Derivation of ecotoxicity thresholds for uranium

Derivation of ecotoxicitythresholds for uraniumSteve C.Sheppard a ,),Marsha I.Sheppard a ,Marie-Odile Gallerand b ,Barb Sanipelli aaECOMatters Inc.,Box 430,Pinawa,Manitoba,Canada R0E 1L0b Andra –Agence nationale pour la gestion des de ´chets radioactifs,Chaˆtenay-Malabry ce ´dex,France Received 11November 2003;received in revised form 21April 2004;accepted 29May 2004AbstractAssessment of the risk of impact from most radionuclides is based on the total radiological dose rate to the organism of concern.However,for uranium (U)there can be greater risk from chemical toxicity than radiological toxicity (depending on the isotopic composition).Chemical ecotoxicity of U is dependent on several environmental parameters.The most important are carbonate content,because of the formation of soluble carbonate complexes,and divalent cation content (Ca CC and Mg CC ),because of their competitive interaction with the uranyl ion (UO 2CC ).This study summarizes the literature available to set PNECs (predicted no-effect con-centrations)for chemical toxicity of U to non-human biota.The corresponding radiological doses were estimated,and as expected chemical toxicity proved to be the greater concern.There were limited data from some types of biota;however,PNECs for the types of biota of interest were as follows:terrestrial plants d 250mg U kg ÿ1dry soil;other soil biota d 100mg U kg ÿ1dry soil;freshwater plants d 0.005mg U L ÿ1water;freshwater invertebrates d 0.005mg U L ÿ1water; freshwater benthos d 100mg U kg ÿ1dry sediment;)Corresponding author.Tel.:C 12047532747;fax:C 12047538478.E-mail address:sheppards@ (S.C.Sheppard).0265-931X/$-see front matter Ó2004Elsevier Ltd.All rights reserved.doi:10.1016/j.jenvrad.2004.05.015Journal of Environmental Radioactivity 79(2005)55–83/locate/jenvrad56S.C.Sheppard et al./J.Environ.Radioactivity79(2005)55–83freshwaterfish at water hardnesses of:–!10mg CaCO3Lÿ1(very soft water)d0.4mg U Lÿ1water;–10–100mg CaCO3Lÿ1(soft water)d2.8mg U Lÿ1water;and–O100mg CaCO3Lÿ1(hard water)d23mg U Lÿ1water;or–as a function of hardness d0.26(hardness as mg CaCO3Lÿ1)mammals d0.1mg U kgÿ1body weight dÿ1.Ó2004Elsevier Ltd.All rights reserved.Keywords:Uranium;Ecotoxicity;Hardness;PNEC;Predicted no-effect concentration1.IntroductionAssessment of the risk of impact from most radionuclides is based on the total radiological dose rate to the organism of concern.However,for natural-isotopic-composition uranium(Nat U)and possibly iodine-129(129I)and technetium-99(99Tc), there is greater risk from chemical toxicity than radiological toxicity.Chemical toxicity exceeds radiological toxicity in importance when the decay half-life is very long(i.e.the radionuclide has very low specific activity),the element is not particularly bioaccumulative,and the radiological emissions have low relative biological impact.Uranium,a member of the actinide series,has an atomic number of92and ten radioactive isotopes,but238U(99.27%),235U(0.72%)and234U(0.0055%)are the three most abundant.The dominant isotope by mass,238U,has a physical half-life of 4.5!109a,giving it a very low specific activity(1.24!104Bq gÿ1U).In at least one detailed analysis of U radiotoxicity and chemical toxicity in humans(Kocher, 1989),the concentration of U in kidney exceeded the recommended limit for chemical toxicity(a value of0.1mg kgÿ1fresh weight kidney)while the correspond-ing annual effective dose equivalent remained below1mSv.Thus,in human risk assessments,the chemical toxicity of Nat U is the focus,and the same is expected for non-human biota.The objective of this work was to summarize the literature available to set PNECs (predicted no-effect concentrations)for chemical toxicity of U to non-human biota found in natural ecosystems of temperate biospheres.This paper summarizes the literature;more detailed discussions of the underlying papers are given in Sheppard (2003).2.Overview of uptake and behaviour of UThe environmental behaviour of U,Ra and other U and Th decay chain radio-nuclides has been extensively studied and the subject of previous reviews(Sheppard, 1980;Gascoyne,1992).A significant recent review of behaviour and toxicity wasundertaken by Environment Canada (PSL2,2002;Bird et al.,2002);some of the background information in this introductory section was drawn from that study.Although U has oxidation states of (C III)to (C VI),the two major oxidation states are U(C VI)and U(C IV).The chemical properties of U(C IV)are similar to Th(C IV)and Pu(C IV),but U(C IV)is readily oxidized to U(C VI)as UO 22C .The uranyl ion,UO 22C ,is the most stable U species in an oxidizing solution and the most prevalent form in the environment (Sheppard,1980).The oxidized U(C VI)(uranyl)ion complexes readily with carbonate,phosphate or sulfate ions.In these forms,U is soluble and readily transported.In contrast,under reducing conditions,such as those found in anoxic waters and sediment,U occurs in the tetravalent (U(C IV))state which has a strong tendency to bind to organic material (Koczy,1963)and to precipitate,and is therefore immobile (Gascoyne,1992).Metallic U and particles of insoluble U compounds are not very bioavailable (Durbin and Wrenn,1975).For detailed information,Ribera et al.