47][47][49]c

[49]c

[49]c

[48][48]

a Data used for derivation of predicted no-effect concentration are italicized. EC50 values are extrapolated to no-effect concentrations by a divisionof 10.

b NOEC 5 no-observed-effect concentration.c As cited in Mieure et al. [50].

using varying PEC values in the range of 0.44 to 10.5 mg/kgdry weight, depending on the degradation rate.

Effect data for soil fauna

The EC10 values were available for 11 invertebrate species(Table 1). For the springtail Folsomia fimetaria, the meanEC10 of seven independent experiments was used for the ex-trapolation [7,23]. The mean and standard deviation of EC10and EC50 values from the seven tests were 96 6 39 and 4426 154 mg/kg, respectively (Table 1). For the earthworm Lum-bricus terrestris and the oribatid mite Platynothrus peltifer,only NOEC values were available. The ratio between 50 and10% effects, indicating the concentration–effects relationshipfor each test species, is shown in Table 1.

Effect data for plants

For plants, a data set of 12 different species could be es-tablished, but only the EC5 and EC50 values were available(Table 2). For three plant species, the EC50:EC5 ratios couldbe calculated, showing values between 1.5 and 6. For deri-vation of the PNEC for plants, the three EC5 values werecombined with ‘‘no-effect concentrations’’ extrapolated fromEC50 values, applying an assessment factor of 10. The use ofan assessment of uncertainty factor of 10 to extrapolate EC50

values to NOEC values is a conservative approach used onlybecause of the low number of EC5 data for plants.

Effect data for microorganisms

For microorganisms, a new data set has been established,including various endpoints of functional as well as structuralparameters [4,5]. The PNEC estimation for microorganisms isbased on the data shown in Table 3. These data represent theshort-term effects of LAS added directly to a sandy agriculturalsoil (i.e., a worst-case situation compared to LAS applied toagricultural soil via sewage sludge) [5]. One of the lowestobserved EC10 values was for the iron reduction. The iron-reduction test should be interpreted separately, however, be-cause the assay conditions differed significantly from those ofthe other tests. Thus, LAS exposure in the iron-reduction testwas via a soil–water slurry, which may have increased thebioavailability to an unrealistic high level and, hence, alsoincreased the toxicity of LAS. The EC10 of the potential am-monium oxidation, the ethylene oxidation, and the growth ofcellulolytic microorganisms, which all are of primeval agro-nomic importance, were also less than or close to the lowesttest concentration of 8 mg/kg.

Risk assessment of LAS Environ. Toxicol. Chem. 20, 2001 1693

Table 3. Effect concentrations (EC10, EC50) of linear alkylbenzene sulfonates for microbial parameters in agricultural soil (mg/kg)

Microbial parameter Exp (d) EC10 EC50EC10:EC50

ratio

Iron reductionAmmonium oxidationDehydrogenase activityb-Glucosidase activityEthylene degradationCellulolytic bacteriaCellulolytic fungiCellulolytic actinomycetes

57770.5777

5a

5a

2247

9115a

8

1740

128.488

24243280

4.3105.8

.10.32.72.26.4

10

a EC10 values below the lowest test concentration are estimated by a simple arithmetic interpolation.

Table 4. Derivation of predicted no-effect concentrations (PNECs)(Kp) for plants and animals separately, and for combined data sets

including all species (mg/kg)a

K5,95% K5,50%

LowestEC10

AnimalsPlantsAnimals and plantsMicroorganisms

1.02.22.1—

4.65.34.6—

696

,10

a The PNECs are derived by the distribution extrapolation of Kp55 witha confidence level of 50 and 95%. Furthermore, the lowest experi-mental EC10 value is presented.

Modifying factors for toxicity of LAS

Most studies of LAS toxicity have been based on experi-mental set-ups using a homogeneous mixing of a water-solubleform of LAS (usually Na-LAS) into a natural or artificial soil.The bioavailability of LAS under such circumstances is, there-fore, expected to be at its highest. Speciation of LAS, the soiltype, and the exposure of LAS in a sludge matrix may beparameters that would modify the toxicity and, thus, shouldbe taken into account during a risk assessment. The degree ofmodification by these factors was studied by Holmstrup et al.[7] and Elsgaard et al. [5] using springtails, earthworms, andmicrobial tests.

