S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 6 ( 2 0 0 8 ) 5 1 8 – 5 2 2

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te / sc i to tenv

Ecological risk assessment in legislation oncontaminated soil in The Netherlands

Alexandra E. Boekhold⁎

Soil Protection Technical Committee, P.O. Box 30947, 2500 GX The Hague, The Netherlands

A R T I C L E I N F O

⁎ Tel.: +31 70 3393034; fax: +31 70 3391342.E-mail address: boekhold@tcbodem.nl.

0048-9697/$ – see front matter © 2008 Elsevidoi:10.1016/j.scitotenv.2008.07.018

A B S T R A C T

Article history:Received 8 July 2008Accepted 14 July 2008Available online 30 August 2008

Recently theDutch soil policywas revised includingnew rules for the relocation of contaminatedsoil and dredged soil material. With these rules, new methods for ecotoxicological riskassessment were implemented. One of the new methods is the assessment of the local toxicpressure of mixtures, also known as the ms-PAF- method, based on the Species SensitivityDistribution concept. Thems-PAFmethod is applied for risk assessment of spreading of dredgedsoil material on adjacent land. Its application will possibly be extended to the derivation of localsoil quality standards relevant in the context of soil relocation. The application of the local toxicpressure will probably increase the reuse of contaminated soil and dredged soil material andhence will reduce the amounts considered to be unfit for use. With this method, local ecologicalrisk limits are derived using pore water concentrations and effects on water organisms. Porewater concentrations are subsequently transferred to total soil concentrations using empiricalrelationships. The methodology does not impose upper limits for total soil concentrations. Insoils with a high sorption capacity, total soil concentrations that are considered to be acceptablemay be several times higher than the current Dutch intervention values. The possibleintroduction of the ms-PAF method will open the door to local soil relocation with soilscontaining large amounts of (semi-permanently soil bound) contaminants. Since the ms-PAFmethod is not yet properly validated, the lack of evidence of ecological effects usingmodels likethems-PAFmethod cannot be regarded as an indication for the absence of effects in reality. TheDutch soil quality decree would gain environmental ambition when the ms-PAF method wascombined with a realistic upper limit on total soil concentrations. This would preventcontamination of land by means of soil relocation.

© 2008 Elsevier B.V. All rights reserved.

Keywords:Ecological risk assessmentRelocationContaminated soil

1. Introduction

Recently the Dutch soil policy was revised and a new soilquality decree will come into force in 2008. In previous yearssoil clean up and the redevelopment of contaminated sitesattractedmost of the attention. The new soil quality decree, asa part of the soil protection act, regulates among others the useand reuse of contaminated (relocated) soils. With this decree abalance is sought between prevention of spreading of con-tamination as a result of relocation of contaminated soil onthe one hand, and reducing waste by maximally reusing

er B.V. All rights reserved

contaminated soils on the other hand. The new rules replacethe presentmethodology as described in a statutory regulationunder the Building Materials Decree. With this revision, thepolicy shifts, ‘from soil protection only to sustainable soil use',and ‘from national to local’, as announced in the soil policyletter of 2003 (State Secretary of Housing, Spatial Planning andthe Environment, 2003). Excavated and contaminated soil is nolonger primarily conceived as a buildingmaterial. Instead, it isseen as material that after relocation becomes integral partof the land again. The significance of this difference is thatquality standards for buildingmaterials are independent of the

.

519S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 6 ( 2 0 0 8 ) 5 1 8 – 5 2 2

quality of the land underneath the building, whereas therelocation of soil depends on both standards for the quality ofthe soil to be relocated as well as standards for the quality ofthe receiving land.

Soil quality standards are derived using both human andecological risk assessmentmethods. This contribution focuseson ecological risk assessment. The possible introduction ofa new method to account for mixture effects is discussed.Ecological risk-based soil quality assessment will probably beimplemented in the member states of the European Unionin the context of an anticipated soil framework directive(Swartjes et al., 2008-this issue). Therefore, the Dutch discus-sions and choices in ecological risk assessment in the contextof legislation and decision-making may be of broader interest.