(1996)provide an excellent review of the behaviour,transfer and fate of U in both terrestrial and aquatic environments,including uptake by plants,animals,fish,aquatic invertebrates and plants and humans.3.Mechanisms of U toxicityAccidental exposures of workers and members of the public to releases of U hexafluoride (UF 6)in the mid-80s reinforced interest in the radiological and chemical toxicity of UF 6and its hydrolysis products.Radiological toxicity of U is primarily manifest in bone sarcomas,whereas chemical and combined chemical and radio-logical effects are reported as nephrotoxicity.Recently,interest in the chemical tox-icity of depleted U (DU)has increased following the Gulf War,where DU was used as strengthening material for various armor piercing munitions (Priest,2001).Also,a large industrial work force engaged in several types of U production operations,having intermittent exposures,has led to several toxicology studies of U to humans.Many of the studies using surrogates,such as rats or dogs,are relevant to this work.Depending on dose,route of administration and chemical form of administered U,exposure to U compounds can produce effects in a variety of mammalian organ systems,including kidney,liver,lung,cardiovascular and central nervous system (Hodge,1973cited by Diamond et al.,1989).The kidney is the most sensitive to uranyl U and is the critical target organ for humans after exposure to UF 6(Morrow et al.,1982).Ribera et al.(1996)state that chemical toxicity is particularly significant in compounds containing Nat U.Hence,UF 6and compounds,soluble in organic fluids,are extremely dangerous,while non-soluble compounds such as oxides (UO 2and U 3O 8),the form most commonly found in the environment,are much less dangerous.Further,Wrenn et al.(1985)stated that only the aqueous hexavalent ion UO 22C causes U chemical toxicity after oral intake.Renal injury observed in most mammalian toxicity studies is usually character-ized histologically by cellular injury and tubular necrosis (Diamond et al.,1989).However,often the repair of acute injury is rapid with complete restoration within 57S.C.Sheppard et al./J.Environ.Radioactivity 79(2005)55–8358S.C.Sheppard et al./J.Environ.Radioactivity79(2005)55–8335d after exposure,depending on the dose.Also associated with the injuries were abnormalities in renal function that are reversible making many of these toxicity studies difficult to assess(Diamond et al.,1989).Most international(USA and Australia)regulatory standards are set at3mg U kgÿ1kidney for injury in humans.The chemical toxicity to other organisms,such as earthworms,bivalves andfish, is usually judged by lack of response(no movement following physical prodding, losing the reflex of valve closure or death).For more detail on the cellular effects of U in animals,the reader is referred to Ribera et al.(1996)and Chassard-Bouchaud (1983,1987).4.Ecotoxicological approachIn general,the intent of ecotoxicology is to describe the potential impact of contaminants on ecosystem components so that good decisions can be made to protect the ecology of an environment.With most species,this means protecting the survival of a population.Thus,it is important to protect against effects on growth and reproduction,not just lethality.For endangered or highly valued species,the objective may be to protect individuals.Not all literature intended to convey ecotoxicity data is of sufficient quality to consider.Conversely,there are data not specifically intended for ecotoxicology that are useful from that perspective.It is recognized that in the absence of directly relevant and fully acceptable data,it is sometimes necessary to consider less-ideal data.Therefore,data acceptance criteria were developed.Data were considered potentially acceptable if they were from a peer-reviewed report or of similar calibre,the culture and experimental organisms and media were reported and were ecologically relevant,the results were consistent within the report (reproducible and monotonic with respect to U concentrations),and there was a credible statistical treatment and interpretation.Obviously,expert judgement was