For springtails and earthworms, the effects of LAS werecompared in three different soil types, ranging from a sandyto a clay soil. Little difference in toxicity to springtails andearthworms was observed. Similarly, effects of Na-, Ca-, andMg-salts of LAS were compared. These studies showed thatno difference in the toxicity of the three different forms ofLAS. The conclusion from these studies is that the level ofexchangeable Ca21 in typical Danish soils is sufficiently highto completely substitute Na1 to form insoluble Ca-LAS whenNa-LAS is spiked into the test soil. Holmstrup et al. [7] didnot find that sludge amendment had any modifying effects onthe toxicity of LAS to soil fauna in the range of concentrationsused. In fact, an increased toxicity was observed during a 14-d anaerobic incubation of sludge (resulting in no degradationof LAS) before it was mixed into the test soil. It was suggestedthat LAS and/or the incubation procedure caused an increasedbioavailability of other (unidentified) toxic compounds of thesludge. Elsgaard et al. [5], on the other hand, found that forfive different microbial soil parameters, the application of LASvia sewage sludge dampened the ecotoxicological effect ofLAS. This effect may be due to reduced availability of LAS,the addition of a large pool of easily accessible nutrient, orthe relatively large number of LAS-adapted microorganismsassociated with the sludge matrix.

In conclusion, speciation of LAS and soil type did notsignificantly modify the toxicity of LAS, whereas the presenceof sludge may, at least for microorganisms, influence LAStoxicity. Because no strong evidence reports against it, toxicitydata with LAS are used directly for the risk assessment.

RESULTS

By use of the log-normal distribution model, the Kp is cal-culated for both soil fauna and plants and a combination ofall these data. To estimate the range of sensitivity, the PNECis estimated with a 95% and a 50% confidence level (Table4). The PNEC with the 50% confidence level is used for riskcharacterization. Additionally, the lowest experimental EC10

value is presented for comparison. Due to the heterogeneousendpoints for microorganisms including functional as well asstructural parameters, the use of sensitivity distributions is notconsidered to be applicable to this group of organisms. Theeffect assessment for this group of organisms is, therefore,based on an expert evaluation of all relevant available infor-mation.

A comparison between EC10 and EC50 values revealedthat in the case of LAS, the no-effect and effect levels aregenerally found within one order of magnitude and are com-parable for all three groups of organisms. The studies presentedby Rundgren and van Gestel [24] confirm a small differencein the sensitivity of lethal and sublethal parameters for LAS.The general mechanism of LAS at environmentally realisticconcentrations is perturbation of membranes [25]. Above thecritical micelle concentration, dissolution of cell membranesmay occur. The mechanisms are the same for all organismsand are often termed narcotic. This is in contrast to morespecific-acting substances, such as insecticides or herbicides.

Risk characterization of LAS in soil

By using the lowest PNEC value derived with a 50% con-fidence level (i.e., 4.6 mg/kg) and different PEC estimations,the RQ (i.e., PEC/PNEC) can be calculated for various sce-narios (Table 5). The RQ values greater than one indicatepotential (i.e., acute) risk of LAS to soil ecosystems. Assuminga simple, first-order rate disappearance of LAS [26,27], thereduction of RQ can be determined as a function of time.Hence, the time necessary to bring the RQ to less than onecan be estimated. Table 5 shows the environmental risk (RQ)and number of days (t) necessary before the RQ is less thanone for a number of risk scenarios. In all RQ calculations, aLAS concentration of 4.6 mg/kg was used as the current bestestimate of the PNEC. Large uncertainties may, however, beinvolved in the estimation of a PNEC [28,29]. Three levels of

1694 Environ. Toxicol. Chem. 20, 2001 J. Jensen et al.

Table 5. The environmental risk (RQ) calculated for different scenarios of varying sludge application rates, linear alkylbenzene sulfonates (LAS)concentrations, and soil degradation half-livesa

LAS concn.(mg/kg3

sludge dry wt)Half-life

(d)

Sludge application rate (ha21)

2

RQinit t (d)

4

RQinit t (d)

