Environmental risk assessment of alkylphenols from offshore produced water on fish reproduction

at SciVerse ScienceDirect

Marine Environmental Research 75 (2012) 2e9

Contents lists available

Marine Environmental Research

journal homepage: www.elsevier .com/locate /marenvrev

Environmental risk assessment of alkylphenols from offshore produced wateron fish reproduction

Jonny Beyer a,b,*, Lars Petter Myhre a,1, Rolf C. Sundt a, Sonnich Meier c, Knut-Erik Tollefsen d, Rune Vabø c,Jarle Klungsøyr c, Steinar Sanni a,b

a IRIS - International Research Institute of Stavanger, Mekjarvik 12, N-4070 Randaberg, NorwaybUniversity of Stavanger, Department of Mathematics and Natural Science, N-4036 Stavanger, Norwayc Institute of Marine Research, P.O. Box 1870, Nordnes, N-5817 Bergen, NorwaydNIVA - Norwegian Institute for Water Research, Gaustadalléen 21, NO-0349 Oslo, Norway

Keywords:Fish reproductionAlkylphenolsProduced waterEnvironmental risk assessment

* Corresponding author. IRIS - International ReseBiomiljø, Mekjarvik 12, N-4070 Randaberg, Norway. T

E-mail address: [email protected] (J. Beyer).1 Current address: Statoil, Stavanger, Norway.

0141-1136/$ e see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.marenvres.2011.11.011

a b s t r a c t

Concern has been raised over whether environmental release of alkylphenols (AP) in produced water(PW) discharges from the offshore oil industry could impose a risk to the reproduction of fish stocks inthe North Sea. An environmental risk assessment (ERA) was performed to determine if environmentalexposure to PW APs in North Sea fish populations is likely to be high enough to give effects on repro-duction endpoints. The DREAM (Dose related Risk and Effect Assessment Model) software was used inthe study and the inputs to the ERA model included PW discharge data, fate information of PW plumes,fish distribution information, as well as uptake and elimination information of PW APs. Toxicodynamicdata from effect studies with Atlantic cod (Gadus morhua) exposed to APs were used to establisha conservative environmental risk threshold value for AP concentration in seawater. By using the DREAMsoftware to 1) identify the areas of highest potential risk and 2) integrate fish movement and uptake/elimination rates of APs for the chosen areas we found that the environmental exposure of fish to APsfrom PW is most likely too low to affect reproduction in wild populations of fish in the North Sea. Theimplications related to risk management of offshore PW and uncertainties in the risk assessment per-formed are discussed.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Produced water (PW) is a major waste water stream fromoffshore oil and gas platforms during hydrocarbon production. Thishighly complex effluent contains dispersed oil and dissolvedorganic compounds, including aromatic hydrocarbons, organicacids, phenols, inorganic compounds as well as traces of chemicalsadded during the production process (Jacobs et al., 1992; Neff et al.,1992; Strømgren et al., 1995). Many PW compounds have beenidentified as aquatic toxicants and the presence of alkylphenols(APs) in these effluents has been considered to represent a possiblerisk factor to successful reproduction of economically importantfish populations in the North Sea. The typical AP composition in PWis dominated by the less alkylated (C1eC3) APs, whereas the higher

arch Institute of Stavanger,el.: þ47 51875507.

ll rights reserved.

molecular weight APs with branched alkyl-chains in the paraposition are only present at trace levels (Boitsov et al., 2007;Harman et al., 2009). The concern associated with PW APs haspredominantly been related to possible estrogenic properties of theAPs, involving potential disruption of estrogen receptor (ER)-mediated processes important for normal growth, developmentand reproduction (White et al., 1994; Gimeno et al., 1997, 1998; Luet al., 2010). A range of APs have been reported to bind to andactivate fish ERs in vitro, and mono-substituted APs with moder-ately sized branched alkyl groups in the para position have beenidentified to have the highest estrogenic potency (Tollefsen et al.,2008; Tollefsen and Nilsen, 2008). Although the estrogenicactivity of APs has been extensively studied, these compoundsdisplay multiple endocrine modes of action (MoA) by interferingwith normal function of the androgen receptor (AR), plasmasex-steroid hormone binding proteins and modulating thebiotransformation and circulatory concentration of natural sex-steroid hormones (Arukwe et al., 1997; Satoh et al., 2005;Tollefsen, 2007). Proteomic based biomarker candidate discoverystudies have indicated that PW and PW APs exposure may induce

Fig. 1. An environmental risk assessment (ERA) study is normally divided into the general steps as shown in this figure. In some ERA versions the hazard identification and hazardcharacterisation parts are combined into one step. The first part of the total ERA (step 1e4) is often termed the Risk Analysis part.

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e9 3

impairment of general growth and development processes of fishfry (Bohne-Kjersem et al., 2009, 2010). As the discharged volumesof PW are large, between 400 and 500 million tons and includingseveral hundred tons of APs per year in the North Sea region (Utvik,1999; OSPARCOM, 2009), a thorough Environmental Risk Assess-ment (ERA) is required in order to identify possible needs formitigation measures.

In an ERA study of a certain chemical stressor, or mixture ofstressors, one normally combines relevant exposure data withrelevant dose-response toxicity information (a brief intro to ERA isprovided below). In risk assessment of PWAPs it is important to beable to evaluate the problem with exposure scenarios that arerealistic for an offshore recipient situation. In addition, the seasonaldistribution of fish populations within the influenced region isa factor of great importance when actual exposure levels ofcontaminants originating from PW outfalls are estimated. Acombination of fish population data and PW plume fate data in ERAmodels may thus enable a prediction of exposure situations and anassessment of actual ecological risks. In this connection, biologicaleffects that are demonstrated to occur at very low exposureconcentrations will be particularly relevant. Hence, several of therecent studies that have addressed PW effects have focussed onnovel effect endpoints related to timing of puberty, gamete devel-opment, timing of spawning and other relevant parts of thereproduction process within a marine fish population (Meier et al.,2007a,b, 2008, 2010, 2011).

