Synthesis of the toxicological impacts of the Exxon Valdez ... · M.G. Carls, G.D. Marty, and J.E. Hose Abstract: Pacific herring (Clupea pallasi) in Prince William Sound (PWS) were

PERSPECTIVE

Synthesis of the toxicological impacts of theExxon Valdez oil spill on Pacific herring (Clupeapallasi) in Prince William Sound, Alaska, U.S.A.

M.G. Carls, G.D. Marty, and J.E. Hose

Abstract: Pacific herring (Clupea pallasi) in Prince William Sound (PWS) were affected by two major events in the past de-cade: theExxon Valdezoil spill in 1989 and a 75% collapse in the adult population in 1993. In this review we compare andreinterpret published data from industry and government sources. Combining site-specific estimates of exposure and recentlaboratory effects thresholds, 0.4–0.7µg·L–1 total polynuclear aromatic hydrocarbons, we conclude that 25–32% of the em-bryos were damaged in PWS in 1989. Significant effects extended beyond those predicted by visual observation of oilingand by toxicity information available in 1989. Oil-induced mortality probably reduced recruitment of the 1989 year class intothe fishery, but was impossible to quantify because recruitment was generally low in other Alaskan herring stocks. Significantadult mortality was not observed in 1989; biomass remained high through 1992 but declined precipitously in winter 1992–1993. The collapse was likely caused by high population size, disease, and suboptimal nutrition, but indirect links to the spillcannot be ruled out. These concepts have broad application to future oil spill assessments. For example, safety standards fordissolved aromatics should reflect the previously unrecognized high toxicity of polynuclear aromatic hydrocarbons toadequately protect critical life stages.

Résumé: Deux événements majeurs, le déversement de mazout de l’Exxon Valdezen 1989 et l’effondrement de l’ordrede 75 % de la population adulte en 1993, ont affecté les Harengs du Pacifique (Clupea pallasi) du détroit du Prince-William (PWS). On trouvera ici une comparaison et une ré-interprétation des données déjà publiées sur le sujet parl’industrie pétrolière et les gouvernements. La combinaison des estimations des expositions aux divers sites et la détermi-nation récente en laboratoire des seuils à partir desquels les effets des hydrocarbures aromatiques polynucléaires (PAH) sefont sentir, soit 0,4–0,7µg·L–1, nous amènent à conclure que 25–32 % des embryons du PWS ont été endommagés en1989. Les effets significatifs ont dépassé ceux qu’on pouvait prédire à l’observation visuelle du déversement de mazout età l’aide des données toxicologiques disponibles en 1989. La mortalité due au mazout a sans doute réduit l’apport de laclasse d’âge de 1989 à la population, mais il est impossible de quantifier cette réduction, parce que le recrutement a aussiété faible chez les autres stocks de harengs de l’Alaska. Aucune mortalité significative n’a été observée chez les adultesen 1989; la biomasse est demeurée élevée jusqu’à la fin de 1992 pour décliner très rapidement durant l’hiver 1992–1993.L’effondrement a été sans doute causé par une densité élevée de la population, par la maladie et par une alimentation sub-optimale; on ne peut cependant pas éliminer les effets indirects du déversement de mazout. Ces concepts pourront, engrande partie, s’appliquer à l’évaluation de déversements pétroliers futurs. Par exemple, les normes de sécurité pour lessubstances aromatiques dissoutes devront tenir compte de la forte toxicité des PAH, inconnue encore récemment, si onveut protéger adéquatement les stades critiques des cycles biologiques.

[Traduit par la Rédaction] Carls et al. 172

Introduction

A decade after theExxon Valdezspilled roughly 42 millionlitres of Alaska North Slope crude oil (ANSCO) into Prince

William Sound (PWS), Alaska, controversy continues aboutthe extent and duration of the impact on Pacific herring(Clupea pallasi). Industry investigators (Bienert and Pearson1995; Pearson et al. 1995, 1999) consistently concluded that

Can. J. Fish. Aquat. Sci.59: 153–172 (2002) DOI: 10.1139/F01-200 © 2002 NRC Canada

153

Received 26 January 2001. Accepted 21 November 2001. Published on the NRC Research Press Web site at http://cjfas.nrc.ca on5 February 2002.J16194

M.G. Carls.1 National Marine Fisheries Service, Auke Bay Laboratory, 11305 Glacier Highway, Juneau, AK 99801, U.S.A.G.D. Marty. Department of Anatomy, Physiology, and Cell Biology, School of Veterinary Medicine, University of California,Davis, CA 95616-8732, U.S.A.J.E. Hose.Occidental College, Los Angeles, CA 93420, U.S.A.

1Corresponding author (e-mail: [email protected]).

J:\cjfas\cjfas59\cjfas-01\F01-200.vpFriday, February 01, 2002 9:15:52 AM

Color profile: DisabledComposite Default screen

© 2002 NRC Canada

154 Can. J. Fish. Aquat. Sci. Vol. 59, 2002

oil effects in 1989 were of smaller magnitude, spatial, andtemporal extent than did Natural Resource DamageAssessment (NRDA) investigators (e.g., Brown et al. 1996a;Hose et al. 1996; Marty et al. 1997). Reconciliation ofNRDA and industry data has been complicated by conflict-ing methods for detectingExxon Valdezoil (EVO) in theenvironment, differing estimates of the magnitude of oil ex-posure, mechanisms of potential toxicity, and difficulties ininterpreting population data. By 1990, the preponderance ofdata from both groups suggested that residual oil impactswere undetectable in embryos and larvae. Stock size in PWSwas at near-record levels when the spill occurred, remainedstrong until 1992, and did not appear to be affected by theoil spill (John Wilcock, Alaska Department of Fish andGame, Cordova, Alaska, personal communication). How-ever, the PWS herring population collapsed over the winterof 1992–1993, closing the multimillion-dollar commercialfishery and raising concerns that the collapse was related tothe spill.

The main purpose of this review is to examine evidencefor short-term consequences of the EVO spill on all lifestages of Pacific herring in an effort to reconcile divergentNRDA and industry conclusions. Exhaustive review of thehydrocarbon literature and other oil spills is beyond thescope of this paper; instead, references are limited to thosedescribing weathered oil or specific constituents of EVO orANSCO. Spatial and temporal distributions of oil in PWS in1989 are considered key to understanding the magnitude andmechanisms of exposure and subsequent short-term effects.We performed a number of new data analyses comparingmethods to detect weathering oil in organisms, coupling oilmeasurements in the environment and in organisms, validat-ing the use of resident mussels as indicators of oil exposurefor herring eggs, and refining the estimates of toxic totalpolynuclear aromatic hydrocarbon (TPAH) concentrations,and thus the extent of biologically relevant exposure in PWSin 1989.

A second objective is to discuss the possibility of long-term spill consequences raised by the collapse of the herringpopulation in 1993. Because our findings generally paral-leled those of industry, we provide only a limited discussionof this topic; interested readers should refer to Pearson et al.(1999) and Carls et al. (2001a) for further detail. Pearson etal. (1999) conclude increases in herring biomass and de-creased food supply, in combination with disease or othernatural factors, led to the collapse. Carls et al. (2001a)agree, but do not rule out indirect links to the spill.

Petroleum hydrocarbon measurement, identification,and weathering

Not only do industry and NRDA researchers reach many

conflicting conclusions regarding the impacts of the EVOspill, they also developed different analytical tools to reachthese conclusions. Consequently, we begin with a brief re-view of oil identification methods. These different methodshave, in part, contributed to the differing conclusions reachedby the two groups.