involved.Many papers will report multiple endpoints;indeed the preference in ecotoxicol-ogy is to apply a battery of bioassays because each contaminant will affect one bioassay more than another.In addition,it is common to report several endpoints for a given organism,such as for plants:emergence,growth,height and root elongation.The approach used in this study to select data for PNEC derivation was: -only the most sensitive,ecologically relevant endpoint for a given study and relevant organism is considered;-if different types of relevant organisms are used in a study,they are considered separately;-if a range of media characteristics,such as a series of soils,are used,and they had an important effect on the toxicity in a systematic way,they are considered separately.For the soils where results are considered the same,the effect concentration may be set as the geometric mean among those soils.Wherever possible,the results of studies in the literature were presented as EC25, the concentration of U where the performance of the endpoint was25%diminished compared to the control.This required some interpolation,but was considered preferable to summarizing results from a variety of effect levels as reported in the original papers.When this was not possible,the effect level reported by the researcher was used.5.Toxicity to terrestrial biota5.1.Toxicity to terrestrial plantsThe study showing the greatest sensitivity of plants to U in soil(Table1)is that of Aery and Jain(1998),who investigated the effect of U applied to a sandy-loam soil in 15-L pots.They measured a number of endpoints,many of which showed a toxic response to U.The most sensitive of these was for number of seeds per spike,with an Table1Toxicity information for terrestrial plants,with effect concentrations as mg U kgÿ1dry soilAuthors Biota Species Endpoint EffectconcentrationAery and Jain (1998)crop plant Triticum aestivum L.var.WH-147EC25for seed numberper spike0.5Shevchenko et al.(1987)crop plants corn,sorghum,cottonand sugar beetEC25for yield5Gulati et al. (1980)crop plants Triticum aestivum,Lycopersicon esculentumEC25for biomass yield6Meyer et al. (1998)native rangegrassAristida purpurea Nutt.(purple threeawn);Buchloedactyloides(Nutt.)Engelm.(buffalograss)andSchizachyrium scoparium(Michx.)Nash(littlebluestem)LOEC for biomassincrease50Zhukov andZudilkin(1971citedby Sheppard,1989)crop plant Probably Triticum aestivum EC11on yield50Sheppard et al.(1985)conifer Pinus sylvestris NOEC for growth O100Bufatin et al. (1986)crop plant rice(probably Oryza sativa)var.Krasnodarskiy424EC25for survival atharvest250Sheppard et al.(1992)crop plantsand coniferPhaseolus vulgaris,Zeamays,Lactuca sativa,Lycopersicon esculentum,Brassica rapa,Pinus strobusEC25for pod yield inbeans300Sheppard et al. (2004)grass Elymus lanceolatus NOEC for emergence,length and biomass100059 S.C.Sheppard et al./J.Environ.Radioactivity79(2005)55–8360S.C.Sheppard et al./J.Environ.Radioactivity79(2005)55–83EC25of0.5mg U kgÿ1dry soil(all soil concentrations in this paper are expressed on a soil dry weight basis).This study meets the data acceptance criteria established,but remains problematic because0.5mg U kgÿ1soil is below background concentrations in many settings.Four other studies(Table1),Shevchenko et al.(1987),Gulati et al. (1980),Zhukov and Zudilkin(1971)and Bufatin et al.(1986),each have significant shortcomings.Methods,especially statistical treatment,are not well described.Sheppard et al.(1985)grew Scots pine in small pots with a range of U concentrations and included a mixed-contaminant soil with U.No effects relative to controls were observed at100mg added U kgÿ1dry soil.As a no-observed-effects-concentration (NOEC)value,this is not itself a definitive study but is supportive of other results.Sheppard et al.(1992)provided the next most sensitive study(Table1).Several different plant species,endpoints,soils and experiments were used,including large outdoor lysimeters and examination of second-generation effects in plants.Although the results are presented in a somewhat unorthodox manner,the study handily meets the acceptance criteria and has been cited by others for regulation development (Fleming et al.,2000;CCME,2002).One shortcoming is that the U concentration series used was an unbalanced design,with4replicates of the control,2–3replicates of treatments between10and100mg U kgÿ1soil,and no replicates at300and 1000mg U kgÿ1soil.The results clearly showed no effects below300mg U kgÿ1soil, but there was less precision to define exact effect levels above300mg U kgÿ1soil. Most of the data were not presented graphically,the EC25was between300and 1000mg U kgÿ1soil,and was500mg U kgÿ1soil for seedling emergence of several crop plants.The soils used by Sheppard et al.