In the current work, a qualitative and quantitative ERA exerciseaddressing the environmental release of APs in offshore PWand theneed for protecting the reproduction in offshore fish populationswas performed. The data used in the ERA included known PWdischarge data, toxicokinetic properties of APs, in vivo doseresponse information from reproduction effect studies with APs inAtlantic cod, as well as North Sea fish stock data of Atlantic cod(Gadus morhua), saithe (Pollachius virens) and haddock (Melanog-rammus aeglefinus). The DREAM (Dose related Risk and Effect

Assessment Model) software (Rye et al., 1996; Neff et al., 2006) wasused for integrating the available information and to identify theprobability of risk for an exposure level of PWAPs that may lead todisturbance of reproduction in field populations of fish.

2. Environmental risk assessment (ERA) method description

2.1. ERA in brief

Environmental Risk Assessment can be defined as a stepwiseprocedure to estimate the adverse effects of an environmentalstressor (often a toxic chemical or pollution mixture) on anecosystem or its components with a known degree of certainty(Depledge and Fossi, 1994; van der Oost et al., 2003; van Leeuwenand Vermeire, 2007). Use of ERA has become increasingly impor-tant in studies of the effects of environmental pollutants. An ERAstudy is normally divided into a scientifically oriented ‘risk analysis’part and a more politically oriented ‘risk management’ part. Therisk analysis part of ERA normally comprises four main elements:(1) a problem definition/hazard identification process, (2) an effectassessment (hazard characterisation, assessment of predicted noeffect concentration (PNEC)), (3) an exposure assessment, by use ofa predicted environmental concentration (PEC) approach and (4)a risk characterisation analysis (Fig. 1). The ‘risk characterisation’includes typically a combined evaluation of the incidence andseverity of the hazardous effects, and the outcome of the riskcharacterisation should serve as a basis for decision making and forthe problem management processes.

2.2. Problem definition/hazard identification

In the current study we performed an ERA of APs in offshore PWin relation to possible effects on the reproduction of fish pop-ulations in the North Sea. The defined “study problem” (or stressor)was: “alkylphenols in offshore PW effluents with alkylation containing

Table 1Discharge volumes and concentrations of alkylphenols divided into two groups for the different offshore installations which are used in the current risk assessment.

2002 2006

PW discharge (m3/day) AP C4eC5 conc (mg/l) AP C6þ conc (mg/l) PW discharge (m3/day) AP C4eC5 conc (mg/l) AP C6þ conc (mg/l)

Tampen regionStatfjord A 34,781 0.11954 0.00132 35,130 0.07170 0.00053Statfjord B 36,277 0.12327 0.00190 27,915 0.07396 0.00076Statfjord C 44,538 0.11980 0.00070 61,996 0.07190 0.00028Gullfaks A 18,185 0.09200 0.00114 18,055 0.05528 0.00046Gullfaks B 33,713 0.07000 0.00070 32,027 0.07000 0.00035Gullfaks C 18,940 0.10500 0.00083 26,082 0.06292 0.00033Veslefrikk 12,852 0.12300 0.00500 20,000 0.07396 0.00214Huldra 521 0.12320 0.00540 1649 0.12320 0.00540Troll A 56 0.06800 0.00125 58 0.06800 0.00125Oseberg F 3800 0.14612 0.00259 1177.5 0.14612 0.00259Oseberg C 3482 0.10110 0.00176 5280 0.10110 0.00176Oseberg Øst 429.9 0.11459 0.01458 986.3 0.11459 0.01458Oseberg Sør 9.3 0.12400 0.00629 690.4 0.12400 0.00629Troll B 21,113 0.11780 0.00260 22,958.9 0.11780 0.00260Troll C 27,437 0.10000 0.00200 23,013.7 0.04000 0.00030Snorre TLP 18,956 0.14053 0.00330 29,742 0.14053 0.00330Snorre B 1027 0.16078 0.00160 271 0.16078 0.00160Brage 2760 0.09597 0.00134 1583.5 0.09597 0.00134Ekofisk regionEkofisk K 4878 0.0119 0.0008 9534 0.0012 0.0001Ekofisk J 8800 0.0213 0.0014 20,546 0.0021 0.0001Ekofisk J sentrif 1360 0.0197 0.0014 Included in Eko JEldfisk B 460 0.0145 0.0011 1764 0.0087 0.0005Eldfisk FTP 942 0.0113 0.0009 1880 0.0068 0.0005Ula 1861 0.0040 0.0009 1475 0.0040 0.0009Gyda 3393 0.0043 0.0004 0Tor 1033 0.0028 0.0010 938 0.0028 0.0010Valhall 887 0.0163 0.0023 216 0.0163 0.0023Sleipner regionBalder 2242 0.00766 0.00024 0 0.00766 0.00024Jotun 7750 0.01843 0.00057 17,400 0.01843 0.00057Sleipner A 350 0.12690 0.00119 858 0.12690 0.00119Sleipner T 230 0.29231 0.00152 1014 0.29231 0.00152Varg 10 0.00437 0.00014 10 0.00437 0.00014Heimdal 0 0.26760 0.00008 1.6 0.26760 0.00008

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e94

four carbons or more”. The “hazard” addressed was identified as:“any known negative effect on reproduction endpoints in fish”. A riskthreshold value of AP exposure was established based on severalreproduction-relevant effect studies with Atlantic cod exposed toAPs (see hazard characterisation section). The target organisms ofthe ERA study were populations in the North Sea of Atlantic cod,saithe and haddock; all three are commercially important fishspecies.