6

RQinit t (d)

8

RQinit t (d)

10

RQinit t (d)

530530530

2,6002,6002,600

16,10016,10016,100

102540102540102540

0.100.100.100.500.500.503.13.13.1

——————164165

0.200.200.201.01.01.06.26.26.2

——————2666

105

0.310.310.311.51.51.59.39.39.3

———

615243281

129

0.410.410.412.02.02.0

121212

———1025403691

145

0.510.510.512.52.52.5

161616

———1333534099

158

a Environmental risk is calculated as the predicted environmental concentration (PEC) divided by the predicted no-effect concentration (PNEC).The PNEC used in all scenarios is 4.6 mg/kg (Table 4). The initial risk quotient (RQinit) is shown, and the time needed to reach RQ , 1 iscalculated assuming first-order soil degradation kinetics. A solid line divides the scenarios showing RQinit . 1 and RQinit , 1 and a bold lineseparates scenarios needing more than one month to reach RQ 5 1.

LAS contamination have been included in the scenarios. Themedian level of LAS in Danish sludge (530 mg/kg), the max-imum allowable level of LAS (2,600 mg/kg), and the highestlevel of LAS found in Danish sludge from 19 wastewatertreatment plants (16,100 mg/kg) as reported by Tørsløv et al.[18]. Three half-lives have been included, ranging from 10 to40 d (rate constants between 0.069 and 0.017), which coversreported degradation rates in temperate regions [1]. Five dif-ferent sludge loads have been used for the calculation, rangingfrom a low level of 2 t/ha to the maximum allowable load inDenmark of 10 t/ha. All in all, these variations may representthree major groups of risk scenarios: a normal scenario cov-ering a median level of LAS in sludge amended at typicaldoses of 2 to 4 t/ha; a realistic worst-case scenario, with sludgecontaining the cut-off value of LAS in sludge and amendedat a rate of 6 t/ha; and a true worst-case scenario, with themaximum registered level of LAS found in Danish sludgeamended up to the theoretically allowable sludge rate (10 t/ha).

In the worst-case scenarios presented in Table 5, an RQ ofgreater than one is estimated in several cases, with a maximumRQ of 16. Sludge with the level of LAS used in this scenariois no longer permitted for use on agricultural soils in Denmark.However, in countries not having restrictions on LAS in sludgeused for soil amendment, this scenario may not be unrealistic.The theoretical time necessary to reach a level of LAS witha negligible risk (RQ , 1) in a Danish realistic worst-casesituation is less than 25 d, and only in scenarios no longerpermissible in Denmark is more than 50 d necessary to reachan RQ of less than one. This, however, depends on the deg-radation rate. Again, the degradation rate depends on the initialconcentration of LAS, the temperature, and other chemical,physical, and biological soil parameters [30]. Half-lives span-ning from 1 to more than 30 d have been observed for LASin soil [1]. In most cases, sludge will be spread during spring,before the growing season. The temperature in this period inDenmark is not likely to drop to a level at which degradationis arrested or hampered. The anticipated half-life of 10 to 40d is, hence, likely to cover most situations.

Exceedance plot

The risk characterization presented above is based on aprobabilistic approach for the toxicity data and a number of

fixed scenarios for the soil concentrations. However, a prob-abilistic approach can also be included in the exposure part ofa risk characterization [31]. In the same manner as the toxicitydata (i.e., PNEC) are assumed to have a log-normal distri-bution, the environmental concentration (i.e., PEC, the soilconcentration after sludge amendment) most likely is distrib-uted according to a log-normal model. The degree of overlapbetween the exposure distribution and the toxicity distributionmay be used to estimate a joint probability of exposure andeffects. This leads to estimates of exceedance probabilities forresponses at a fixed effects assessment criterion (e.g., the con-centration equivalent to the fifth percentile of the species dis-tribution [Kp]). In other words, it summarizes the probabilityof a random sludge application causing a risk to the mostsensitive species within an ecosystem. Instead of estimatingthe likelihood that a specific effect criterion will be exceeded,exceedance probabilities can be presented as a continuum oflikelihoods. This is done by combining the frequency distri-butions for the probability of a concentration being exceededwith the probability of a species being affected (or not affected)in a joint probability curve. In this way, the likelihood thatspecies will exceed their EC10 or NOEC as a result of sludgeapplication can be expressed graphically. In turn, this makesit possible to judge the outcome over a range of possible com-binations (e.g., with what frequency sludge poses a risk formore than 5, 10, or 50% of species). The advantage of thisapproach is that it is less rigid and conservative in its calcu-lations, because both the distribution of species sensitivity andsludge content are included. Furthermore, it gives a betterimpression regarding the actual consequences of sludgeamendment, because the classical risk characterization (i.e.,RQ) primarily indicates risk and not the potential magnitudeof effects.