2. Quality standards for the reuse ofcontaminated soil in The Netherlands

The new soil quality decree prescribes a tiered approach inrisk assessment of contaminated soils to be reused. Besides anational soil policy, municipalities are encouraged to developa local soil policy, using higher tier risk assessment methods.A locally developed soil policy subsequently replaces (part of)the national soil policy. In the next paragraph the nationalsoil policy is described, the paragraph thereafter addresses thelocal soil policy.

2.1. National references for soil quality

National soil policy includes the use of national soil qualitystandards. These national soil quality standards are called soilreferences or maximum values. These soil quality standardswere derived for different formsof landuse (Wezenbeek, 2007).Based on model calculations, the maximum allowed concen-trations of contaminants were derived at which land use is notrestricted due to human or ecological risks. These models usenumerous assumptions regarding human exposure and theimportance of ecological functioning as a function of land use.Land in use for agriculture and nature is considered to be of agood quality for that use when it holds to its standards. Theselands must keep their good quality. The accompanying soilquality standards for these soils equal the current backgroundconcentrations of contaminants in The Netherlands. Agricul-ture and nature form are two categories of land use. Natureis supposed to be ‘natural’ and ‘not contaminated’, whereasgood soil quality for agriculture is required with regard to foodsafety and the international reputation of Dutch agriculturalproducts. Residential soils are allowed to contain slightlyraised concentrations of contaminants. However, these mustremain at sufficiently low concentrations so that human risksare acceptable with residential land use (housing, garden-ing, playing children). The ecological protection goals in resi-dential land are less ambitious than in agricultural land ornature. Land used for industry is even allowed to containsomewhat higher concentrations of contaminants, assuminga lowhumanexposure since the soil is primarily used as a basefor buildings. Industrially used land is of limited ecologicalimportance, since the presence of ‘green’ areas is limited andsupposedly of little ecological quality. Based on these types of

land use, four classes of soil are distinguished on the basis ofthe concentration of contaminants:

1. soil fit for use in all situations;2. soil fit for use in residential and industrial areas;3. soil fit for use in industrial areas;4. soil not fit for relocation.

Class 4 soils must be cleaned or treated as waste.Receiving land is classified in a similar manner:

1. the land is fit for all uses;2. the land is fit for residential and industrial use;3. the land is only fit for industrial use;4. the land is not fit for any use and must be remediated.

The actual or intended land use ultimately determinesthe quality of soil that can be relocated onto receiving landwith a certain soil quality. Land used in a way that poses lowhuman and ecotoxicological risks may receive soil from asimilar quality class or better. However, this is limited by theso-called stand still principle. Stand still in this context meansthat soil cannot be relocated on receiving land of a better soilquality class, even when the intended soil use would allow so.In this way, the quality of the land does not change into alower quality class due to relocation of contaminated soil.Some deterioration of soil quality may occur within the soilclass. For instance, consider a soil classified on the basis of acertain substance. When this soil is relocated onto land ofthe same class that was classified on the basis of anothersubstance, the quality class does not change. Nevertheless,the receiving land is now contaminated with two substancesinstead of one. In the process of decision-making, this effectwas considered to be ecologically insignificant, as long as classlimits are not exceeded.

The class boundaries are based on human and ecotoxico-logical risk limits. The class boundaries are called ‘references’or ‘maximum values’. In this contribution only the ecotox-icological risk assessment is discussed.

2.2. Local references for soil quality

The decree on soil quality encourages local authorities (mu-nicipalities and the district water boards) to develop a localpolicy on the relocation of soils. A local soil policy involves,among others, the derivation of local soil class boundaries onthe basis of human and ecological risk limits that are derivedusing local information on soil characteristics and soil use.Municipalities may use a methodology prescribed by the stategovernment offered to them in an electronic format calledthe risk toolbox. With the risk toolbox these local soil qualitystandards (local references or local maximum values) can bederived. In short, the risk toolbox calculates the risks associatedwith a certain concentration of the contaminant in soil, as afunction of local parameters such as soil type, soil propertiesand soil use. The methodology can also be used to perform thecalculations vice versa, so that themaximumconcentrationof acontaminant can be calculated based on a predefined poten-tially affected fraction of species in the soil ecosystem that isregarded to be acceptable for the corresponding land use. The