2.3. Hazard characterisation of PW APs

A direct biological effect of PWAPs in an offshore fish populationwill imply that the seawater in which the fish lives contains APsfrom PW at a certain concentration level. There is a shortage ofempirical data about AP concentrations in seawater down-currentfrom PW outfalls. Model estimates are most often used and theseare based on analyses of the AP composition in PW streams at thepoint of recipient entry and a prediction of the fate influence ofdifferent recipient factors on the dilution of the PW plume. Modelstudies published earlier by Rye et al. (1996) indicated that a worstcase in vivo exposure scenario in fish living down-current from PWoutfalls was a PW AP body burden (BB) level of approximately10 mg/kg fish.

An essential point in the hazard characterisation of PWAPs is toestablish NOEC and LOEC data, i.e. the highest stressor concentra-tion with no effect and the lowest stressor concentration givinga statistically significant effect in the organisms of concern,respectively. Several dose-response experiments conducted by

Meier et al. with Atlantic cod exposed via contaminated feed toamixture of four C4eC7 APs (4-tert-butylphenol, 4-n-pentylphenol,4-n-hexylphenol and 4-n-heptylphenol) have revealed reproduc-tion-relevant effects; such as reduction in the amount of sperma-tozoa in male fish and delayed gonadal maturation in females andmales at a nominal body burden dose of 20 mg AP/kg (Meier et al.,2007a,b) and alteration of onset of puberty in females at a nominaldose of 4 mg AP/kg (Meier et al., 2011). Several other effects havebeen observed at higher exposure doses; such as impaired oocytedevelopment, reduced estrogen levels and a 17e28 days delay inestimated time of spawning in female fish, whereas in malesa reduction of 11-keto-testosterone concentrations, induction ofvitellogenin levels, impaired testicular development and anincrease in the amount of spermatogoniawere observed (ibid.). TheAtlantic cod in the studies of Meier et al. (2011) were fed withcontaminated feed pastes containing the four APs for up to 20weeks. The observed LOEC in these studies was at a nominal dose of4 mg AP mix/kg bodyweight, and the observed NOEC level was oneorder of magnitude lower (0.4 mg AP mix/kg bodyweight). Meieret al. estimated that the LOEC of 4 mg AP mix/kg corresponds toa theoretical water concentration of approximately 8 ng/l whenusing an average bioconcentration factor (BCF) of 500 (Meier et al.,2011). However, since the LOEC estimations were based on nominalexposure doses given by oral administration and since theabsorption efficiency of the APs over the gut wall in fish is relativelylow, typically about 10% (Pickford et al., 2003; Sundt et al., 2009), itis possible that the real body burdens (and hence the LOEC andNOEC) were one order of magnitude lower than estimated. In that

Fig. 2. Examples of fish population distribution maps in the North Sea for: (A) Atlanticcod in third quarter 2000; (B) saithe in first quarter 2000 and (C) haddock in firstquarter 2003. Fish population data obtained from ICES - the International BottomTrawl Survey (IBTS) database. The regions with highest fish densities are shown in redcolour. (For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e9 5

case, the real body burden LOEC for these C4eC7 APs in Atlantic codcould theoretically be as low as 0.4 mg AP/kg, corresponding toa water LOEC value of 0.8 ng/l and a NOEC of 0.08 ng/l, again byemploying a mean BCF of 500. On the basis of a review of these andother available studies about the issue of APs in PWwe chose to use40 ng/l and 4 ng/l for C4eC5 APs þ C6þ APs in seawater as the riskthreshold values in the present risk assessment study; this two-value approach was used in order to allow for both a realistic anda conservative risk estimate.

2.4. Exposure prediction of PW APs in offshore fish populations

Knowledge of the concentration of PWAPs that fish are exposedto in the field is instrumental for predicting the risk for adverseeffects. In this study, the reported discharge volumes of PW (and APconstituents) from the years 2002 and 2006 from major oilproducing platforms in the Norwegian sector (Table 1) were used tosimulate the environmental concentrations of APs in differentregions of the North Sea. The AP exposure simulations were doneusing the DREAM software (Rye et al., 1996; Neff et al., 2006). TheAP discharge data was sub-divided into two groups (C4 þ C5 andC6þ) APs. Relevant factors related both to the physical characteris-tics of the recipient waters (e.g. sea current and wind conditions,etc), as well as abundance and distribution of selected fish pop-ulations were used as inputs to the exposure simulations. Tox-icokinetic properties of APs (uptake and elimination constants,bioconcentration factors) had been investigated in separate studies(Tollefsen et al., 1998; Sundt et al., 2009) and mean values of thesecharacteristics were used for the prediction of body burden of APsin fish.

Fish abundance indexes for the different study regions wereestimated by means of fish stock distribution data for Atlanticcod, saithe and haddock obtained from the International bottomtrawl survey’s (IBTS) database (from the years 1999e2003) whichwere available from the International Council for the Explorationof the Sea (ICES) (Fig. 2). A detailed evaluation of the fish distri-bution pattern was made for the three study areas “Tampen”,“Ekofisk” and “Sleipner”. The contribution of the fish stocks inthese regions to the total North Sea stocks is shown in Table 2.The fish distribution data was organized into two-dimensionalgrids and each grid cell was assigned with an abundance indexvalue indicating the relative fish abundance within that specificarea. The DREAM model has the capability to simulate a three-dimensional semi-randomised movement of so-called “fishparticles” within a concentration grid. A “fish particle” can in thisconnection be defined as a small shoal of fish, but the term “fishparticle” is simply a technical term for how fish are representedduring the DREAM simulation. In this study, we used 10,000 fishparticles within each exposure simulation. Each “fish particle”was characterised by a set of variables (horizontal and verticalswimming speeds, depth range, etc.) which the DREAM modeluses for simulating semi-randomised temporal and spatialmovements during the simulation period (details not shown). Theintegration time unit (time-step) in the exposure simulation wasset to 20 min and the duration of each simulation period in eachstudy region was 30 days. Due to the semi-randomisation thedifferent fish particles within the simulation were exposed todifferent levels of (PW AP) contaminants.