An exceedance plot of sewage sludge associated LAS ispresented in Figure 1. The data for plants and invertebratesare combined; however, their separate exceedance plot doesnot differ significantly from the combined plot. Ideally, theplot should be based on a wide number of monitored soilconcentrations. These data are, however, scarce. Therefore, thesoil concentration has been estimated by applying a log-normaldistribution of LAS in sludge, a fixed sludge load of 6 t/hadry weight, and a uniform distribution within the first 15 cmof soil. The data for estimating the log-normal distribution of

Risk assessment of LAS Environ. Toxicol. Chem. 20, 2001 1695

Fig. 1. Exceedance plots for soil fauna and plants. The magnitude ofeffect (% of species) is plotted against the exceedance frequency(probability or relative frequency that species exceed their no-effectlevel).

LAS in sludge are found in Tørsløv et al. [18]. In most situ-ations, only a small fraction of plant and invertebrate speciesare likely to be exposed to LAS concentrations exceeding theirno-effect level (i.e., EC10 or NOEC). When applying sludgeat a rate of 6 t/ha, less than 0.6% of all species have a likelihoodof more than 50% to exceed their EC10. Organisms amongthe 5% most sensitive species are at risk of exceeding theirEC10 in approximately one-third of the cases, whereas ap-proximately 10% of sludge applications lead to situations inwhich more than 50% of all species are exposed to a concen-tration exceeding their EC10 or NOEC. These numbers arebased on situations with a fixed application of 6 t/ha dryweight, but without a ceiling for LAS in sludge. In Denmark,these numbers will be different due to the existing cut-off valuefor LAS at 1,300 mg/kg.

Field observations

Reports on field observations of the effects of LAS in sludgehave been reviewed by Jensen [1]. The available informationshows that no changes are normally observed in the field com-munities of plants, animals, or soil functions after sludge ap-plication. Malkomes and Wohler [32] observed, however, anoverall negative effect on the soil respiration at LAS concen-trations of 10 kg/ha applied via solutions. During the first fourmonths, effects on the adenosine triphosphate content and de-hydrogenase activity were also observed. Litz et al. [33] didnot find any reduction in yield of ryegrass after harvest. Theyobserved considerable short-term, acute physiological damageafter application of aqueous LAS solutions at concentrationsof up to 500 kg/ha. No damage was reported for the applicationrate of 50 kg/ha. An application of 10 kg/ha is expected tofall within the range of a normal sewage sludge application,whereas a rate of 500 kg/ha is unlikely to occur in practice.Using the same assumptions as in the PEC calculations above,LAS doses of 10 and 500 kg/ha approximately correspond tosoil concentrations of 4.5 and 225 mg/kg, respectively. Mostfield studies have observed that the high content of organicmaterial and essential nutrients in sewage sludge generallystimulate the microbial activity and, hence, also indirectly ordirectly the abundance of soil fauna and growth of plants.Sewage sludge has been shown to stimulate collembolans,mites, and nematodes [34–36]. The stimulation of these groupsof soil invertebrates by sewage sludge is generally explainedby an increase in microbial biomass, thus increasing the avail-able food resources [37]. On the other hand, Brendecke et al.[38] did not find any detectable long-term changes in microbialpopulations or activity after four years of repeated sludge ap-

plications of 2 or 6 t/ha/year. Jensen and Krogh [39] did notobserve any short-term or long-term (four years) adverse ef-fects on nine different microbial functions/processes or theabundance and diversity of microarthropods and earthwormsafter sludge application of up to 21 t/ha dry weight, corre-sponding to a LAS dose of approximately 35 kg/ha, or ap-proximately 15 mg/kg dry weight. Thus, based on field ob-servations, it is concluded that LAS doses of 10 to 50 kg/ha,or average soil concentrations of 5 to 15 mg/kg, do not seemto be critical for soil ecosystems on the long term. This isconsistent with the statistically derived PNEC of the presentstudy.