(1992)were stored moist,outdoors in covered, in-ground containers for over ten years,and were then used again for toxicity tests (Sheppard et al.,2004).The concentrations of U were re-measured and remained unchanged.In order to assure a toxic response,aliquots of the‘‘aged’’soil with 1000mg U kgÿ1soil were further spiked with U to achieve3000mg U kgÿ1.The plant bioassay,formally developed by Environment Canada,was a22-d bioassay with northern wheatgrass(Elymus lanceolatus),where emergence and shoot and root lengths and total plant biomass were measured.There was a statistically significant decrease in shoot and root length and in total plant biomass at3000mg U kgÿ1 compared to controls in a neutral pH loam,and a statistically significant decrease in root length at3000mg U kgÿ1compared to controls in afine sand with pH6.2. The solid/liquid partition coefficients,K d,for U in these soils were0.060and 0.29m3kgÿ1,respectively.The results,after aging the U in the soil for10years,show even less toxicity to plants than in the original study(Sheppard et al.,1992).The remaining studies of plant response to soil U(not shown in Table1,see Sheppard,2003),Meyer and McLendon(1997)and Meyer et al.(1998),showed EC25values above10000mg U kgÿ1soil,but are of marginal relevance because they used pulverised U metal from munitions that is relatively insoluble,or U as the mineral schoepite.They did note a hormesis-like response,with increased growth at 50mg U kgÿ1.Several other authors reported hormesis(e.g.,Gulati et al.,1980).In some cases,this could be because uranyl nitrate was used and the plant is responding to the nitrate(Gulati et al.,1980used uranyl acetate,which is less likely to cause aneffect as a result of the anion).There remains uncertainty in the ecotoxicology literature on how to interpret hormesis.In this analysis,it is assumed that positive response to U,regardless of mechanism,is not a regulatory concern and so only detrimental responses are considered.Because of the paucity of reliable studies on the toxicity of U in soil,it is relevant to consider the results of solution culture studies (Boileau et al.,1985;Cannon,1952;Murthy et al.,1984;Panda et al.,2001;Prister and Prister,1970;Weinberger and Shiva Shankara Murthy,1985).These must be interpreted with considerable caution as there is no proof that effects at a specific solution U concentration have any bearing on potential effects at the same U concentration in soil pore water.This is especially the case with U,where very marked accumulation of U in cell walls of the root epidermis is documented.In solution culture,it is possible that the U in solution is almost fully entrained on the root surface,resulting in a markedly different concentration at the uptake membranes compared to what was intended to be in the bulk solution.In general,solution culture results are expected to show toxicity at relatively low concentrations because of the absence of soil mediation.The studies listed above showed effect concentrations from 5to 200mg L ÿ1.The pH of the corresponding solutions was not consistently controlled (or reported),but in several it was about or above pH 7.Based on several reports of soil solid/liquid partition coefficients,a value of 4–30L kg ÿ1(geometric mean of 10L kg ÿ1)would correspond to the pH of the solution culture solutions.Thus,5–200mg L ÿ1in soil pore water might correspond to total soil concentrations of 50–2000mg U kg ÿ1dry soil.These are broadly consistent with the effect levels observed by Sheppard et al.(1992)in whole soils,and so support conclusions from that study.Solution culture data are not sufficiently relevant to indicate the possibility of ecologically relevant toxicity at concentrations lower than those observed in studies with soils,so the estimated effect at 50mg U kg ÿ1soil (based on an EC 25for tomato root growth)is considered only as guidance.Summarizing the studies above,the most reliable and relevant study (Sheppard et al.,1992)indicates an overall EC 25in the range of 300–500mg U kg ÿ1dry soil,and this is supported by the solution culture studies.Given that there are few studies,it is appropriate to apply a safety factor to these results.However,Sheppard et al.