520 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 6 ( 2 0 0 8 ) 5 1 8 – 5 2 2

SSDand thems-PAFmethod formthebasis of theecological riskassessment in the risk toolbox. SSD is the abbreviation ofSpecies Sensitivity Distribution, ms-PAF stands for multisub-stance Potentially Affected Fraction, see paragraph 3 on themethodology forderiving risk limits for further explanation.Therisk toolbox will eventually consist of first, second and highertiered risk assessment methods. The first tier is the most ge-neral method. The higher tiers use gradually more site-specificinformation, for instance local information on soil type, pH,and bioassays. Ultimately even ecological field observationscanbeused.TheobservationbyPosthumaetal. (2008-this issue)that second tier risk assessment using site-specific analyses isneeded to solve current policyproblems, is alreadypartly settledin the Dutch soil policy instruments for risk assessment in thehigher concentration ranges (above intervention values), and isnow being adopted in the context of soil relocation.

So far, thems-PAFmethodmay only be used in the decisionmaking process to determine the ecotoxicological risk of thedeposition of dredged soil material onto adjacent land. It iscurrently under investigation by the Dutch government whe-ther the ms-PAF method may also be applied to give a betterscientific basis to local references. A strongly policy driven sti-mulus to this investigation is the expectation that local soilquality standards will become less stringent, which reduces theeffort needed by society for soil remediation.

The national soil quality standards (or national references)were derived using the same overall risk assessment metho-dology as offered by the risk toolbox, but because of the nation-wide application of these standards, they were derived usingfirst tier risk assessment only. As with all Dutch soil qualitystandards, the national references are defined for a soil with10% organic matter and 25% clay content. With so-called soiltype correction formulas they can be converted to standardsapplicable to soils with other percentages organic matter andclay. These formulas were originally derived for the interven-tion values and the target values and now also apply to thenational references, with the exception of the references forsoils fit for use in all situations. The latter references arebased on background concentrations of substances at loca-tions where no history of contamination is suspected.

3. The methodology for deriving risk limits

The basic scientific approach used in the derivation ofecotoxicological risk limits defining class boundaries is offeredby the Species Sensitivity Distribution (SSD) method (Post-huma et al., 2002). The related multisubstance potentiallyaffected fraction, ms-PAF, is used to regulate the placement ofsludge from ditches (dredged soil material) on adjacent, oftenagricultural land (De Zwart and Posthuma, 2005; Osté andWintersen, 2006). Both methods use the distribution of thesensitivities of different soil organisms to chemical stress asan indicator of ecological risks. Detailed information on thesemethods can be found in the cited literature, here only thefeatures relevant for this paper are highlighted.

The derivation of ecotoxicological risk limits using the SSD-andms-PAFmethod dependshighly on the availability of toxicitydata as reported in international literature. It also involves anumber of methodological decisions. For instance, the SSDs for

the different contaminants are generally based on No ObservedEffect Concentrations (NOECs). An alternative may be the use ofeffect parameters, as opposed to no effect. The relevance ofdifferent effects for ecosystem services including the relevantmagnitudeof the effectmust thenbe assessed. An example is theoften-used EC50 on mortality, which is the concentration atwhich 50% of the species die. Also the use of either chronic oracute toxicity data or a combination thereof must be considered.Other decisions involve the selection of data (which organismsare included, how to deal with the natural background concen-trations, the number and quality of the data). For most con-taminants data are only available for a limited amount of species.Nevertheless, out of necessity the available data are regardedas being representative for the overall spectrum of speciessensitivities.