Toxicokinetic data is key information in the assessment ofbioaccumulation (bioconcentration) and biotransformationpotential of chemicals. Alkylphenols, which have been reported toaccumulate fairly rapidly and depurate efficiently, have a bio-concentration potential ranging over several orders of magnitude(McLeese et al., 1981; Tollefsen et al., 1998). In the presentmodelling study, the most important properties of the selected

alkylphenols were taken into account by including mean kineticrate constants for uptake (k1) and elimination (k2) and bio-concentration factors corresponding to the selected C4eC5 APsand C6þ APs (Table 3).

Table 2Mean values of fish distribution of Atlantic cod, saithe and haddock from threeselected study areas in the North Sea used as input for the exposure and effectsimulation in the current ERA study. The total area of the North Sea consideredcorresponds to about 1,138,908 km2.

Tampen Ekofisk Sleipner

Area, km2 10,260 11,313 24,511% of total area 0.9 0.99 2.15Cod, % of total population 1.70 1.50 3.60Saithe, % of total population 12.30 0.04 4.80Haddock, % of total population 1.90 1.20 9.90

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e96

2.5. Risk characterisation

The DREAM simulation was performed in a two-step approach,with each step differing in complexity. The first stepwas performedas a simplified approach to provide an initial assessment of thepotential risk. In this initial screening, identifications of the totalseawater volumes in which the area-specific PEC/NOEC ratio indi-cated a positive risk situation (e.g. PEC/NOEC > 1) were performed.This was done to identify the “worst case” regions for subsequentDREAM analysis. Standard data-files for sea current conditions(“May 90.DIR”) and wind conditions (“Gullfaks.wnd” for theTampen area and “Ekofisk.wnd” at Ekofisk and Sleipner) were usedfor simplicity reasons. Fish stock distribution data that weresimplified to a certain degree and PEC/NOEC calculations based onearlier LOEC and NOEC studies with C4eC7 APs (Rye et al., 1996)were used as model inputs.

In a second and more advanced mode of the DREAM simulation,the Tampen area was studied in greater detail since this regionwasfound to be the “worst case” in the screening step. In this secondstep simulation the dispersal and dilution fate of the alkylphenolsin the sea and data of fate properties of APs (e.g. uptake andelimination constants, biodegradation rates) were taken intoaccount, thus providing a more realistic exposure prediction andrisk estimate. From the hazard characterisation part (see earliersection) minimum and maximum risk threshold values for PWAPsbody burden of 2 mg/kg 20 mg/kg had been established. These BBvalues represented a C4eC5 þ C6þ AP concentration in seawater of4 ng/l and 40 ng/l, respectively; given a mean BCF of 500 for the APtarget compounds. Hence, if “fish particles” at any time-pointduring the simulation were predicted to reach an exposure levelof >4 ng/l, and consequently a theoretical AP body burden of 2 mg/kg, they were considered to be at risk according to the mostconservative risk threshold level.

3. Results

The oil producing platforms that are located in the Tampenregion are characterised by considerably higher PW dischargevolumes than the platforms located in the Ekofisk and Sleipnerregions (Table 1). This relative difference between the differentregions in the current study was also illustrated by the resultsobtained in the PEC/NOEC screening runs, which showed that theTampen area came out much higher than the Sleipner and Ekofiskregions with regard to potentially affected seawater volumes and

Table 3Estimated average values for uptake (k1) and elimination (k2) rate constants andbioconcentration factors (BCF) of alkylphenols for use in the simulations. See text forreference to uptake and elimination studies.

Compound k1 k2 BCF

AP C4eC5 200 1 200AP C6þ 680 1.5 453

the size of risk areas (Table 4). Based on these results it was decidedthat the more advanced DREAM simulations were performed onlyfor the Tampen region, since this clearly represented the “worstcase” exposure scenario.

The results of the more advanced DREAM simulation for theTampen region with PW discharge data from 2002 and 2006 areshown in Table 5. When the DREAM simulation in this case reportsa “0” value it means that none of the 10,000 fish particles includedactually reached the risk threshold level. This “zero prediction”situation was the result in all simulations which used the AP bodyburden of 2 mg/kg as risk threshold value; and similar alsowhen therisk threshold was lowered to 0.2 mg/kg. It was only when the riskthreshold value in the simulation was lowered two orders ofmagnitude (i.e. to 0.02 mg/kg) that a relatively small number of fishparticles within the Tampen area reached a level that indicateda risk (e.g. 16 fish particles out of 10,000 in the Statfjord/Gullfakssimulation based on the 2002 discharge data as the highest)(Table 5). No risk scenarios were identified for the Oseberg/Trollarea, even when reducing the risk threshold two orders of magni-tude. In Fig. 3, the temporal exposure and body burden predictionresults of the top 10 fish particles from the simulation of the “worstcase” Statfjord/Gullfaks 2002 are shown. The results show thatnone of the 10,000 fish particles were predicted to reach a bodyburden of 0.1 mg/kg at any timewithin thewhole 30 days timeframeof the simulation.