Oil identification models independently developed by in-dustry researchers (Bence and Burns 1995) and NRDA re-searchers (Short and Heintz 1997) successfully recognizepure and weathered EVO in sediment and mussels (Fig. 1),but identification is complicated by changes in polynucleararomatic hydrocarbon (PAH) composition in water and othertissue, such as herring eggs, as a result of differential hydro-carbon accumulation and (or) metabolism. Both models suc-cessfully detected EVO in contaminated mussels, whichhave little ability to metabolize hydrocarbons (Vandermeulenand Penrose 1978; Stegeman 1985; Livingstone et al. 1989).For example, the Short and Heintz model detected EVO inall analyzable Outside Bay mussel samples in 1989, and theBence and Burns model found EVO in 36 of the 46 samples(78%) (Fig. 1). The differences between PAH composition innonmolluscan tissue and source oil that were observed ineggs and adult herring tissues experimentally exposed toANSCO (Carls et al. 1998, 1999, 2000) were anticipated byBence and Burns (1995), who stated that their model wasmost applicable to samples of external surfaces and gastroin-testinal tracts and that shifts in PAH composition in internaltissues lead to uncertainties in identifying EVO in tissue.Similarly, data from the same laboratory tests could not beanalyzed with the Short and Heintz model because somePAH essential for identification were absent. The Bence andBurns model correctly identified EVO in only 4% of theseeggs and did not identify EVO in herring muscle tissue orova, thus identification of contaminant hydrocarbons wasusually erroneous because a source of oil is always assigned.Differences between PAH composition in water and sourceoil also cause difficulties in identification of the source oilbecause dissolution rates decline as the number of aromaticrings increases (Page et al. 1995; Short and Heintz 1997).Thus, composition of PAH in organisms contaminated byexposure to dissolved PAH may frequently not resemblecomposition in the source oil for multiple reasons. Becausethe Short and Heintz (1997) model is less likely to misiden-tify the source of contaminant oil, we prefer it and use itwhere applicable throughout the remainder of this paper.

Oil weathering also alters PAH composition. Although therate at which oil weathers in PWS is extremely variable (theweathering index (w) of oil in beach sediment ranged from 0to 10.7 in 1995 (Carls et al. 2001b)), the close correspon-dence between PAH composition in the field and laboratorydemonstrates that the weathering processes are fundamen-

Fig. 1. Mean total polynuclear aromatic hydrocarbon (TPAH) concentration in (a–b) seawater, (c) mussel tissue, and (d) sediment, asfunctions of time (a) throughout Prince William Sound (Neff and Stubblefield 1995) and (b–d) at specific locations within the NakedIsland area (Short et al. 1996). Where mussel and sediment symbols are solid,Exxon Valdezoil (EVO) was confirmed as the source ofcontamination in one or more samples (Short and Heintz 1997). (Outside Bay mussels (only) were analyzed with two PAH identifica-tion models; EVO was confirmed by both models where symbols are solid; half-filled symbols indicate additional samples where onlythe Bence and Burns (1995) model identified EVO.) Solid symbols for seawater data indicate that one or more samples containednaphthalenes, fluorenes, dibenzothiophenes, and phenanthrenes; (a) composition data from Neff and Stubblefield (1995) were not avail-able. Lines labeled bg are estimated background concentrations. Timing of herring spawn and estimated hatch times (assuming a 24-day incubation time) are indicated. The T/VExxon Valdezspilled oil March 24, 1989.



tally identical (Short and Heintz 1997). Carls et al. (2001b)provide the following verbal descriptions forw: un-weathered,w = 0; slightly weathered, 0 <w ≤ 2; moderatelyweathered, 2 <w ≤ 8; and highly weathered,w > 8. Whereas

the weathering model developed by Short and Heintz (1997)does not explicitly address the weathering of other petro-leum hydrocarbons, e.g., alkanes and the unresolved com-plex mixture (UCM), there is no evidence at present to

© 2002 NRC Canada

Carls et al. 155



suggest that changes in concentration and composition ofthese compounds in laboratory tests would be substantivelydifferent from changes in the field.

Short-term spill consequencesHydrocarbon data from industry and NRDA sources were

reexamined to determine the temporal and spatial extent ofEVO in areas where Pacific herring spawned in 1989. Dataavailable and appropriate for assessment of site contamina-tion include visual observations (e.g., Gundlach et al. 1990;Neff et al. 1995) and chemical analyses of seawater (Neffand Stubblefield 1995; Short and Harris 1996a), sediment(O’Clair et al. 1996; Short and Babcock 1996), molluscs(e.g., Brown et al. 1996a; Short and Babcock 1996), andherring eggs (Pearson et al. 1995; Short et al. 1996). TPAHconcentrations in seawater, sediment, and mussels fromNRDA studies were obtained directly from the NRDA hy-drocarbon database (Short et al. 1996). Published data, in-cluding text and graphics from industry-sponsored researchwere included as appropriate, but detailed data from industrystudies were generally not available. Sites where herringspawned that were nominally classified as oiled (the NakedIsland area and northern Montague Island) were examinedmost closely. There was no argument between industry andNRDA that sites in the northern and northeastern portions ofPWS were not oiled (e.g., Fairmont Island and Tatitlek Nar-rows), and these sites were used in NRDA studies as EVO-free reference areas.

Spatial distributions of oil within spawning areasPacific herring spawn biomass and distances spawned in

PWS were estimated by the Alaska Department of Fish andGame (Brady et al. 1991), and both industry and NRDAresearchers accepted these estimates (Bienert and Pearson1995; Brown et al. 1996a, 1996b). Of the total spawned bio-mass, 32% was in the northern Montague Island area, 19%was in the Naked Island area, and the remainder was in thenorthern and northeastern areas (Brady et al. 1991). Dis-tances of shoreline with spawn in oiled areas were relativelysmaller, 29% and 14% of total spawn length, for theMontague Island and Naked Island areas, respectively(Brady et al. 1991). Industry estimates of potentially oiledherring spawn were based on spawn distance (Pearson et al.1995), whereas NRDA estimates were based on spawn bio-mass (Brown et al. 1996a, 1996b).

Designation of potentially oiled spawn areas by NRDAand industry were identical (Pearson et al. 1995; Brown etal. 1996a). The Naked Island area and northern MontagueIsland were both within the observed distribution of oil(Gundlach et al. 1990) (Figs. 2–3). Oiling was documentedalong the northern shoreline of Montague Island, includingareas in Zaikof Bay, Rocky Bay, Montague Point, GraveyardPoint, and Stockdale Harbor (Brady et al. 1991). Islandsmoderately or heavily impacted by oil included Naked,Storey, and Ingot islands (Brady et al. 1991). In agreementwith Brown et al. (1996a, 1996b), Bienert and Pearson(1995) found 43% of the total miles of spawn were depos-ited within the slick trajectory, but based on a visual shore-line survey of oiling by Neff et al. (1995), Pearson et al.(1995) concluded that only 4% of the total spawn distancewas at risk of coming into direct physical contact with oil.

Biological availability of oil at spawning sites in 1989EVO was identified as the source of contamination in

1989 in herring spawning sites throughout the Naked Islandand Montague Island areas, except for Zaikof Bay, whereinsufficient data were available (Figs. 2–3). In the NakedIsland area, EVO was confirmed in mussel samples fromCabin Bay, Outside Bay, Bass Harbor, on the north shore ofStorey Island, and on a small islet to the south of Storey Is-land (Short and Harris 1996b; Short and Heintz 1997). Mus-sels tethered 25 m below the water surface in Outside Baywere contaminated with EVO (Short and Harris 1996b). OnMontague Island, EVO was confirmed in mussel tissue fromRocky Bay and Stockdale Harbor and inLittorina samplesnear Graveyard Point. Water samples provided additional ev-idence of EVO in Cabin Bay and Rocky Bay. In contrast,EVO was rarely identified in sediments collected from thesespawning areas, and TPAH concentrations were only slightlyelevated (<0.5µg·g–1) if at all (Figs. 1 and 4). Thus, EVOwas identified in samples from most Pacific herring spawnsites (Short and Heintz 1997) within the slick trajectory(Gundlach et al. 1990), providing justification for designat-ing individual sites within the trajectory as oiled.