DISCUSSION AND CONCLUSIONS

A realistic worst-case estimation of the LAS concentrationin soil after sludge amendment is predicted to be 7 mg/kg dryweight, which should be compared to the estimated PNEC of4.6 mg/kg. Under Danish circumstances, this leads to an initialrisk ratio (i.e., PEC/PNEC) of approximately 1.5 in a realisticworst-case consideration. The concentration of LAS will dropto a level below the PNEC 6 to 24 d after sludge application,depending on the degradation rate of LAS. In countries nothaving similar sludge regulations, the estimated risk ratio mayinitially be considerably higher. However, also in very excep-tional situations, the environmental risk (i.e., PEC/PNEC) willbe reduced after sewage application due to a relatively rapiddegradation of LAS in soil. Even in the most extreme scenario(i.e., annual LAS load . 160 kg/ha and half-life of 40 d), thelevel of LAS is expected to be well beyond the estimated PNECone year after application (i.e., ;0.1 mg/kg vs 4.6 mg/kg).This may give the ecosystem time to recover from potentialdamage caused by LAS before an eventual, new soil amend-ment.

The exceedance profile of sewage sludge–associated LAS(Fig. 1) reveals that at a sludge dose of 6 t/ha dry weight, only0.5% of all soil fauna have a probability greater than 50% ofbeing exposed to LAS concentrations exceeding their EC10or NOEC. Approximately 50% of the species will, in 9 of 10cases, be exposed to LAS concentrations below their EC10 orNOEC, whereas only 1% of all species will hardly ever (.99%of cases) be exposed to concentrations above their EC10 orNOEC.

The final interpretation of whether risks caused by LAS insewage sludge are acceptable is a matter of what should beprotected and the level of certainty on which that protectionshould be based. However, based on the shape of the exceed-ance profiles and the risk characterization presented here, therisk could be regarded as being acceptable.

That the risk may be acceptable is confirmed by monitoringstudies of effects in field situations. Application of LAS-con-taining sludge has typically not caused adverse effects on soilecosystems, even at LAS concentrations showing significanteffects on various groups of organisms in laboratory studies[1]. Sludge is not likely to be homogeneously distributed intothe soil. Therefore, soil organisms may avoid exposure, whichis not possible in laboratory experiments with homogeneouslydistributed sludge and LAS in the test soils. Soil ecosystemsin agricultural fields may also be more ductile than anticipatedin laboratory risk-assessment procedures. Another explanationof the discrepancy between risk assessment as based on effectsobserved in the laboratory and on effects observed in the fieldmay be that the bioavailability of LAS is significantly reducedwhen LAS is applied via sludge. Immobilization of LAS may

1696 Environ. Toxicol. Chem. 20, 2001 J. Jensen et al.

be enhanced as a result of the organic content and/or a largenumber of particles in sludge. On the other hand, LAS in-creases the mobility of, for example, heavy metals, polycyclicaromatic hydrocarbons, or persistent organic pollutants in soil,which is an ability used during bioremediation of soils pollutedwith heavy metals and polycyclic aromatic hydrocarbons [40].The LAS may also stimulate the uptake of organic pollutantsor pesticides into plants or animals [41]. Haigh [42] and Foxet al. [43] have, however, shown that indirect effects are notlikely to occur at LAS concentrations normally present in do-mestic sewage sludge. In recent experiments, Holmstrup et al.[7] found that sewage sludge did not influence the toxicity ofLAS to soil fauna. Toxicity toward microorganisms, on theother hand, was reduced but not eliminated [5]. Based on theseresults, modifying effects of sewage sludge were excluded inthe present effect assessment.