4. Discussion

According to the environmental regulation of offshore oil andgas industry activities in Norway, it is generally not permitted torelease chemicals that lead to measurable endocrine disruption orreproduction disturbance in organisms which are present in therecipient ecosystem (see Petroleum Safety Authority of Norwaywebsite for regulation documents). The presence of high concen-trations of different AP isomers and findings of estrogenic activityin PW from offshore oil production platforms in the Norwegian andUK sector of the North Sea (Boitsov et al., 2004, 2007; Thomas et al.,2004; Tollefsen et al., 2007), have suggested a possible risk forendocrine disruption in fish living inwaters receiving PWeffluents.However, other investigators have argued that the risk of negativeeffects on fish populations by AP exposure is likely to be very low(Gray, 2002). In this connection, it is important to differentiatebetween “perceived” and “real” risk of PW discharges, in particularthis is essential for the development of an environmentallyappropriate but also cost effective risk management of PWeffluents.

The overall results of DREAM simulations presented in thisarticle imply that fish populations living in offshore waters close tooil production platforms are unlikely to be exposed to PW APs (i.e.the sum of C4eC5 APs and C6þ APs) at the concentrations requiredfor causing adverse effects on fish reproduction. In our study, weincluded the major oil producing regions of Tampen, Ekofisk andSleipner. The most detailed risk assessment was performed only forthe Tampen area, since this area had considerably higher dischargevolumes of PW than the Ekofisk and Sleipner regions in the initialscreening step of the study. In this first step, PEC grids of APs inseawater were established and these were subsequently comparedwith PNEC grids which were made based on two alternative riskthreshold values, i.e. 40 and 4 ng/l APs in seawater. When wesubsequently compared the resulting risk areas with fish distribu-tion data for Cod, Saithe and Haddock we could estimate prelimi-nary risk potential for the regional sub-populations of these threefish species. For the second step of the risk assessment we intro-duced the three-dimensional and semi-random movement asa factor for the 10,000 “fish particles” which were included in the

Table 4Sum of results from the PEC/NOEC screening simulations showing the obtained Environmental Impact Factors and risk areas (as km2) by study regions using both 40 ng/l and4 ng/l of PW APs in seawater as risk threshold values.

Region Environmental impact factor Risk area, time-step with largest area, km2

2002 2006 2002 2006

40 ng/l 4 ng/l 40 ng/l 4 ng/l 40 ng/l 4 ng/l 40 ng/l 4 ng/l

Tampen 8.4 947.9 8.39 662.7 x x x xStatfjord/Gullfaks 10.5 572.5 2.1 314.6 0.084 4.788 0.021 2.415Oseberg/Troll 0 195.5 0 134.4 0 1.734 0 1.196Snorre 2.41 68.3 3.21 131 0.0243 0.632 0.032 1.34Sum Tampen area 12.91 836.3 5.31 580 0.1083 7.154 0.053 4.951Sleipner 0 0 0 0 x x x xSleipner zoom 0 0 0 3.254 0 0 0 0.03Jotun Balder 0 1.03 0 5.127 0 0.01 0 0.05Sum Sleipner area 0 1.03 0 8.381 0 0.01 0 0.08Ekofisk 0 0 0 0 0 0 0 0Ekofisk zoom 0 0

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e9 7

simulation of risk scenarios at Tampen. The advanced simulationprovided a more field-realistic estimate of the potential exposureregime, as it is unlikely that wild fish at an offshore location willstay at the same position within a concentration gradient for longperiods of time. The results of this more advanced exposuresimulation showed that none of the “fish particles” during thesimulations of the Tampen area reached a positive risk level whenthe decided risk threshold value of 2 mg/kg APs on basis of bodyburden was used.

The results of the simulation agree with observation data fromfield surveys. Field studies using passive samplers have founddetectable amounts of short chain APs (up to 192 ng/l forthe

PC1eC3 APs) in seawater around different Norwegian oil

platforms, but longer chain length APs were only present in verylow concentration (<ng/l) (Harman et al., 2009). This has beencorroborated by studies in fish caged close to PW outfalls. Themeasurement of bile metabolites of short chain APs (C1eC3 APs),a sensitive exposure marker for PWAPs (Beyer et al., 2011), showedan exposure clearly above background levels in cod which werecaged for 6 weeks close to a PW outfall at a North Sea oil platform,but no long chain APs were detected (Sundt et al., 2011). TheInstitute of Marine Research have both in 2002 and 2005 collectedwild fish from the Tampen area for AP analysis and did not finddetectable levels (>1 mg/kg) of APs in cod and haddock liver nor inherring muscle (Grøsvik et al., 2007).

Several simplifications and assumptions were made during therisk assessment process performed in this study. A low resolutionin the fish stock data in the close proximity of the oil productionplatforms led to the assumption of an even distribution of fish inthese areas. This is not necessarily a valid assumption as higherdensities of fish are known to occur in connection with platformsand other underwater installations in the North Sea (Jørgensenet al., 2002; Soldal et al., 2002). The most intense dilution ofPW plumes will occur in relatively close proximity to the plat-forms (Washburn et al., 1999), and this will generate mixing zoneswith an AP concentration level higher than the threshold of 4 ng/l.

Table 5Numbers of fish particles (out of a total of 10,000) in the different oil producing regions wand 0.02 mg/kg, respectively.