Although there was little overlap between visibly oiledshoreline and herring spawn, there was strong evidence ofbiologically available oil in spawn areas in 1989 (Figs. 2–3).Industry estimates that 4% of the spawn length occurredalong visibly oiled shorelines (Pearson et al. 1995) are simi-lar to our estimate (6%). However, widespread contamina-tion of mussels throughout much of the spawn lengthprovides evidence that EVO was biologically available inareas without visible shoreline oiling, as do elevated TPAHconcentrations in herring eggs-on-kelp (Pearson et al. 1995).For example, mean TPAH concentrations in mussels at thehead of Outside Bay rose to a maximum of 7.0µg·g–1 dryweight at a distance of 0.6–1.0 km from the nearest small(<0.1 km) patch of visible oil identified in shoreline surveys.TPAH concentrations in intertidal sediment at the head ofOutside Bay remained≤0.2 µg·g–1, providing further evi-dence of minimal sediment contamination where oil was bi-ologically available. Mussels tethered 0.5 km offshore fromthe nearest oiled shoreline in Outside Bay at depths of 1–25 m accumulated EVO (mean TPAH concentrations rangedfrom 0.6 to 1.5µ·g–1 dry weight in May 1989) (Short andHarris 1996b), demonstrating that dispersed oil was biologi-cally available in the water column. Thus, oil was biologi-cally available over a much broader range than suggested byvisibly oiled shorelines or nearby sediment contamination.

Temporal changes in PAH

SeawaterTPAH concentrations in seawater peaked in the Naked Is-

land (1.9 µg·L–1) and Montague Island (2.6µg·L–1) areassoon after the EVO spill, which was shortly before or at thetime Pacific herring spawned in these areas (Brown et al.1996a; Short and Harris 1996a; Figs. 1 and 4). Total PAHconcentrations in surface water, averaged throughout PWS,peaked in late April (Neff and Stubblefield 1995). Seawatersamples at peak concentrations in Rocky Bay and all watersamples in the Naked Island area contained naphthalenes,fluorenes, dibenzothiophenes, and phenanthrenes at concen-

© 2002 NRC Canada




trations above the method detection limit, strongly suggest-ing the source of contamination was EVO. The rapid increasein aqueous TPAH concentrations after the spill, followed by

a decline, is unambiguous evidence for a single-event spill,and also suggests that EVO was the source of contamina-tion. Peak TPAH concentrations in water occurred about the

© 2002 NRC Canada

Carls et al. 157

Fig. 2. Location of herring spawn and water, mussel, and sediment hydrocarbon collection sites in 1989 in the Naked Island area. Themapped area was entirely within theExxon Valdezoil (EVO) slick trajectory (Gundlach et al. 1990). Herring spawn data (bold linesdrawn slightly offshore) are from Brady et al. (1991). Where symbols are solid, EVO was confirmed as the source of contamination inone or more samples (Short and Heintz 1997). Sites labeled O1–O16 were sampled by Brown et al. (1996a), T1A–T33A by Pearson et al.(1995), and unlabeled sites represent a mixture of other studies (Short et al. 1996). Visibly oiled samples are annotated (e.g., medium oil),and + mark approximate positions of oil observed during herring spawn surveys (Brown et al. 1996b). Visible shoreline oiling (thick linesdrawn slightly onshore) was categorized as light (L), medium (M), or heavy (H) (Gundlach et al. 1990; Neff et al. 1995).



same time as peak concentrations in mussels in the NakedIsland area (Fig. 1), but earlier than in mussels farther southat Rocky Bay (Fig. 4).

SedimentEVO was infrequently detected in intertidal sediment from

spawn areas within the oil trajectory, and TPAH concentra-tions were generally low throughout 1989 (Figs. 1 and 4). In1989, TPAH concentrations were not elevated in sediment atthe head of Outside Bay (≤0.2 µg·g–1), but concentrations atRocky Bay were slightly greater than estimated backgroundconcentrations (<0.5µg·g–1) and remained elevated untilmid-1991 (Short and Babcock 1996). However, divers ob-served oil in sediment in 1989 at Outside Bay, Bass Harbor,Cabin Bay, and Montague Point to Graveyard Point (Pearson

et al. 1995), but intertidal sediments were often not collecteduntil 1990. EVO was detected in sediment in 1990 at two ofthese sites (Cabin Bay and Storey Island; Short and Heintz1997). Although the extent of sediment sampling in spawnareas was very limited, these data generally corroborate visi-ble evidence of shoreline oiling in 1989 (Gundlach et al.1990; Neff et al. 1995) or absence thereof.

MusselsMussel samples provided strong evidence that oil was bio-

logically available in herring spawn areas within or adjacentto the slick trajectory in 1989 (Figs. 1 and 4). Total PAHconcentrations in mussel tissue increased after the spill,peaked at about the time Pacific herring spawned in the Na-ked Island area, and peaked at about the time of embryo

© 2002 NRC Canada


Fig. 3. Location of herring spawn and water, mussel,Littorina, and sediment hydrocarbon collection sites in 1989 on northernMontague Island. Herring spawn data (bold lines drawn slightly offshore) are from Brady et al. (1991). Where symbols are solid,Exxon Valdezoil (EVO) was confirmed as the source of contamination in one or more samples (Short and Heintz 1997). Sites sampledby Brown et al. (1996a) for mussels and herring are labeled as originally designated by the authors (O17–O19); unlabeled sites andsite M16 represent a mixture of studies (Short et al. 1996). Visibly oiled samples are annotated (e.g., medium oil), and + markapproximate positions of oil observed during herring spawn surveys (Brown et al. 1996b). Visible shoreline oiling (lines drawn slightlyonshore) was categorized as light (L), medium (M), or heavy (H) (Gundlach et al. 1990; Neff et al. 1995). Oil slick data (crosshatch)are from Gundlach et al. (1990).



hatch in the Montague Island area. The complete time seriesdata for Outside Bay and Rocky Bay provide a frame of ref-erence against which other mussel samples could be judgedin the these areas. At Outside Bay, TPAH concentrations inmussels increased before herring spawned, were high duringspawning (>6.1µg·g–1 dry weight), and remained elevatedthrough hatch (>2.5µg·g–1) (Fig. 1). Total PAH concentra-tions in mussels at all other Naked Island sites were simi-

larly high at about the time of peak concentration in OutsideBay (3.4–7.5µg·g–1) (Fig. 1). Total PAH concentration inmussels collected near the Ingot Island herring spawn, a siteincluded in the Naked Island area, ranged from 9.9 to15.4µg·g–1 in September 1989, suggesting high exposures inthe spring. (Both industry (Pearson et al. 1995) and NRDA(Brown et al. 1996a) agree that the Ingot Island area wasmoderately to heavily oiled.) At Rocky Bay, TPAH concen-

© 2002 NRC Canada

Carls et al. 159

Fig. 4. Mean total polynuclear aromatic hydrocarbon (TPAH) concentration in (a) seawater, (b) mussel tissue, and (c) sediment asfunctions of time in Rocky Bay, Stockdale Harbor, and Zaikof Bay. Also included are TPAH concentrations inLittorina spp. nearGraveyard Point. Where mussel and sediment symbols are solid,Exxon Valdezoil (EVO) was confirmed as the source of contamination inone or more samples (Short and Heintz 1997). Solid symbols for seawater data indicate that one or more samples contained naphthalenes,fluorenes, dibenzothiophenes, and phenanthrenes. Lines labeled bg are estimated background concentrations. Timing of herring spawn andestimated hatch times (assuming a 24-day incubation time) are indicated. The T/VExxon Valdezspilled oil March 24, 1989.



trations in mussels increased during spawning and peaked at4.4 µg·g–1 at about the time of embryo hatch (Fig. 4).Around the time of peak concentration in Rocky Bay, TPAHwas high (9.2µg·g–1) in Littorina spp. near Graveyard Point,but only slightly elevated (0.5µg·g–1) in mussels at Stock-dale Harbor. Mussel collection in Zaikof Bay began 31 July1989, well after expected concentration peaks.

WeatheringIncreases in oil weathering (w), estimated from mussel tis-

sue, were roughly linear in the first three months (P < 0.001),but correlation betweenw and time was low (r 2 = 0.28;Fig. 5). Weathering (w) of EVO ranged from 0.7 to 5.7 inherring-spawn areas during the critical egg incubation periodand averaged 3.7 (note:w is a unitless number).