The present risk assessment concludes that LAS does notpose a significant risk to fauna, plants, and essential functionsof agricultural soils as a result of normal sewage sludge amend-ment. However, risks have been identified in worst-case sce-narios. When evaluating and managing the risk of LAS toterrestrial ecosystems, it is, therefore, important to considerthe following aspects: First, with a half-life of generally lessthan one month, LAS is relatively rapidly degraded in the soilsystem and not susceptible to accumulation in soil or biota asa result of normal agricultural practice. Second, LAS does notaccumulate in crops. Even at sewage sludge doses exceedingDanish practice by more than an order of magnitude, LAS isnot taken up by edible plants in harmful or intolerable amounts[44]. Third, the toxic mechanism of LAS is well known andmost likely due to perturbation of biomembranes. Fourth, nolong-term, adverse effects of LAS or LAS-containing sewagesludge in the field have been reported. Fifth, the predominatingroute of LAS to soils is via sewage sludge amendment of arableland. The impact of inorganic fertilizers, lime, pesticides, an-nual tilling, compaction by large vehicles, and other soil al-terations associated to normal agricultural practice may, inmany cases, significantly exceed the potential effects of LASand other constituents of sewage sludge on the ecosystems ofagricultural soils.

Acknowledgement—The financial support given by Association In-ternationale de la Savobnnerie de la Detergence des produitsd’Entretien, Comite Europeen des Agents de Surface et leurs Inter-mediaires Organiques, European Centre of studies on LAB/LAS,Council for LAB/LAS Environmental Research and the Danish En-vironmental Research Programme is acknowledged.

REFERENCES

1. Jensen J. 1999. Fate and effects of linear alkylbenzene sulfonate(LAS) in the terrestrial environment—A review. Sci Total En-viron 226:93–111.

2. Kloepper-Sams P, Torfs F, Feijtel T, Gooch J. 1996. Effects as-sessments for surfactants in sludge-amended soils: A literaturereview and perspectives for terrestrial risk assessment. Sci TotalEnviron 185:171–185.

3. De Wolf W, Feijtel T. 1998. Terrestrial risk assessment for linearalkyl benzene sulfonate (LAS) in sludge-amended soils. Che-mosphere 36:1319–1343.

4. Elsgaard L, Petersen SO, Debosz K. 2001. Effects and risk as-sessment of linear alkylbenzene sulfonates in agricultural soil. 1.Short-term effects soil microbiology. Environ Toxicol Chem 20:1656–1663.

5. Elsgaard L, Petersen SO, Debosz K. 2001. Effects and risk as-sessment of linear alkylbenzene sulfonates in agricultural soil. 2.Effects on soil microbiology as influenced by sewage sludge andincubation time. Environ Toxicol Chem 20:1664–1672.

6. Holmstrup M, Krogh P. 2001. Effects and risk assessment of linearalkylbenzene sulfonates in agricultural soil. 3. Sublethal effectson soil invertebrates. Environ Toxicol Chem 20:1673–1679.

7. Holmstrup M, Krogh PH, Løkke H, de Wolf W, Marshall S, FoxK. 2001. Effects and risk assessment of linear alkylbenzene sul-fonates in agricultural soil. 4. The influence of salt speciation,soil type, and sewage sludge on toxicity using the collembolanFolsomia fimetaria and the earthworm Aporrectodea caliginosaas test organisms. Environ Toxicol Chem 20:1680–1689.

8. European Commission. 1996. Technical guidance document insupport of commission directive 93/67/EEC on risk assessmentfor new notified substances and commission regulation (EC)1488/94 on risk assessment for existing substances. Part II. Eco-logical risk assessment. Brussels, Belgium.

9. Wagner C, Løkke H. 1991. Estimation of ecotoxicological pro-tection levels from NOEC toxicity data. Water Res 25:1237–1242.

10. Aldenberg T, Slob W. 1993. Confidence limits for hazardous con-centrations based on logistically distributed NOEC data. Ecotox-icol Environ Saf 25:48–63.

11. Scott-Fordsmand JJ, Pedersen MB, Jensen J. 1996. Setting a soilquality criterion. Toxicol Ecotoxicol News 3:20–24.

12. Hoekstra JA, van Ewijk PH. 1993. Alternatives for the no-ob-served-effect level. Environ Toxicol Chem 12:187–194.

13. Van der Hoeven N. 1997. How to measure no effect? Part III.Statistical aspects of NOEC, ECx and NEC estimates. Environ-metrics 8:255–261.