(1992)placed most emphasis on U concentrations in the range of 10–100mg U kg ÿ1,and on over 10soils and 6plant species there was no effect at these concentrations.Thus,any safety factor applied should not lower the PNEC to below 100mg U kg ÿ1.A value of 250mg U kg ÿ1dry soil is proposed as the PNEC to be protective of terrestrial plants.5.2.Toxicity to other soil biotaThis encompasses a large number of organisms,from microbes to animals,and is not fully independent of effects on plants.In an ecological setting,an effect on one soil-related organism may well affect several or all others,including plants.There have been few studies of effects of U on other soil biota,and the range of results (Table 2)reflects in part the range of types of biota.61S.C.Sheppard et al./J.Environ.Radioactivity 79(2005)55–83Cannon (1952)cited early work by Stoklasa and Penkava (1928)that showed effects on soil N-fixing microbes (Azotobacter )at 5mg U kg ÿ1dry soil.This work is difficult to confirm.In contrast,Sheppard et al.(1992)observed (qualitatively)no effect of 300mg U kg ÿ1on nodulation of legumes,another N-fixing process.Sheppard et al.(1992)also measured phosphatase enzyme activity in soils,following the logic that effects on nutrient cycling need to be examined and that the chemicalTable 2Toxicity information for soil biota other than plants,with effect concentrations as mg U kg ÿ1dry soil,except as notedAuthorsBiota Species Endpoint Effect concentration Cannon (1952)soil N-fixing microbes Azotobacter 5Sheppard et al.(2004)collembolan Onychiurus folsomi EC 20for survival and reproduction in fine sand,pH 6.292–190Sheppard et al.(1992)soil enzyme activity phosphatase enzyme activity EC 25for amount of p -nitrophenol released for phosphatase enzymes 100Meyer et al.(1998)soil microbes soil microbes EC 25for respiration,corresponding LOEC was 500mg/kg 160Sheppard andEvenden (1992)earth-worm Lumbricus sp.LOEC for implied survival 54d,soils !10%clay 300Sheppard et al.(2004)collembolan Folsomia candida EC 20for survival in fine sand,pH 6.2350Sheppard et al.(2004)collembolan Onychiurus folsomi EC 20for survival and reproduction in 2loam soils,pH 7.5O 390Sheppard et al.(2004)collembolan Folsomia candida EC 20for survival and reproduction in 2loam soils,pH 7.5O 710Sheppard et al.(1992)earthworm Lumbricus sp.EC 25for earthworm survival 1000Sheppard andEvenden (1992)earthworm Lumbricus sp.NOEC for implied survival 54d,soils O 10%clay 1000Sheppard et al.(2004)earthworm Eisenia andrei NOEC for survival and reproduction,3soils 1000Norberg andMolin (1983)bacteria Zoogloea ramigera ,a common aerobic organism found in sewage treatmentplantsprolonged lag phase and decreased growth rate,solution culture5mg L ÿ1Labrot et al.(1996)earthworm Eisenia fetida EC 25for biomarkers,contact test1m g cm ÿ2Labrot et al.(1999)earthworm Eisenia fetida LC 5096h in contact test13m g cm ÿ2Ribera et al.(1996)earthworm Eisenia fetida andrei LC 50contact test 40m g cm ÿ262S.C.Sheppard et al./J.Environ.Radioactivity 79(2005)55–83analogy of U and P might result in some effects of U on P cycling in soil.The EC 25was as low as 100mg U kg ÿ1soil in an acidic sand and 400mg U kg ÿ1soil averaged across 10soils.Although consistent with effect levels observed for plants,there remains difficulty in the ecological interpretation of effects involving a soil enzyme system.Meyer et al.(1998)observed an LOEC for soil respiration at 500mg U kg ÿ1soil,with a corresponding EC 25of 160mg U kg ÿ1soil.This was at a specific time in their soil respiration study,at other times the EC 25was 1600mg U kg ÿ1soil.For other endpoints related to decomposition rate,the EC 25values were 1600–7000mg U kg ÿ1soil.These results suggest that the EC 25of 160mg U kg ÿ1soil may not be representative of the overall effect levels in the study:a higher value was more common.Ecological relevance is also less certain because of both the form of U,pulverised munitions metals,and the nature of the endpoints.The next most sensitive was a series of studies.The earliest examined survival of earthworms (Sheppard et al.,1992),and in a sandy soil they observed an EC 25of 1000mg U kg ÿ1soil.This soil was originally an acid forest soil,as fine sand,that was limed to pH 6.2to accommodate earthworm and plant bioassays.There was no effect of U in another neutral pH sandy-loam soil at U concentrations up to 1000mg U kg ÿ1soil.These were large,mature earthworms,and those surviving in the sandy-loam soil were actively processing soil,so that exposure to the soil U was assured.As a long-term exposure study,this is a credible NOEC value with direct ecological relevance.In a related study using 11different soils,Sheppard and Evenden (1992)reported on worm/soil U concentration ratios (CRs)for worms exposed to the soils for 54d.They did not report survival statistics,but concentrations in worms imply survival and the CR values indicate effects on the uptake process.In the absence of much ecotoxicology literature for U,this study is considered a surrogate.The results suggest survival (NOEC)at 1000mg U kg ÿ1soil if the soil clay content is above 10–20%,and a possible lowest observed effect concentration (LOEC)of 100–300mg U kg ÿ1soil if the clay content is less.The results for the organic soil are not useful.The effects on linearity of uptake were similar U concentrations,with an LOEC of 300–1000mg U kg ÿ1soil when the clay content is above 10%and an LOEC as low as 50–100mg U kg ÿ1soil for soils with lower clay contents.The limitations of these data are:the ecological relevance of the observations about linear uptake is not obvious,and the implied survival information is binary (yes/no)data rather than numerical scores.The soils used by Sheppard et al.