Exposure of herring eggs to oilPearson et al. (1995) concluded that mean TPAH concen-

trations in eggs-on-kelp did not differ significantly betweenoiled and reference areas in 1989, despite obvious differ-ences between many of the oiled sites and the reference sites(Fig. 6). Mean TPAH concentrations by tide elevationranged from approximately 27 to 69 ng·g–1 at reference sitesand 13 to 342 ng·g–1 wet weight at oiled sites; maximumconcentrations ranged up to 1000 ng·g–1 at Cabin Bay, wheresome eggs were visibly coated by oil (Pearson et al. 1995).The highest mean TPAH concentrations were consistently inthe upper tide zone (Pearson et al. 1995), but correlation ofTPAH among tide zones was high (r 2 = 0.93), demonstratingthat surface and subsurface oil concentrations were related(our analysis). Less than 4% of the eggs, all from Cabin Bay,were visibly coated with oil, and the proportion of developedeggs was inversely related to concentration at this site.Pearson et al. (1995) concluded that only a minor portion ofthe 1989 spawn, the 2% of the spawn length at Cabin Bay,was affected by the spill.

Herring eggs collected by Brown et al. (1996a) were ana-lyzed for hydrocarbons, but detection problems limited theusefulness of this collection and the results were not pub-lished. Total PAH concentrations in herring eggs collectedby Brown’s group were significantly elevated at Cabin Bay(PANOVA < 0.001), but not at other oiled sites (our analysis).Brown et al. (1996a) concluded that the mass of eggs sam-pled was insufficient for adequate detection of hydrocarbons,and thus relied on hydrocarbon concentrations in musselscollected within herring spawn areas as surrogate measuresof exposure.

The source of PAH in herring eggs within the slick trajec-tory in PWS in 1989 was probably EVO because seawater,sediment, and mussels were all contaminated with EVO inthese areas. We cannot directly test the hypothesis that theherring eggs-on-kelp samples collected by Pearson et al.(1995) were contaminated with EVO because the data wereunavailable. Pearson et al. (1995) reported that EVO wasidentified in only three eggs-on-kelp samples, based on theBence and Burns (1995) model, but this model failed to cor-rectly identify ANSCO as the source of contamination inherring eggs exposed to this oil in laboratory tests by Carlset al. (1999, 2000). Moderately weathered EVO was identi-fied as the source of contamination in a NRDA herring eggsample from Cabin Bay (w = 2.3; Short and Heintz 1997).

Exposure of mussels and eggs to oil in spawn areas waswater mediated

We infer that EVO was biologically available primarilyfrom the water column in herring spawn areas because hydro-carbon concentrations were elevated in water but not inintertidal sediment. Mackay et al. (1980) suggested the ma-jor toxic effects of an oil spill may be from hydrocarbonsdissolved from dispersed oil. Oil can be dispersed in the wa-ter column by wind or wave activity (Mackay et al. 1980;Payne et al. 1991; Wolfe et al. 1994), and three days afterthe Exxon Valdezspill a three-day storm dispersed substan-tial quantities of oil into the water (Wolfe et al. 1994). Thisdispersed oil was clearly accumulated by mussels tethered atdepths to 25 m(Short and Harris 1996b). Phytane was presentin 77% of Naked Island area mussels and 58% of Montague

© 2002 NRC Canada


Fig. 5. Estimated state of oil weathering at herring spawn sitesin Prince William Sound in 1989. Weathering (w) was estimatedfrom polynuclear aromatic hydrocarbon (PAH) composition inmussel tissue collected from oiled areas by Brown et al. (1996a)with the methods of Short and Heintz (1997). Circles representdata from the Naked Island area; diamonds represent data fromthe Rocky Bay area on Montague Island. Timing of the herringegg incubation period is indicated. The T/VExxon Valdezspilledoil March 24, 1989.



Island area mussels in 1989, confirming the presence of par-ticulate oil in the majority of samples (Short et al. 1996).Dispersion of oil into water greatly increases the surfacearea of the oil, thus accelerating the weathering processes

described by Short and Heintz (1997) and increasing theconcentrations of dissolved hydrocarbons, particularly of thehigher molecular weight, less soluble compounds.

All available evidence indicates that the primary source ofherring egg contamination by oil in PWS was water medi-ated. The PAH in nearly all eggs-on-kelp samples from PWS(Pearson et al. 1995) must have been accumulated from sur-rounding water because (i) TPAH were present in the watercolumn (Neff and Stubblefield 1995; Short and Harris1996a); (ii ) TPAH were accumulated from water by bothcaged (Short and Harris 1996b) and native mussels (Brownet al. 1996a) in spawn areas; (iii ) composition of TPAH inthe only herring egg sample available to us (Short et al.1996) is consistent with contamination by dissolved PAH;(iv) <4% of herring eggs had visible tarry oil deposits(Pearson et al. 1995); and (v) oil in beach sediment was in-frequently encountered and present only at low concentra-tions where detectable (Pearson et al. 1995; Short andBabcock 1996; Short et al. 1996). The argument put forwardby Pearson et al. (1999) that the ascites (an accumulation offluid in the body cavity) observed in herring larvae in PWSby Marty et al. (1997) must have resulted from direct coat-ing of eggs by oil is unlikely. A coating incidence of <4%cannot explain the >16% elevation in the incidence ofascites in oiled areas. Furthermore, Hay et al. (1995) re-ported that herring eggs in direct contact with oil invariablydied, an observation corroborated by Pearson et al. (1995),who found higher mortality in eggs coated with oil at CabinBay. Thus, the damaged larvae observed by Marty et al.(1997) likely resulted from exposure of eggs to aqueousPAH, not oil coating.

EVO concentrations in mussels and herring eggs werecorrelated in 1989

Total PAH concentrations in mussels collected by Brownet al. (1996a) from spawn sites were correlated with TPAHconcentrations in herring eggs (Pearson et al. 1995) at thesame sites, justifying the use of mussel data as surrogates forherring egg exposure to EVO in PWS. Brown et al. (1996a)and Hose et al. (1996) did not report TPAH concentrationsin herring eggs, but rather used TPAH concentrations inmussels adjacent to herring egg collections as surrogatemeasures of hydrocarbon exposure of herring eggs becausemussels and other suspension-feeding bivalves are fre-quently used for monitoring sporadically distributed hydro-carbons in seawater (e.g., NRC 1980; Wolfe et al. 1981;Armstrong et al. 1995). To determine if using TPAH in thesemussels as surrogates for herring egg exposure was justifi-able, mussel and herring egg data from sites sampled incommon by both research teams were regressed (Fig. 7).These sites included Cabin Bay, Outside Bay, Storey Island,Bass Harbor, and Stockdale Harbor. (Total PAH data formussels in Stockdale Harbor were collected by the U.S. Fishand Wildlife Service and reported by Short et al. (1996).Transects O1, O14, and O15 (McGurk et al. 1990; Brown etal. 1996a) were considered outside areas of overlap andwere not included (Fig. 2)). Mean TPAH concentrations inmussels (Brown et al. 1996a) were correlated with those inherring eggs (Pearson et al. 1995) (r 2 = 0.92, P = 0.011)(Fig. 7). Bienert and Pearson (1995) previously argued thatmussel data may be of limited value in estimating exposure

© 2002 NRC Canada

Carls et al. 161

Fig. 6. Mean total polynuclear aromatic hydrocarbon (TPAH)concentration in Pacific herring eggs-on-kelp (Pearson et al.1995a). Error bars indicate range of means by tide zone.Background TPAH concentrations in herring egg tissue estimatedfrom (a) laboratory exposures (Carls et al. 1999) and (b) PWS(Pearson et al. 1995a) are indicated. To compensate fordifferences in background concentrations, estimates of damagingconcentrations were scaled to field background estimates. (c) Theminimum estimated zone of concentrations sufficient to causeabnormalities is based on several estimates of the lowestobserved effective concentrations in the laboratory (Carls et al.1999) applied to two estimates of field background concentrations(see text). (d) Estimation of minimum concentrations sufficient tobe directly lethal were calculated similarly.



of herring eggs deposited subtidally, but demonstration(i) that hydrocarbons were available to subsurface marinefauna after theExxon Valdezspill (Short and Harris 1996b);(ii ) that TPAH concentrations in herring eggs were corre-lated across tide zones; and (iii ) that TPAH concentrationsin mussels were correlated with those in herring eggs sug-gests otherwise. Thus, the use of TPAH concentrations inmussels as surrogates for herring egg exposure to EVO inPWS by Brown et al. (1996a) and Hose et al. (1996) was valid.