Ecotoxicological risk limits based on effect data for aqua-tic organisms may differ from limits based on effect data forsoil organisms (Van Beelen et al., 2002). However, although wa-ter organismsmay differ significantly from soil organisms,manyorganisms are assumed to be physiologically related. Moreover,muchmorewaterdataareavailable, and it is argued thatbothsoiland water organisms are exposed to contaminants via the waterphase. This may be true for soft-bodied organisms like earthworms, but is seldom tested for hard-bodied organisms. Equili-briumpartitioningdoesnot consider (active) uptakemechanismsinduced by soil ingestion (Sijm et al., 2002). Nevertheless, afterample discussion in the soil policy arena it was chosen to derivethe soil ecological risk limits on SSDs derived using pore waterconcentrations and effects on water organisms (Wezenbeek,2007).

It is therefore relevant to be able to predict pore waterconcentrations from total soil concentrations and vice versa.Many efforts have been taken to developmodels for prediction ofthe concentration in the pore water of the soil from total soilconcentrations as a function of soil properties, including pH, claycontent and organic matter content. For Cd and Zn empiricalrelationships appeared to be quite successful, andmany transferfunctions have been published (for instance Janssen et al., 1997;Römkens et al., 2004). For other heavy metals the relationshipsbetween total concentrations and concentrations in the porewater as a function of soil properties appeared to be less straight-forward. Only Cu and Ni show some correlation, although with amuch lower correlation coefficient than Cd and Zn. Instead ofthese statistically derived empirical relationships, mechanisticmodels are preferred because only then the soil processes thatdictate bioavailability can really be understood, andmanipulatedwhen necessary (Van Riemsdijk, 2006). Unfortunately, mechan-istic models are still under development and not yet ready to beimplemented in legislation. International guidance is now beingdeveloped for a harmonized framework of methods to assessbioavailability for different target species with regard to severalclasses of contaminants (Harmsen, 2007), so that risk assessmentis based on common methodological concepts regarding expo-sure and effects.

4. Discussion

The SSD-method has proven to be a very successful tool inquantifying ecological risks for decision-making purposes.

521S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 6 ( 2 0 0 8 ) 5 1 8 – 5 2 2

The presence of a ready-to-usemethodology for ecological riskassessment made it possible to incorporate ecological con-siderations into theDutch soil policy and legislation during thepast 20 years. Themethodology was thoroughly improved andextended during that period. Recently the Dutch governmentdecided to use these improvements in the new soil qualitydecree, as an instrument to help local authoritieswith derivinglocal soil quality standards with which the spreading ofdredged soil material onto adjacent land is regulated. In thecontext of regulation of relocation of contaminated soil, themethod is currently discussed as a tool for the decisionmakingprocess. Question is whether the ms-PAF-method is an ade-quate method to be used in the risk toolbox to derive localreferences. Besides scientific discussion on the methodology,also policy-oriented arguments are relevant. Questions to beansweredare dual: Pragmatically: does thenewmethod lead toa significant change in volume of relocated soils, what costs orbenefits are involved, and when, and for what party concer-ned. Environmentally: does the new methodology lead to asignificant better decision in terms of sustainable soil use onboth local and national scale, and what is needed to furtherimprove risk assessment of contaminants?

In general the new local soil quality standards based on thems-PAFmethod will be less rigid than before. The added risk ofall substancespresent usually leads tomore stringent risk limitsthan a separate assessment for each substance individually.However, the new method also assumes that soil binding ofcontaminants protects organisms against exposure to contami-nants in soil. This relaxing effect appears to be stronger thanthe opposite effect of multisubstance risk assessment (Post-huma et al., 2006).

Summarizing, thems-PAFmethodology as currently imple-mented in the risk toolbox establishes the ecological riskof mixtures of contaminants in soil on the basis of pore wa-ter concentrations and the effects of contaminants on waterorganisms.