(1992),including a loam soil not previously used in earthworm bioassays,were ‘‘aged’’moist and used again for toxicity tests (as described above for the plant bioassays and in Sheppard et al.,2004).Bioassays,formally developed by Environment Canada,for three different soil organisms were completed:an acute (14-d)survival and a chronic (56-d)reproduction bioassay with the earthworm Eisenia andrei ;a 28-d reproduction test with even-aged collembolan,Folsomia candida ,from a synchronous culture;and a 35-d reproduction test with mature adults of collembolan Onychiurus folsomi .There was no effect of 1000mg U kg ÿ1soil compared to controls on earthworm survival in any of the63S.C.Sheppard et al./J.Environ.Radioactivity 79(2005)55–8364S.C.Sheppard et al./J.Environ.Radioactivity79(2005)55–83soils,and no effect of1000mg U kgÿ1on reproduction in two of the soils.There was no reproduction in any treatment in the third soil.Clearly,E.andrei was not particularly sensitive to U.For F.candida,an EC20of350mg U kgÿ1soil for survival was observed in the limedfine sand with pH6.2,whereas reproduction was less sensitive in that soil and both survival and reproduction were less sensitive in the other soils(EC20from710to O3000mg U kgÿ1).Results for O.folsomi showed similar EC20in the neutral pH loam andfine sandy loam,from390to O1000mg U kgÿ1.Results for O.folsomi,a root-eating springtail,showed much greater sensitivity in the limedfine sand soil.In two different series of trials,survival and reproduction of this species in this soil was significantly affected.The EC20values were as low as 92mg U kgÿ1.There is no obvious explanation why this organism was especially sensitive.The soil was an acidic forest sand that had been limed10years earlier,and may not be typical habitat for O.folsomi.The remaining tests with earthworms are based on the OECD contact test,where worms are put into contact with moist,U-treatedfilter paper for a short interval and reported as m g U cmÿ2paper.Depending on the specific methods used,there is about 0.02cm3water cmÿ2paper.This test has some relevance for lipophilic organic compounds,especially pesticides,but is of severely limited value for inorganic con-taminants where effects are typically much slower and require some accumulation time in the tissues.The only reported sub-lethal effect level(Labrot et al.,1996)was for biomarkers at1m g cmÿ2,which assuming the above moisture content would (very crudely)coincide with50mg Lÿing a soil solid/liquid partition coefficient, K d,of10L kgÿ1,this would imply a soil concentration of500mg U kgÿ1soil.This is a highly speculative extrapolation,but somewhat supports the EC25from Sheppard et al.(1992).The study of Norberg and Molin(1983)is also somewhat removed from soil studies;they reported an EC25of5mg Lÿ1soil for Zoogloea ramigera,a common aerobic organism found in sewage treatment plants.The effect was a lag in growth in agar when U was present at pH7,and unfortunately did not change with U ing the5mg Lÿ1and the soil solid/liquid partition coefficient used earlier for this pH range(10L kgÿ1),a corresponding concentration in soil might be50mg U kgÿ1soil.This is lower than reported in the more relevant studies, and is not considered a definitive value.Summarizing the studies above,there is some indication from the phosphatase enzyme study of Sheppard et al.(1992),the collembolan test of Sheppard et al. (2004)and the soil respiration study of Meyer et al.(1998)that certain soil processes might be more sensitive to U than corresponding plant responses.Indeed,all3 studies included plants in their overall programs,and all reported some incidents of greater sensitivity in non-plant endpoints.However,the data are few,and the ecological relevance of the microbial endpoints is uncertain.The earthworm survival test of Sheppard et al.(1992)is more ecologically relevant and showed less sensitivity than corresponding plant bioassays.As a result,a PNEC of100mg U kgÿ1dry soil is proposed,to be lower than the PNEC for plants,but recognizing that there is a large diversity among the soil processes involved.。