Oil effects in PWS herring eggsAdverse reactions of Pacific herring eggs to EVO in PWS

were documented at hatching and in newly hatched larvaeby NRDA studies (Brown et al. 1996a), but the industrystudy concluded that oil effects were generally negligible(Pearson et al. 1995). Response data in all studies focused

primarily on the condition of larvae hatched from exposedeggs. (Eggs naturally spawned in PWS were collected late indevelopment by NRDA and industry researchers and incu-bated in laboratories until hatch.) Premature hatch (Brown etal. 1996a) and abnormalities in Pacific herring larvae (Hoseet al. 1996), effects consistent with exposure to crude oil(e.g., Smith and Cameron 1979; Weis and Weis 1989; Carlset al. 1999), were significantly more frequent in oiled areasthan in reference areas of PWS. Compared with referenceembryos, oiled embryos hatched earlier, producing less ma-ture larvae (Brown et al. 1996a). The proportion of larvaewith jaw deformities varied significantly with oil concentra-tions in resident mussels (Brown et al. 1996a; Hose et al.1996). Oiled larvae were longer at hatch, but weighed lessthan reference larvae (Brown et al. 1996a). Hose et al.(1996) found that the severity of skeletal, craniofacial, and

© 2002 NRC Canada


Fig. 7. Mean total polynuclear aromatic hydrocarbon (TPAH) concentrations in herring eggs-on-kelp (ng·g–1 wet weight; Pearson et al.1995) and in adjacent mussels (µg·g–1 dry weight; Brown et al. 1996a) were correlated (r 2 = 0.92; bounding curves are the 95% confi-dence bands). Data are means ± standard error. Minimum concentrations in mussels sufficient to predict damage in herring embryoswere estimated from minimum concentrations sufficient to cause (a) abnormalities or (m) mortality in Prince William Sound herringembryos (see Fig. 6 and text).



finfold abnormalities was significantly higher in oiled areascompared with a reference site, Fairmont Bay. The severityof skeletal abnormalities, certain types of craniofacial de-fects (jaw abnormalities, microphthalmia, and absence ofotic capsules), and the total severity index were significantlycorrelated to log-transformed TPAH concentrations in adja-cent mussels (Hose et al. 1996). Not clear is why Pearson etal. (1995) generally did not observe similar effects in embryo-larval herring; possibly the statistical power of these testswas too low. The only correlation to oil that Pearson et al.(1995) found was a reduction in the proportion of developedeggs at Cabin Bay. Comparison of biological responsesamong sites by Pearson et al. (1995) was limited to a singlestatistical contrast between oiled and reference areas, but be-cause mean TPAH concentrations in eggs-on-kelp (and inmussels) varied considerably among oiled sites, this compar-ison may have missed significant intersite variation. BecausePearson et al. (1995) only presented the proportions of bio-logical response variables without reporting the total numberof individuals examined, and generally did not provide site-specific and tide zone detail, direct comparison with NRDAdata is not possible.

Oil effects in laboratory-exposed eggsExtensive research has consistently demonstrated that

aqueous exposure of fish eggs to oil is damaging (e.g., seereviews by Moore and Dwyer 1974; Neff and Anderson1981; Weis and Weis 1989); however, much of this literaturefocused on acute toxicity and one- to two-ring aromatics.Because one-ring compounds were quickly lost from the wa-ter of PWS (Wolfe et al. 1994; Neff and Stubblefield 1995),subsequent toxicity research was specifically tailored to studyconstituents remaining after the initial weathering (e.g., Martyet al. 1997; Carls et al. 1999; Heintz et al. 1999).

The quantitative laboratory studies of Carls et al. (1999)provide a link between field and laboratory toxicity forembryolarval herring and can be used to hindcast biologicalresponses in PWS to EVO in 1989. Herring eggs were ex-posed to concentration series of less weathered and moreweathered oil in two consecutive experiments where seawater(32 ppt) was contaminated by passage through oiled gravel.Originally reported weathering estimates ranged from 0.04to 1.28; amended ranges are 0.1≤ w ≤ 1.0 (less weatheredtest) and 0.5≤ w ≤ 1.5 (more weathered test). The resultantoil concentrations and composition closely modeled thoseobserved in PWS after the spill (Short and Heintz 1997) andTPAH concentrations declined exponentially during tests.Total PAH accounted for 36–79% of total hydrocarbons dis-solved in seawater in the less weathered test and 3–13% inthe more weathered test. Exposure of eggs to 0.7µg·L–1

TPAH of the more weathered oil caused malformations, ge-netic damage, mortality, decreased size, and inhibited swim-ming in herring larvae. Total aqueous PAH concentrations aslow as 0.4µg·L–1 caused sublethal responses such as edema(ascites) and immaturity consistent with premature hatching.Responses to less weathered oil, which had relatively lowerproportions of high molecular weight PAH, generally paral-leled those of the more weathered oil, but lowest observedadverse effect concentrations (LOAECs) were higher(9.1 µg·L–1), demonstrating the importance of composition.Based on total measured aqueous hydrocarbons (TPAH +

total alkanes + UCM), LOAECs in the less and moreweathered herring egg tests were 25µg·L–1 and 14µg·L–1

(previously unreported by Carls et al. 1999). Correlationsbetween biological responses and TPAH were consistentlygreater than correlations to either total alkanes or the UCM.

To properly match laboratory and field results, the state ofoil weathering, summarized byw, was considered. Meanw,as estimated from mussel tissue, was 3.7 (range 0.7–5.7,n =51) during the critical egg development period in PWS(1989) and 2.3 in herring eggs (n = 1), thus generally ex-ceeding weathering in both experiments reported by Carls etal. (1999). However, Heintz et al. (1999) provide evidencethat ANSCO remains toxic with further weathering (w =4.9), and continues to be more toxic per unit mass thanunweathered oil. Thus, use of the more weathered test byCarls et al. (1999) for interpretation of field results is morejustifiable than use of the less weathered test, but may un-derestimate the true toxicity owing to continued weathering.True toxicity may also be underestimated because exposureto ultraviolet light was negligible during laboratory incuba-tion, but developing embryos in PWS were likely exposed toultraviolet light from sunlight (Barron and Ka’aihue 2001),and absorption of ultraviolet light can increase PAH toxicityby two to 1000 times (Pelletier et al. 1997). Tidally influ-enced fluctuations in salinity were not modeled by the labo-ratory experiment, but could also have influenced toxicity inPWS. For example, Vines et al. (2000) found that the in-creased incidence of abnormalities of herring embryos ex-posed to creosote was less at 28 ppt than at lower salinities(16 ppt and 8 ppt).

Laboratory toxicity estimates suggest toxicity occurredin PWS

Three independent estimates of herring egg exposure tohydrocarbons in PWS in 1989, interpreted by laboratory ex-posures, provide evidence that the oil spill was damaging:TPAH concentrations in water, herring eggs, and mussels.These predictions of damage were confirmed by observationof larvae collected from oiled areas in PWS with abnormali-ties consistent with exposure to oil (Marty et al. 1997).

WaterDirectly measured aqueous TPAH concentrations in PWS

in 1989 exceeded concentrations detrimental to herringembryos and contaminated water contained high molecularweight compounds (e.g., phenanthrenes and chrysenes).Peak aqueous TPAH concentrations in Cabin Bay and RockyBay (1.9–2.6µg·L–1; Short and Harris 1996a) exceededLOAECs causing abnormalities in herring embryolarvae(0.4µg·L–1) in controlled laboratory experiments (Carls et al.1999). Concentrations of TPAH in surface water averagedover a wider area also exceeded LOAECs (Neff andStubblefield 1995); intertidally spawned eggs were poten-tially exposed to surface sheens as tides fluctuated. Argu-ments that imply that EVO did not harm marine organismsbecause TPAH concentrations never exceeded water qualitystandards (e.g., Neff and Stubblefield 1995; Neff and Burns1996; Pearson et al. 1999) are invalidated by postspill fieldand laboratory observations to the contrary (e.g., Hose et al.1996; Marty et al. 1997; Carls et al. 1999).