Region # fish particles body burden > 2 mg/kg # fish particles

2002 2006 2002

Statfjord/Gullfaks 0 0 0Snorre 0 0 0Oseberg/Troll 0 0 0

Another relevant aspect is whether other effect endpoints rele-vant to fish reproduction and endocrine disruption could haveresulted in different risk predictions. Studies have suggested thatPW contains compounds that may interfere with other endocrinefunctions such as endogenous steroid transport, steroid metabo-lism and androgen receptor-mediated processes and thuscontribute to the endocrine disrupting potential of PW (Arukweet al., 1997; Satoh et al., 2005; Tollefsen et al., 2006; Tollefsen,2007). Findings that only approximately 35% of the total estro-genicity in PW can be attributed to APs introduce the possibilitythat other compounds may also contribute significantly to theendocrine disrupting potential of PW (Thomas et al., 2009).Although the identity of these compounds has not been fullyrevealed yet, use of high resolution analytical methods has iden-tified certain PAHs and novel classes of compounds in oil-relatedeffluents such as naphthenic acids as being potentially endo-crinologically active (Thomas et al., 2009; Rowland et al., 2011). Asthese compounds are expected to act in combination with otherendocrine disruptors to produce combined toxicity as seen forxenoestrogens elsewhere (Petersen and Tollefsen, 2011), thresh-olds for the effects of individual compounds or small groups ofcompounds could potentially be lower than those proposed here.Other recent studies have shown that effects relevant to thesexual development process and the timing of the spawningprocess in fish populations may represent relevant endpoints instudies on PW impact (Meier et al., 2007a). However, use ofconservative estimates throughout the ERA and identification ofrisk thresholds being two orders of magnitude lower than thatobtained under the exposure scenarios studied indicate that thelikelihood of reproductive effects in wild populations of fish by APexposure is low.

It is a complex task to address whether the environmentalrelease of PW to the North Sea has negative implications foroffshore fish populations and their reproduction. But by using theprinciples of ERA as a foundation for hazard and risk assessment,this study has shown that it’s possible to discriminate better

hich the DREAM simulation indicated will accumulate a body burden of APs of 2, 0.2

body burden > 0.2 mg/kg # fish particles body burden > 0.02 mg/kg

2006 2002 2006

0 16 10 4 60 0 0

Top ten fish particles Statfjord gullfaks 2002

0.00

0.02

0.04

0.06

0.08

0.10

0.0 5.0 10.0 15.0 20.0 25.0 30.0timestep

bo

dyb

urd

en

p

pb

.

1951374038284535524659416968698089749545

Fig. 3. The ten highest predictions of accumulated body burden of alkylphenols pr fishparticle based on DREAM simulations of the PW discharges released from platforms atthe Statfjord and Gullfaks fields (Tampen region). The individual time-steps for inte-grating exposure levels were set to 20 min and the total simulation period was 30 days.

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e98

between “perceived risk” and the “real risk” scenarios. Thedistinction of these two types of risk information is crucial forimplementation of best possible environmental practice, manage-ment and regulation of PW discharges (Gray, 2002). This studyrepresents a type of risk assessment that does not in detail followa standardized procedure, but where the best available tools areused for simulating the temporal and spatial exposure of fishpopulations to APs originating from PW discharges. The data fromexposure simulations using the DREAM software were used incombination within a general ERA scheme to provide a conceptualframework for “weight of evidence” approaches that may assist theenvironmental management of PW discharges.

5. Conclusion

An environmental risk assessment study was performedaddressing the possible effect of offshore PWalkylphenols (APs) onreproduction in North Sea fish populations, using the DREAMsoftware. Inputs to the model included discharge data of APs frommajor offshore oil production platforms, an effect thresholdconcentration of C4eC5 APs and C6þ APs (body burden of 2 mg/kgand predicted NOEC of 4 ng/l), predictions of environmentalconcentrations of APs in seawater and fish stock resource data forthe different study regions and the North Sea as awhole. Themodelsimulations indicated that, evenwhen usingmore environmentallyconservative input data, the predicted exposure of fish in the mostimpacted region (Tampen) was approximately two orders ofmagnitude lower than the exposure level required for causingnegative effects on fish reproduction. Future studies shouldaddress some of the uncertainties identified in this study toprovide better risk estimates for the whole effluent toxicity inorder to safeguard the reproductive health of North Sea fishpopulations.

Acknowledgements

This work has been financed by the oil companies with PWdischarges in Norwegian waters. The authors thank Mathijs Smit(TNO-MEP, current affiliation Statoil), Odd Gunnar Brakstad, MarkReed, and Marinela Gerea (SINTEF), and representatives fromStatoil, Hydro, ConnocoPhillips, BP, Esso and Talisman for theirsignificant contributions to the study. Thanks to Emily Lyng (IRIS)for a final language vetting of this manuscript.

References

Arukwe, A., Förlin, L., Goksøyr, A., 1997. Xenobiotic and steroid biotransformationenzymes in Atlantic salmon (Salmo salar) liver treated with an estrogenic

compound, 4-nonylphenol. Environmental Toxicology and Chemistry 16 (12),2576e2583.

Beyer, J., Sundt, R., Sanni, S., Sydnes, M., Jonsson, G., 2011. Alkylphenol metabolitesin fish bile as biomarkers of exposure to offshore oil industry produced water inferal fish. Journal of Toxicology and Environmental Health - Part A - CurrentIssues 74 (7e9), 569e581.

Bohne-Kjersem, A., Bache, N., Meier, S., Nyhammer, G., Roepstorff, P., Sæle, O.,Goksøyr, A., Grøsvik, B.E., 2010. Biomarker candidate discovery in Atlantic cod(Gadus morhua) continuously exposed to North Sea produced water from egg tofry. Aquatic Toxicology 96 (4), 280e289.

Bohne-Kjersem, A., Skadsheim, A., Goksøyr, A., Grøsvik, B.E., 2009. Candidatebiomarker discovery in plasma of juvenile cod (Gadus morhua) exposed tocrude North Sea oil, alkyl phenols and polycyclic aromatic hydrocarbons (PAHs).Marine Environmental Research 68 (5), 268e277.

Boitsov, S., Meier, S., Klungsøyr, J., Svardal, A., 2004. Gas chromatography-massspectrometry analysis of alkylphenols in produced water from offshore oilinstallations as pentafluorobenzoate derivatives. Journal of Chromatography A1059 (1e2), 131e141.

Boitsov, S., Mjøs, S.A., Meier, S., 2007. Identification of estrogen-like alkylphenols inproduced water from offshore oil installations. Marine Environmental Research64 (5), 651e665.