14. Van Straalen NM, van Rijn JP. 1998. Ecotoxicological risk as-sessment of soil fauna recovery from pesticide application. RevEnviron Contam Toxicol 154:3–141.

15. SAS Institute. 1993. SAS/INSIGHTt User’s Guide, Ver 6, 2nded. Cary, NC, USA.

16. Van Beelen P, Doelman P. 1997. Significance and application ofmicrobial toxicity tests in assessing ecotoxicological risks of con-taminants in soil and sediment. Chemosphere 34:455–499.

17. Van Straalen NM, van Gestel CAM. 1993. Soil invertebrates andmicro-organisms. In Calow P, ed, Handbook of Ecotoxicology,Vol 1. Blackwell Scientific, Oxford, UK, pp 251–277.

18. Tørsløv J, Samsøe-Petersen L, Rasmussen JO, Kristensen P. 1997.Use of waste products in agriculture. Contamination level, en-vironmental risk assessment and recommendations for quality cri-teria. Environmental Project 366. Danish Environmental Protec-tion Agency, Ministry of Environment and Energy, Copenhagen,Denmark.

19. Prats D, Ruiz F, Vazquez B, Zarzo D, Berna JL, Moreno A. 1993.LAS homolog distribution shift during waste-water treatment andcomposting—Ecological implications. Environ Toxicol Chem 12:1599–1608.

20. Marcomini A, Capel PD, Lichtensteiger TH, Brunner PH, GigerW. 1989. Behaviour of aromatic surfactants and PCBs in sludge-treated soil and landfills. J Environ Qual 18:523–528.

21. Holt MS, Matthijs E, Waters J. 1989. The levels and fate of linearalkylbenzene sulphonate (LAS) in sludge amended soils. InQuaghebeur D, Temmerman I, Angeletti G, eds, Organic Con-taminants in Waste Water, Sludge and Sediment. 1. Waste Water.Organic Pollutants. Elsevier Science, Amsterdam, The Nether-lands, pp 161–182.

22. Waters J, Holt MS, Matthijs E. 1989. Fate of LAS in sludgeamended soils. Tenside Surfactants Detergents 26:29–135.

23. Holmstrup M, Krogh PH. 1996. Effects of an anionic surfactant,linear alkylbenzene sulfonate, on survival, reproduction andgrowth of the soil living collembolan, Folsomia fimetaria. En-viron Toxicol Chem 15:1745–1748.

24. Rundgren S, van Gestel CAM. 1998. Comparison of species sen-sitivity. In Løkke H, van Gestel DAM, eds, Handbook of SoilInvertebrate Toxicity Tests. John Wiley, Chichester, UK, pp 41–55.

25. Schwuger MJ, Bartnik FG. 1980. Interaction of anionic surfac-tants with proteins, enzymes, and membranes. In Gloxhuber C,eds, Anionic Surfactants—Biochemistry, Toxicology, Dermatol-ogy. Marcel Dekker, New York, USA.

26. Larson RJ, Payne AG. 1981. Fate of the benzene ring of linearalkylbenzene sulfonate in natural waters. Appl Environ Microbiol41:621–627.

27. Swisher RD. 1987. Surfactant Biodegradation. Marcel Dekker,New York, NY, USA.

28. Chapman PM, Fairbrother A, Brown D. 1998. A critical evalu-

Risk assessment of LAS Environ. Toxicol. Chem. 20, 2001 1697

ation of safety (uncertainty) factors for ecological risk assessment.Environ Toxicol Chem 17:99–108.

29. Forbes TL, Forbes VL. 1993. A critique of the use of distribution-based extrapolation models in ecotoxicology. Funct Ecol 7:249–254.

30. Schoberl P. 1989. Basic principles of LAS biodegradation. Ten-side Surfactants Detergents 26:86–94.

31. Gisey JP, Solomon KR, Coats JR, Dixon KR, Giddings JM, Ke-nega EE. 1999. Clorpyrifos: Ecological risk assessment in NorthAmerican aquatic environments. Rev Environ Contam Toxicol160:1–129.