Corresponding total soil concentrations are calculatedusing empirical soil-pore water relationships that are hardlyvalidated for most compounds. The major underlying as-sumption is that pore water concentrations and effects aspredicted for water organisms are adequate predictors of soilecological functioning. A certain percentage of the soil speciesis allowed to be potentially affected by the soil contami-nants, under the implicit assumption that the remaining spe-cies form a soil ecosystem that is sufficiently resilient andmaintains its vital soil functions. This assumption appliesequally to the more classical single substance risk limits, butis not validated. On the contrary, evidence is growing thatinterspecific functional dissimilarity among species may bemore important than species number for soil ecosystemprocesses (Heemsbergen et al., 2004). Therefore it is not at allsure that the contaminated soil is indeed harmless to theintended land use. Moreover, environmental conditions maychange over time due to changes in land use or processes likeglobal warming. This could lead to mobilization of contami-nants and an increase of risks. As proposed by Posthuma et al.(2008-this issue) it is evident that local soil authorities shouldaddress various scenarios of future land use too, in order toavoid allowance of soil relocation that will hamper future soiluse.

Although location specific risk assessment addresses re-levant risk-related aspects that cannot be captured in a ge-neric compound-oriented approach, the true understandingof the outcome of an ecological risk assessment is limitedby the lack of understanding of its meaning for ecologicalservices. The introduction of a more detailed and sophisti-cated methodology in the decision making process, i.e. thems- PAF-method, suggests a result of higher quality, but this isof course not necessarily the case.

The lack of proper validation is especially relevant in soilswith a high sorption capacity. These soils combine high totalsoil concentrations with relatively low pore water concentra-tions. According to the ms-PAF method, these soils are con-sidered to be fit for use, and the relocation of these soils issubsequently not limited by local soil quality criteria. In soilswith high sorption capacity total soil concentrationsmay thenrise orders of magnitude above the current Dutch interventionvalues. A high sorption capacity can have different physicaland chemical causes, i.e. high in content of clay, organicmatteror metal oxides, high in pH, etcetera. Environmental condi-tionsusually influence the actual sorption capacity. The lack ofevidence of ecological effects as calculated using the ms-PAF-methodology cannot be regarded as a lack of possible effects inreality now, nor in the future. The decision maker cannot beconfident that allmethodological assumptionsmade are valid,which is of specific relevance at high total soil concentrations.The scientific community should avoid anymisunderstandingon this. Without knowledge of the relevant mechanisms ofecosystem functioning and the main aspects that influenceecosystem services, legislation on soil use and reuse shouldnot rely on methods that allow the reuse of soils with a hightotal concentration of contaminants.

A poor functioning soil ecosystemmay lead to a poor waterinfiltration capacity, or to a poor water filtering capacity, or toa poor habitat for agriculture and gardening. Land should beprotected against relocation of soil that is in fact contami-nated although thems-PAF-methodology does not indicate so.This protection can easily be achieved.

The soil quality decreewould gain environmental ambitionif a realistic upper limit on total soil concentrations were set asa boundary for soil relocation and the spreading of dredged soilmaterial on land adjacent to waterways. It is recommended todefine a maximum total soil concentration of contaminantsabove which soil relocation is not permitted, not even in localsoil policy. The current intervention values are not necessarilya good choice for this maximum value. The derivation ofintervention values is based on the same SSD-methodologyand thus also assumes that species numbers are important.Even more so, many consultants doubt the alleged ecologicalrisks of soil above the intervention values, at least for certainsubstances, based on their field experience (Labee and Bos,2000). An alternative maximum value may thus be developedin the specific context of risk assessment of contaminated soilusing multi-substance approaches and bioavailability asmodel concepts. It may be difficult to give a solid scientificbasis to such a maximum value. Nevertheless it preventsseverely contaminated soils from being labeled as ‘little risks’on an uncertain scientific basis.

In fact during the recent years the scientific communityhas not been able yet to predict ecotoxicological risks using

522 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 6 ( 2 0 0 8 ) 5 1 8 – 5 2 2

ecological soil parameters, despite the progress made on riskassessment methodologies. The SSD-method is elegantindeed, but only chemical information is used to derive eco-logical soil quality standards. In the Dutch soil quality act,primarily information on pore water concentrations and ef-fects on water organisms is used, instead of soil concen-trations and effects on soil organisms. Even though thearguments to do so are valid, the SSD-method will becomeeven more elegant when ecological soil quality is definedusing soil ecological parameters. These soil ecological para-meters should describe the relevant ecological processes inview of ecosystem services. This is one of the challenges forfuture research in the field of ecological science applied topolicy instruments. The research findings presented in thisspecial issue are a valuable step forward.