Depledge, M.H., Fossi, M.C., 1994. The role of biomarkers in environmental assess-ment. 2. Invertebrates. Ecotoxicology 3, 161e172.

Gimeno, S., Komen, H., Gerritsen, A.G.M., Bowmer, T., 1998. Feminisation of youngmales of the common carp, Cyprinus carpio, exposed to 4-tert-pentylphenolduring sexual differentiation. Aquatic Toxicology 43 (2e3), 77e92.

Gimeno, S., Komen, H., Venderbosch, P.W.M., Bowmer, T., 1997. Disruption of sexualdifferentiation in genetic male common carp (Cyprinus carpio) exposed to analkylphenol during different life stages. Environmental Science & Technology 31(10), 2884e2890.

Gray, J.S., 2002. Perceived and real risks: produced water from oil extraction.Marine Pollution Bulletin 44 (11), 1171e1172.

Grøsvik, B.E., Meier, S., Westrheim, K., Skarphéðinsdóttir, H., Liewenborg, B., Balk, L.,Klungsøyr, J., 2007. Condition Monitoring in the Water Column 2005: OilHydrocarbons in Fish from Norwegian Waters. Institute of Marine Research,Bergen, Norway, p. 34.

Harman, C., Thomas, K.V., Tollefsen, K.E., Meier, S., Bøyum, O., Grung, M., 2009.Monitoring the freely dissolved concentrations of polycyclic aromatic hydro-carbons (PAH) and alkylphenols (AP) around a Norwegian oil platform byholistic passive sampling. Marine Pollution Bulletin 58 (11), 1671e1679.

Jacobs, R.P.W.M., Grant, R.O.H., Kwant, J., Marquenie, J.M., Mentzer, E., 1992. Thecomposition of producedwater from Shell operated oil and gas production in theNorth Sea. In: Ray, J.P., Engelhardt, F.R. (Eds.), Produced Water: Technological/Environmental Issues and Solutions, vol. 46. Plenum Press, New York, pp. 13e21.

Jørgensen, T., Løkkeborg, S., Soldal, A.V., 2002. Residence of fish in the vicinity ofa decommissioned oil platform in the North Sea. ICES Journal of Marine Science59, 288e293.

Lu, G.H., Song, W.T., Wang, C., Yan, Z.H., 2010. Assessment of in vivo estrogenicresponse and the identification of environmental estrogens in the Yangtze River(Nanjing section). Chemosphere 80 (9), 982e990.

McLeese, D.W., Zitko, V., Sergeant, D.B., Burridge, L., Metcalfe, C.D., 1981. Lethalityand accumulation of alkylphenols in aquatic fauna. Chemosphere 10 (7),723e730.

Meier, S., Andersen, T.C., Lind-Larsen, K., Svardal, A., Holmsen, H., 2007a. Effects ofalkylphenols on glycerophospholipids and cholesterol in liver and brain fromfemale Atlantic cod (Gadus morhua). Comparative Biochemistry and PhysiologyC-Toxicology & Pharmacology 145 (3), 420e430.

Meier, S., Andersen, T.E., Norberg, B., Thorsen, A., Taranger, G.L., Kjesbu, O.S., Dale, R.,Morton, H.C., Klungsøyr, J., Svardal, A., 2007b. Effects of alkylphenols on thereproductive system of Atlantic cod (Gadus morhua). Aquatic Toxicology 81 (2),207e218.

Meier, S., Morton, H.C., Andersson, E., Geffen, A.J., Taranger, G.L., Larsen, M.,Petersen, M., Djurhuus, R., Klungsøyr, J., Svardal, A., 2011. Low-dose exposure toalkylphenols adversely affects the sexual development of Atlantic cod (Gadusmorhua): acceleration of the onset of puberty and delayed seasonal gonaddevelopment in mature female cod. Aquatic Toxicology 105 (1e2), 136e150.

Meier, S., Morton, H.C., Misund, O.A., 2008. Does the discharge of alkylphenols inproduced water cause oestrogenic endocrine disruption among fish in theNorth Sea? Exploration & Production: Oil and Gas Review 6 (II), 64e67.

Meier, S., Morton, H.C., Nyhammer, G., Grøsvik, B.E., Makhotin, V., Geffen, A.,Boitsov, S., Kvestad, K.A., Bohne-Kjersem, A., Goksøyr, A., Folkvord, A.,Klungsøyr, J., Svardal, A., 2010. Development of Atlantic cod (Gadus morhua)exposed to produced water during early life stages effects on embryos, larvae,and juvenile fish. Marine Environmental Research 70 (5), 383e394.

Neff, J.M., Johnsen, S., Frost, T.K., Utvik, T.I.R., Durell, G.S., 2006. Oil well producedwater discharges to the North Sea. Part II: comparison of deployed mussels(Mytilus edulis) and the DREAM model to predict ecological risk. Marine Envi-ronmental Research 62 (3), 224e246.

Neff, J.M., Sauer, T.C., Maciolek, N., 1992. Composition, fate and effects of producedwater discharges to nearshore marine waters. In: Ray, J.P., Engelhardt, F.R. (Eds.),Produced Water: Technological/Environmental Issues and Solutions, vol. 46.Plenum Press, New York, pp. 371e385.

OSPARCOM, 2009. Discharges, Spills and Emissions from Offshore Oil and GasInstallations in 2007, Including the Assessment of Data Reported in 2006 and2007. Offshore Industry Series. OSPAR Commission, London, UK, p. 58.

J. Beyer et al. / Marine Environmental Research 75 (2012) 2e9 9

Petersen, K., Tollefsen, K.E., 2011. Assessing combined toxicity of estrogen receptoragonists in a primary culture of rainbow trout (Oncorhynchus mykiss) hepato-cytes. Aquatic Toxicology 101 (1), 186e195.