© 2002 NRC Canada

Carls et al. 163



Herring eggsInterpretation of 1989 field results requires comparison of

observed TPAH concentrations in eggs-on-kelp from PWSwith biologically significant concentrations observed in lab-oratory tests. Recently published estimates by Carls et al.(1999) indicate that the mean LOAEC causing abnormalitieswas 22 ± 4 ng·g–1 wet weight TPAH (mean of 4–16 days ex-posure data). Average area under the curve extending overall exposure days (0–16 days) might have been a more ap-propriate estimate of LOAEC causing abnormalities,19.7 ng·g–1, and provides a lower estimate for minimumdamaging exposure concentrations. The peak LOAEC caus-ing abnormalities was 26.0 ng·g–1, providing an upper esti-mate of minimum damaging exposure concentrations. Thepublished LOAEC causing embryolarval mortality was108 ± 35 ng·g–1 in herring egg tissue (mean of 1–16 daysdata; Carls et al. 1999). Again, a more appropriate estimatewould be the average area under the curve (0–16 days),123 ng·g–1. The peak LOAEC causing mortality was226 ng·g–1.

Complicating application of biologically significant labo-ratory concentrations to field results is a difference in fieldand laboratory baseline concentrations. The inferred back-ground TPAH concentration for PWS eggs-on-kelp was themean reference concentration (50.4 ng·g–1; our calculation).Reasons why this estimate of background concentration ineggs-on-kelp was higher in PWS (Pearson et al. 1995) thanin the laboratory (10.7 ± 1.9 ng·g–1; Carls et al. 1999) arenot clear, but the possibility of contamination at referencesites in PWS cannot be discounted. Other possibilities in-clude differences in analytical and sampling procedures(e.g., kelp was included in samples analyzed by Pearson etal. (1995), but not in laboratory samples (Carls et al. 1999)).To compensate for these background differences, we calcu-lated the difference between each laboratory estimate andthe mean laboratory background concentration, and addedthese differences to the field background concentration. Toestimate an upper limit for minimum concentrations causingabnormalities, we increased the estimated background con-centration for PWS to the maximum mean concentration re-ported for any reference tide zone in 1989 (69 ng·g–1;Pearson et al. 1995).

Direct measurement of TPAH in Pacific herring eggs in1989 (Pearson et al. 1995), interpreted by recent experiments(Carls et al. 1999), suggests that exposure of eggs to EVO infive of six oiled sites in PWS was detrimental. In contrast,Pearson et al. (1995) concluded that mean TPAH concentra-tions in eggs-on-kelp from PWS in 1989 did not differ sig-nificantly between oiled and reference areas and that onlythree samples, representing two sites, were contaminatedwith EVO. Estimated minimum concentrations that causedabnormalities in PWS herring eggs, 59.4–84.6 ng·g–1 wetweight, included four or five of six oiled sites; observed con-centrations in Bass Harbor fell within the zone of uncer-tainty (Fig. 6). However, larvae at Bass Harbor were shorterand grew more slowly than larvae from reference sites, and anumber of histopathological and cytogenetic scores were ele-vated (finfold, cytologic, and craniofacial lesions, yolk,ascites, and anaphase aberration; Marty et al. 1997).McGurk and Brown (1996) found mean egg–larval mortality

at Bass Harbor was greater than at reference sites, also dem-onstrating that these effects were biologically significant.Although growth differences could have been influenced byunrelated site-specific variables, all of these differences areconsistent with exposure to oil and confirm that the modelaccurately predicts occurrence of negative effects, includingthose in Bass Harbor, a site within the zone of uncertainty.One or two sites may have experienced minimum concentra-tions sufficient to be directly lethal (147–285 ng·g–1), CabinBay and portions of Outside Bay (Fig. 6). Observation ofeggs in contact with oil at these two sites (Pearson et al.1995) together with the observation by Hay et al. (1995) thateggs in direct contact with oil invariably die early in devel-opment suggests the accuracy of our interpretation that oilconcentrations at these two sites could be directly lethal. Weconclude that oil-induced abnormalities, which likely pres-aged delayed mortality, were plausible at 83% of the oiledsites sampled by Pearson et al. (1995) and were confirmedin the oiled site least likely to produce significant differ-ences, Bass Harbor (Marty et al. 1997).

MusselsInterpretation of TPAH concentrations in mussels, which

were correlated with TPAH concentrations in herring eggs,provides a third estimate of toxicity, and demonstrates thatdamage to herring embryos was plausible in Rocky Bay, asite not sampled by Pearson et al. (1995, 1999). Minimumconcentrations in mussel tissue sufficient to predict adversereaction in adjacent herring embryos were estimated fromthe regression between TPAH concentration in herring eggs-on-kelp and mussels (Fig. 7). Ranges of critical exposureconcentrations estimated from TPAH in mussel tissue aregreater than corresponding ranges estimated directly fromherring eggs due to uncertainties in the linear fit. Minimumconcentrations sufficient to predict abnormalities rangedfrom 1.0 to 4.8µg·g–1 dry weight and those sufficient to bedirectly lethal ranged from 4.2 to 15.3µg·g–1 (Fig. 7). Themean TPAH concentration in Rocky Bay mussels was>3.4µg·g–1 but <4.8µg·g–1, suggesting negative effects wereplausible. Indeed, Rocky Bay larvae were shorter and grewmore slowly than larvae from reference sites and exhibitedhistopathological and cytogenetic damage consistent with oilexposure (Marty et al. 1997). McGurk and Brown (1996)found mean egg–larval mortality at Rocky Bay was greaterthan at reference sites, demonstrating that these biologicaleffects were consequential. Elevated TPAH concentration inNaked Island mussels was also consistent with observed lar-val damage.

Laboratory exposure experiments provide usefulmodeling of real-world effects

That the laboratory results of Carls et al. (1999) predictedresponse of PWS herring larvae to EVO with considerableprecision also demonstrates that these tests accurately modelreal-world spill conditions and are not misleading, as im-plied by Neff et al. (2000). Hydrocarbons other than PAHmay contribute to petroleum toxicity, but the toxicity of PAHis well recognized. At a minimum, TPAH serves as an indexof the presence of oil and was the only measure of oil con-centration common to the studies presented in this synthesis.

© 2002 NRC Canada




In most cases, acute toxicity of a petroleum product is di-rectly correlated to its content of soluble aromatic deriva-tives (Anderson et al. 1974; Moore and Dwyer 1974; Neff1979). Neff et al. (2000) reported that PAH accounted for 3–94% of estimated hazard indices of four oils and that thecontribution increased with weathering: TPAH contributionto toxicity in the most weathered fractions ranged from 57 to94% (omitting a 99% observation for one oil because of ana-lytical errors). Phenols contributed <15% to the toxicity inany test by Neff et al. (2000), and did not vary consistentlyas a function of weathering. Similarly, Barron et al. (1999)also found that the least toxic of three oils had by far thehighest phenol concentration. Retention of the higher molec-ular weight, more refractory, and more toxic PAH in aque-ous solution as the oil weathers appears to best explain theincreased toxicity of weathered oil (Carls et al. 1999; Heintzet al. 1999, 2000).

Although microorganisms throughout the marine environ-ment have evolved the ability to metabolize hydrocarbons(Atlas 1981; Floodgate 1984), and the resultant metabolicbyproducts can be toxic (Middaugh et al. 1998; Shelton etal. 1999), the estimated microbial contribution to toxicity intests by Carls et al. (1999) is minimal. Concentration of thewater-soluble fraction (WSF) varied among microbial cul-tures and increased with time (Middaugh et al. 1996, 1998;Shelton et al. 1999); concentration changes explained >70%of the variation in toxicity (our analysis). Furthermore, mi-crobial degradation of oil is typically nutrient limited (e.g.,Gibbs and Davis 1976; Atlas 1995), as was the case inwater-column tests (Middaugh et al. 1996, 1998; Shelton etal. 1999) and in experimentally oiled sandy-gravel columns(Gibbs and Davis 1976). Biodegradation rates may be in-creased by five to 10 times by nutrient additions (Bragg etal. 1994; Atlas 1995; Shelton et al. 1999). The time requiredfor hydrocarbon-using microbes to begin to degrade substan-tive quantities of hydrocarbons in beach substrate may re-quire roughly one to three weeks (Gibbs and Davis 1976).Degradation rates for a given microbial population can be farless at low temperatures than at high temperatures (e.g.,ZoBell 1969; Gibbs and Davis 1976; Atlas 1981). Experi-mental conditions used by Carls et al. (1999) were less fa-vorable for microbial growth than those by Middaugh et al.(1996, 1998) and Shelton et al. (1999): temperatures werecolder, (4–7°C versus 20°C), water flow was rapid (5–6 L·min–1 versus static), and nutrients were not added.Dividing the potential contribution of metabolic byproducts(<30%; estimated from Shelton et al. 1999) by 3–5 to ac-count for slower degradation rates without nutrient supple-ment (Atlas 1995) and by 3.7 to account for coldertemperatures (estimated from Gibbs and Davis 1976), we es-timate that metabolic byproducts accounted for <2–3% ofthe toxicity reported by Carls et al. (1999).