32. Malkomes HP, Wohler, B. 1983. Testing and evaluating somemethods to investigate side effects on environmental chemicalson soil micro-organisms. Ecotoxicol Environ Saf 7:284–294.

33. Litz N, Doering HW, Thiele M, Blume HP. 1987. The behaviourof linear alkylbenzenesulfonate in different soils: a comparisonbetween field and laboratory studies. Ecotoxicol Environ Saf 14:103–116.

34. Glockemann B, Larink O. 1989. Einfluß von Klarschlammdun-gung und schwermetallbelastung auf Milben, speziell Gamasiden,in einem Ackerboden. Pedobiologia 33:237–246.

35. Pimentel D, Warneke A. 1989. Ecological effects of manure, sew-age sludge and other organic wastes on arthropod populations.Agricultural Reviews 3:1–30.

36. Weiss B, Larink O. 1991. Influence of sewage sludge and heavymetals on nematodes in an arable soil. Biol Fertil Soils 12:5–9.

37. Fließbach A, Martens R, Reber HH. 1994. Soil microbial biomassand microbial activity in soils treated with heavy metal contam-inated sewage sludge. Soil Biol Biochem 26:1201–1205.

38. Brendecke JW, Axelson RD, Pepper IL. 1993. Soil microbialactivity as an indicator of soil fertility: Long-term effects of mu-nicipal sewage sludge on an arid soil. Soil Biol Biochem 25:751–758.

39. Jensen J, Krogh PH. 1999. Ecological assessment of sewagesludge application. Proceedings, Nordiska JordbruksforskaresForening, Seminar 292. Jokionen, Finland, November 23–25,1998, pp 98–100.

40. Aronstein BN, Alexander M. 1992. Surfactants at low concen-trations stimulate biodegradation of sorbed hydrocarbons in sam-ples of aquifer sands and soil slurries. Environ Toxicol Chem 11:1227–1233.

41. Lichtenstein EP, Fuhremann TW, Scopes NEA, Skrentny RF.1967. Translocation of insecticides from soils into pea plants. JAgric Food Chem 15:864–869.

42. Haigh SD. 1996. A review of the interaction of surfactants withorganic contaminants in soil. Sci Total Environ 185:161–170.

43. Fox KK, Chapman L, Solbe J, Brennand V. 1997. Effect of en-vironmentally relevant concentrations of surfactants on the de-sorption or biodegradation of model contaminants in soil. TensideSurfactants Detergents 34:436–441.

44. Grøn C, Laturnus F, Mortensen GK, Egsgaard H, Bennetzen S,Ambus P, Jensen ES. 1999. Plant uptake and soil degradation oforganic contaminants in sludge amended soil. Proceedings, 217thAmerican Chemical Society National Meeting, Anaheim, CA,USA, March 21–25, p U716.

45. Løkke H, van Gestel CAM. 1998. Handbook of Soil InvertebrateToxicity Tests. John Wiley, Chichester, UK.

46. Monsanto Chemical Company. 1990. Acute toxicity of linear al-kylbenzene sulfonate to earthworms (Eisenia foetida). Report38316. St. Louis, MO, USA.

47. Marschner A. 1992. Phytotoxizitatsuntersuchungen mit zwei an-ionischen Detergentien (TPBS, LAS) und einem Herbizid (Atra-zin). Schriftenr Ver Wasser Boden Lufthygiene 89:459–483.

48. Gunther P, Pestemer W. 1992. Phytotoxicity of surfactants to high-er plants. In Hall JE, Sauerbeck DE, Hermite PL, eds, Effects ofOrganic Contaminants in Sewage Sludge on Soil Fertility, Plantsand Animals. Office for Official Publications of the EuropeanCommunities, Brussels, Belgium, pp 103–111.

49. Unilever. 1987. Effects on the growth of Sorghum bicolor, He-lianthus annuus and Phaselous aureus. Study Report BL/B/3078.Unilever Research Port Sunlight Laboratory, Port Sunlight, UK.

50. Mieure JP, Waters J, Holt MS, Matthijs E. 1990. Terrestrial safetyassessment of linear alkylbenzene sulfonate. Chemosphere 21:251–262.