Acknowledgements

This article is written in a private capacity. The Soil ProtectionTechnical Committee has no responsibility for the content of thispaper. I would like to thank my colleagues Dr. Joke vanWensemand Drs. Jaap Tuinstra for commenting on this manuscript.

R E F E R E N C E S

De Zwart D, Posthuma L. Complex mixture toxicity for single andmultiple species: proposed methodologies. Environ ToxicolChem 2005;24:2665–72.

Harmsen J. Measuring bioavailability: from a scientific approach tostandard methods. J Environ Qual 2007;36:1420–8.

Heemsbergen DA, BergMP, LoreauM, Van Hal JR, Faber JH, VerhoefHA. Biodiversity effects on soil processes explained byinterspecific functional dissimilarity. Science2004;306:1019–20.

Janssen RPT, Peijnenburg WJGM, Posthuma L, Van den HoopMAGT. Equilibrium partitioning of heavy metals in Dutch fieldsoils. I. relationship between metal partition coefficients andsoil characteristics. Environ Toxicol Chem 1997;16:2470–8.

Labee F, Bos SC. Eco(toxico)logische risico's vanbodemverontreinigingen veelvuldig overschat. Bodem2000;1:16–7 [in Dutch].

Osté L, Wintersen A. Nieuw klassenindeling waterbodems,normen voor verspreiden en toepassen. Concept of July 5.Lelystad, the Netherlands, RWS-RIZA; 2006 [in Dutch].

Posthuma L, Suter GW, Traas TP. Species sensitivity distributionsin ecotoxicology. Boca Raton, Florida: Lewis Publishers; 2002.

Posthuma L, Lijzen J, Otte PF, De Zwart D,Wintersen A, Osté L, et al.Beslissen over bagger op bodem deel 3. Modellering van risico'sna verspreiding bagger. Report 7117046/2006, National Institutefor Public Health and the Environment, Bilthoven,the Netherlands, RIVM; 2006 [in Dutch].

Posthuma L, Eijsackers HJP, Koelmans AA, Vijver MG. Ecologicaleffects of diffuse mixed pollution are site-specific and requirehigher-tier risk assessment to improve site managementdecisions: a discussion paper. Sci Total Environ 2008;406:503–17(this issue).

Römkens PFAM, Groenenberg JE, Bonten LTC, De Vries W, Bril J.Derivation of partition relationship to calculate Cd, Cu, Ni, Pb,and Zn solubility and activity in soil solutions. Report no. 305,Alterra Green World Research, Wageningen; 2004.

Sijm D, Lijzen J, Peijnenburg W, Sneller E, Traas T, Verbruggen E.Biobeschikbaarheid in beleid … wat er aan vooraf ging en watnog komt, resultaten van een workshop en het beleidsvervolg.Report 607220006/2002, National Institute for Public Health andthe Environment, Bilthoven, the Netherlands; 2002 [in Dutch].

State Secretary of Housing, Spatial Planning and the Environment.Letter setting out the new policy for soil. Letter to the Presidentof the lower house of the States General; 2003.

Swartjes FA, Carlon C, De Wit NHSM. The possibilities for theEU-wide use of similar ecological risk-based soil contaminationassessment tools. Sci Total Environ 2008;406:523–9 (this issue).

Van Beelen P, Verbruggen EMJ, Peijnenburg WJGM. The evaluationof the equilibrium partitioning method using sensitivitydistributions of species in water and soil or sediment. Reportno. 607220005, National Institute for Public Health and theEnvironment, Bilthoven, the Netherlands; 2002.

Van RiemsdijkWH. Een fundamenteel bodemchemisch perspectief,basis voor normstelling voor bodem en water. Bodem2006;4:137–8 [in Dutch].

Wezenbeek J. Normstelling en bodemkwaliteitsbeoordeling,onderbouwing en beleidsmatige keuzes voor de bodemnormenin 2005 en 2006. NOBO-2007-056, eindversie. Ministerie VROMen Grontmij; 2007 [in Dutch].