Prediction of oil effects in PWS herring eggs-on-kelp usinglaboratory results was restricted to 1989 because the primaryoil reservoir, PWS water, was essentially uncontaminated inspawn areas after 1989 (Fig. 1), except at two islands whereintertidal sediment had been heavily oiled. Total PAH con-centrations in mussels at Cabin Bay and Rocky Bay fell in1989 and remained at baseline in the following years(Figs. 1 and 4). EVO was not verifiable in 47 of 48 mussel

samples collected from oiled spawn sites in 1990 (Short etal. 1996). Biologically available EVO was plausible in some(7 of 13) mussel samples at Green Island (0.02–4.8µg·g–1)and Smith Island (0.07–5.3µg·g–1), sites that received spawnin 1990 but not in 1989; however, samples were too weath-ered and the source was verifiable only once (w = 7.70 atGreen Island, Short et al. 1996). These areas received a rela-tively small proportion of the total spawn (roughly 6% atGreen Island, and 3% at Smith Island; our estimates fromBrady et al. (1991)). Because mussels are effective watersamplers, there is little reason to believe that aqueous PAHconcentrations were elevated a year or more after the spill atmost herring spawn sites, and Neff and Stubblefield (1995)reported little or no TPAH in PWS water in 1990 (Fig. 1).The 1990 herring eggs-on-kelp data are perplexing becausethere is continued evidence of elevated TPAH concentra-tions at both oiled and reference sites in PWS (Pearson et al.1995), yet concentrations in eggs did not correlate withthose in mussels at sites in common geographically and tem-porally (r 2 = 0.19, P = 0.562,n = 4). The composition ofthis contamination in PWS and at Sitka Sound referencesites is unknown to us because the data were not published(Pearson et al. 1995). Available evidence suggests <10% ofthe herring eggs were exposed to EVO in 1990. Thus, appli-cation of laboratory results (Carls et al. 1999) to PWS datais not advisable beyond 1989 and the lack of embryo re-sponse in 1990 (Hose et al. 1996) suggests contaminant hy-drocarbon composition was not the same as in 1989.

Differences in industry and NRDA perspectives explainconflicting conclusions

Some of the differences in industry and NRDA embryo-larval studies may be explained by intensity of effort. Theparallel studies in 1989 lead by Pearson et al. (1995) (indus-try) and McGurk (1992) (NRDA) both initially reported mi-nor impacts. Industry studies stopped at this point, butNRDA studies were extended. Hose et al. (1996) examinedlarvae hatched in the McGurk study for genetic and morpho-logical damage. Field efforts to directly examine PWS her-ring adults and larvae were also added (McGurk and Brown1996; Marty et al. 1997, 1999). These extended studies dem-onstrated that significant damage occurred in embryolarvaefrom oiled areas compared with those from reference areas.In their update of McGurk (1992), McGurk and Brown(1996) reported that differences in growth and mortalityrates between oiled and nonoiled areas supported the hy-pothesis of injury to herring embryos and larvae. The philos-ophy behind the science was apparently different; industryinitially found minor effects, stopped testing, and concludedthe spill caused little damage, but the NRDA group was notconvinced that initial equivocal results could be interpretedas not harmful, continued testing, and ultimately concludedthat there were major spill effects.

The effects of the EVO spill on herring eggs extended be-yond the restricted areas where oil was in contact with eggs.Pearson’s group concludes that oil must adhere to (or coat)herring eggs to cause effects (Pearson et al. 1985, 1995,1999), but this perspective conflicts with the large body ofliterature indicating significant oil toxicity in the absence ofcoating (e.g., Moore and Dwyer 1974; Neff and Anderson

© 2002 NRC Canada

Carls et al. 165



1981; Grimmer 1983). Hay et al. (1995) provide a link be-tween coating and chemical toxicity effects. They report thatdirect contact with oiled substrate invariably killed eggs, butbecause oiled substrate negatively impacted many eggs eventhough most were not in direct contact with oil, the primaryexposure mechanism must have been via contaminated incu-bation water (Hay et al. 1995). Dissolved hydrocarbons areclearly toxic to fish embryos (e.g., Mironov 1967; Kuhnhold1974; Carls et al. 1999). Our conclusion is that herring eggsin PWS were exposed primarily to dissolved oil in water andthat this was the principal route of toxicity after theExxonValdezspill.

Although both industry and NRDA researchers relate eggexposure to TPAH concentrations, the two groups draw upondifferent literature to interpret EVO toxicity and reach oppo-site conclusions concerning its toxic potential in PWS. In-dustry researchers estimated the toxic potential of dissolvedEVO using water-soluble and water-accommodated fractionresearch (Pearson et al. 1999; Neff et al. 2000). Aromaticcompounds present in short-term, unweathered water-solubleor water-accommodated fraction tests are typically domi-nated by single-ring compounds (e.g., Rice et al. 1979; Neffand Anderson 1981; Rice et al. 1987). Neff et al. (2000)conclude that “under mild conditions with little physicaldispersion of petroleum into the water column, volatilemonocyclic aromatic hydrocarbons dissolving from the sur-face slick would be the main contributors to any toxicity ob-served in water column organisms”. In their review, Pearsonet al. (1999) argue that “the main hydrocarbons present inthe water-soluble fraction include the low relative molecularmass PAHs that are responsible for most of the toxicity ofcrude oil (Rice et al. 1977; Neff and Stubblefield 1995)” andconclude that the WSF used in toxicity tests can be com-pared with the composition of the hydrocarbon fraction pres-ent in the PWS water column following the spill. Problemswith these assertions are that (i) oil and water in PWS didnot mix under calm conditions; (ii ) monoaromatic hydrocar-bons all evaporated rapidly from PWS; and (iii ) toxic highmolecular weight PAH were present in PWS in addition tolow molecular weight PAH. Three days after theExxonValdezspill, a three-day storm dispersed substantial quanti-ties of oil into the water and accelerated the evaporation pro-cess (Wolfe et al. 1994). By the time herring spawned,composition of oil in PWS did not resemble the WSFs be-cause monoaromatics had evaporated (Wolfe et al. 1994;Neff and Stubblefield 1995) and high molecular weight PAHwere present (Neff and Stubblefield 1995; Short and Harris1996a; Short and Heintz 1997). The technique used by Carlset al. (1999), passage of water through oiled gravel, pro-duced water with a PAH composition similar to that ob-served in PWS (naphthalenes through chrysenes), includingthe weathering patterns characterized by Short and Heintz(1997). Because high molecular weight PAH are more toxicthan low molecular weight PAH (Rice et al. 1977; Neff1979; Black et al. 1983), Carls et al. (1999) and Heintz et al.(1999) observed that PAH from EVO was about 1000 timesmore toxic than previously studied WSFs. Reliance of indus-try on traditional acute toxicity tests, which focus on narco-sis induced by the most water-soluble components of oil,does not explain the suite of toxic effects in embryonic lifestages caused by exposure to PAH, where the toxicity mech-

anism is oxidative cellular damage that can be passed on todaughter cells during development (Livingstone et al. 1990;Akcha et al. 2000). Thus, whereas industry concludes thatthe roughly 1µg·L–1 aqueous concentrations of EVO in PWSwere not toxic, NRDA researchers conclude that 1µg·L–1

aqueous concentrations were sufficient to have caused sig-nificant damage.