Pickford, K.A., Thomas-Jones, R.E., Wheals, B., Tyler, C.R., Sumpter, J.P., 2003. Routeof exposure affects the oestrogenic response of fish to 4-tert-nonylphenol.Aquatic Toxicology 65 (3), 267e279.

Rowland, S.J., West, C.E., Jones, D., Scarlett, A.G., Frank, R.A., Hewitt, L.M., 2011.Steroidal aromatic ‘naphthenic acids’ in oil sands process-affected water:structural comparisons with environmental estrogens. Environmental Science& Technology. doi:10.1021/es202606d.

Rye, H., Reed, M., Slagstad, D., Melbye, A., Johnsen, S., 1996. Modeling Transport andDilution of Produced Water and the Resulting Uptake and Biomagnification inMarine Biota Soc. Petrol. Eng. SPE 35911, New Orleans, Louisiana, 9e12 June1996.

Satoh, K., Nonaka, R., Ohyama, K., Nagai, F., 2005. Androgenic and antiandrogeniceffects of alkylphenols and parabens assessed using the reporter gene assaywith stably transfected CHO-K1 cells (AR-EcoScreen System). Journal of HealthScience 51 (5), 557e568.

Soldal, A.V., Svellingen, I., Jørgensen, T., Løkkeborg, S., 2002. Rigs-to-reefs in theNorth Sea: hydroacoustic quantification of fish in the vicinity of a “semi-cold”platform. ICES Journal of Marine Science 59, 281e287.

Strømgren, T., Sørstrøm, S.E., Schou, L., Kaarstad, I., Aunaas, T., Brakstad, O.G.,Johansen, Ø., 1995. Acute toxic effects of produced water in relation tochemical composition and dispersion. Marine Environmental Research 40 (2),147e169.

Sundt, R., Ruus, A., Jonsson, H., Skarphéðinsdóttir, H., Meier, S., Grung, M., Beyer, J.,Pampanin, D., 2011. Biomarker responses in Atlantic cod (Gadus morhua)exposed to produced water from a North Sea oil field: laboratory and fieldassessments, Marine Pollution Bulletin, doi:10.1016/j.marpolbul.2011.10.005.

Sundt, R.C., Baussant, T., Beyer, J., 2009. Uptake and tissue distribution of C4-C7alkylphenols in Atlantic cod (Gadus morhua): relevance for biomonitoring ofproduced water discharges from oil production. Marine Pollution Bulletin 58,72e79.

Thomas, K.V., Balaam, J., Hurst, M.R., Thain, J.E., 2004. Identification of in vitroestrogen and androgen receptor agonists in North Sea offshore producedwater discharges. Environmental Toxicology and Chemistry 23 (5),1156e1163.

Thomas, K.V., Langford, K., Petersen, K., Smith, A.J., Tollefsen, K.E., 2009. Effect-directed identification of naphthenic acids as important in vitro xeno-estrogensand anti-androgens in North Sea offshore produced water discharges. Envi-ronmental Science & Technology 43 (21), 8066e8071.

Tollefsen, K.E., 2007. Binding of alkylphenols and alkylated non-phenolics to therainbow trout (Oncorhynchus mykiss) plasma sex steroid-binding protein. Eco-toxicology and Environmental Safety 68 (1), 40e48.

Tollefsen, K.E., Blikstad, C., Eikvar, S., Finne, E.F., Gregersen, I.K., 2008. Cytotoxicity ofalkylphenols and alkylated non-phenolics in a primary culture of rainbow trout(Onchorhynchus mykiss) hepatocytes. Ecotoxicology and Environmental Safety69 (1), 64e73.

Tollefsen, K.E., Finne, E.F., Romstad, R., Sandberg, C., 2006. Effluents from oilproduction activities contain chemicals that interfere with normal function ofintra- and extra-cellular estrogen binding proteins. Marine EnvironmentalResearch 62, S191eS194.

Tollefsen, K.E., Harman, C., Smith, A., Thomas, K.V., 2007. Estrogen receptor (ER)agonists and androgen receptor (AR) antagonists in effluents from NorwegianNorth Sea oil production platforms. Marine Pollution Bulletin 54 (3), 277e283.

Tollefsen, K.E., Ingebrigtsen, K., Olsen, A.J., Zachariassen, K.E., Johnsen, S., 1998.Acute toxicity and toxicokinetics of 4-heptylphenol in juvenile Atlantic cod(Gadus morhua L.). Environmental Toxicology and Chemistry 17 (4), 740e746.

Tollefsen, K.E., Nilsen, A.J., 2008. Binding of alkylphenols and alkylated non-phenolics to rainbow trout (Oncorhynchus mykiss) hepatic estrogen receptors.Ecotoxicology and Environmental Safety 69 (2), 163e172.

Utvik, T.I.R., 1999. Chemical characterisation of produced water from four offshoreoil production platforms in the North Sea. Chemosphere 39 (15), 2593e2606.

van der Oost, R., Beyer, J., Vermeulen, N.P.E., 2003. Fish bioaccumulation andbiomarkers in environmental risk assessment: a review. Environmental Toxi-cology and Pharmacology 13 (2), 57e149.

van Leeuwen, C.J., Vermeire, T.G., 2007. Risk Assessment of Chemicals: An Intro-duction. Springer, New York.

Washburn, L., Stone, S., MacIntyre, S., 1999. Dispersion of produced water in a coastalenvironment and its biological implications. Continental Shelf Research 19 (1),57e78.

White, R., Jobling, S., Hoare, S.A., Sumpter, J.P., Parker, M.G., 1994. Environmentallypersistent alkylphenolic compounds are estrogenic. Endocrinology 135 (1),175e182.