Reassessment of the fraction of herring eggs exposed toEVO in PWS

We reassessed the fraction of PWS herring eggs exposedto EVO in 1989 by determining toxic body burdens of EVOat various oiled sites and related them to those of industry(4–10% oiled, Pearson et al. 1995, 1999) and NRDA (40–52%, Brown et al. 1996a, 1996b) (Table 1). (Zaikof Bay wasincluded in the high estimate by Brown et al. (1996a,1996b), but excluded from the low estimate because of in-sufficient chemical analysis.) Earlier we noted that the in-dustry assessment was based on spawn distance, whereas theNRDA estimate was based on spawn biomass; the differencein estimation methods is <10%. We concur that biomass isthe more appropriate measure of egg exposure and reportonly this measurement in the following text. Evidence fromall portions of the Naked Island area suggested biologicallysignificant oiling, i.e., TPAH concentrations in mussels andherring eggs were high enough to cause developmental ab-normalities in Pacific herring embryos. Concentrations weresimilarly significant in the outer portion of Rocky Bay, al-though concentrations in the inner portion of the bay mayhave been below biological significance. (Total PAH concen-trations in mussels in the inner portion of the bay (generallywest of and including site O18 in Fig. 3) were less than the2.5 µg·g–1 level estimated to signal biological significance,but TPAH concentrations in water samples collected in theeastern portion of the inner bay were consistently above0.4µg·L–1, a level that caused adverse biological response inherring embryolarvae in laboratory tests (Carls et al. 1999).)We conclude that 25–30% of the spawn biomass was oiled atbiologically significant levels and that oiling was likely sig-nificant in an additional 2% of the spawn biomass atMontague Point. Oil was visually identified at MontaguePoint (Gundlach et al. 1990; Brady et al. 1991; Neff et al.1995), but not documented by chemical analysis. GraveyardPoint may have acted as a natural barrier to shield StockdaleHarbor from most of the oil slick, where evidence suggeststhat oil concentrations were too low to be detrimental to her-ring egg development (6% of spawn mass). There were in-sufficient data to determine if herring egg exposure wassignificant in Zaikof Bay (13% of spawn mass). Our esti-mate of the percentage of Pacific herring eggs exposed to bi-ologically meaningful amounts of EVO in 1989 (25–32% ofspawn mass) falls between the original estimates.

Identical oil effects were observed in older herringlarvae from oiled PWS sites

Pacific herring larvae from oiled areas of PWS were alsoadversely affected by EVO in 1989, and the causal relation-ship between oil exposure and detrimental effects was con-firmed through laboratory study. Major oil-associated effectsin larvae captured from oiled sites in spring 1989 includedsmall size, ascites, pericardial edema, delayed development,

© 2002 NRC Canada




and genetic damage (Marty et al. 1997). Microscopic lesionswere consistent with decreased growth and increased mortal-ity of herring larvae collected near oiled beaches (McGurkand Brown 1996) and were similar to those observed in lab-oratory studies of herring larvae exposed to ANSCO as eggs(e.g., Kocan et al. 1996; Marty et al. 1997; Carls et al.1999). Because of the low aqueous concentrations of hydro-carbons in PWS and variable environmental conditions, thelink between ascites and oil exposure in field-sampled larvaewas controversial (Pearson et al. 1999). Recent laboratorystudies, however, provide evidence that aqueous PAH con-centrations of 0.4–1µg·L–1 cause ascites and edema in her-ring exposed as developing eggs (Carls et al. 1999),concentrations well within documented TPAH levels inPWS in April and May 1989 (Neff and Stubblefield 1995;Short and Harris 1996a). Growth of older larvae from off-shore areas also decreased throughout PWS in 1989, but adirect link to oil exposure was not possible because theselarvae likely originated from a mixture of oiled and unoiledareas (Norcross et al. 1996). In contrast, the frequency ofgenetic defects was low and jaw size was within normal lim-its in PWS larvae six years after the spill (Norcross et al.1996). Pacific herring larvae that hatched in PWS were notstudied by industry. We conclude that NRDA informationon embryolarval effects, coupled with identical responsesobserved in the laboratory and in free-swimming herringlarvae, provides consistent documentation of oil toxicity in

early life stages of PWS herring in 1989. (Effects of EVOon juvenile herring were not evaluated by either group.)

Adult PWS herring accumulated hydrocarbons andexhibited histopathological changes

In 1989, adult Pacific herring in oiled areas of PWS ex-hibited significant oil-associated lesions and evidence of hy-drocarbon exposure, but fish from reference sites did not(Moles et al. 1993; Marty et al. 1999). When the oil spilloccurred, herring were beginning to congregate in shallowbays for their annual mass spawning in April. Fish fromoiled sites had hepatic necrosis and elevated PAH concentra-tions (primarily naphthalenes) in their tissues (Marty et al.1999). Naphthalenes were also preferentially accumulated inmuscle tissue in laboratory exposures of adult herring toPAH in water (Carls et al. 2000), evidence of metabolism(Thomas et al. 1997) and differential uptake. Herring fromoil-exposed areas had fewer nematode parasites in their bodycavities than did fish from reference sites (apparently be-cause stressed nematodes migrated into muscle tissue) andthe potential link to acute oil exposure was confirmedthrough laboratory study (Moles et al. 1993). In a recentstudy, wild-caught Pacific herring exposed to crude oil hadhigher prevalence of hepatic necrosis, increased mortality,and viral hemorrhagic septicemia virus (VHSV) was isolatedfrom exposed fish but not control fish (Carls et al. 1998).Also, aqueous PAH concentrations in the range of those

© 2002 NRC Canada

Carls et al. 167

Spawn Percent

Oiling Distance (km) Mass (metric tons) Based on distance Based on mass

Industry estimatea

Moderate–heavy 0.9 NE 1 NEVery light–light 5.2 NE 3 NENone 152.2 NE 96 NE

Total 158.3NRDA estimateb,c

Oiledd 54.7 20 886 35 40Zaikof Bay 13.8 5 998 9 12Not oiled 89.8 25 351 57 48

Total 158.3 52 235ReassessmentSignificante 30.6–38.1 13 230–15 911 19–24 25–30Likely significantf 2.7 984 2 2Unknowng 19.0 6 823 12 13Likely not significanth 8.8–16.2 3 167–5 847 6–10 6–11None 89.8 25 351 57 48

Total 158.3 52 235

Note: Published Natural Resource Damage Assessment (NRDA) estimations were based both on spawn distance and eggmass; the industry estimate was based only on spawn distance. NE, not estimated.

aPearson et al. (1995a).bBrown et al. (1996a).cBrady et al. (1991).dBrown et al. (1996b) stated range of eggs exposed to oil was 41–52%, based on spawn mass.eIncludes the Naked Island area plus Rocky Bay and Graveyard Point.fOiling at Montague Point was likely biologically significant and most of the spawn overlapped visibly oiled shoreline

(Neff et al. 1995), but TPAH (total polynuclear aromatic hydrocarbon) concentrations were not documented by chemicalanalysis.

gInsufficient data were collected in Zaikof Bay to either support or refute exposure of Pacific herring eggs in 1989.hThere were indications of slight oiling in Stockdale Harbor, but the weight of evidence suggests concentrations were

probably too low to damage herring eggs.

Table 1. Reassessment of percentages of Pacific herring eggs exposed toExxon Valdezcrude oil in PrinceWilliam Sound after the spill and comparison to previous estimates.



measured in PWS in spring 1989,≥0.6 µg·L–1, cause immu-notoxicity in rainbow trout (Karrow et al. 1999). Althoughsignificant mortality of adult herring in 1989 as a result ofthe spill was never documented in PWS, lesions in fish sam-pled from oiled sites in 1989 were consistent with lesions infish from which VHSV was isolated in more recent studies(Marty et al. 1998, 1999), and a small fraction of the popula-tion probably died as a result of this viral outbreak. In 1990and 1991, fish sampled from oiled sites had neither oil-related lesions nor significant PAH concentrations in theirtissue (Marty et al. 1999).

Long-term spill consequences were indirect or notassociated with the spill

Adult biomass in PWS was at historically high levels in1989 and continued at high levels through 1992. During thisperiod, the spawning population was composed mostly offish that were adults during the 1989 oil spill because PWSherring first spawn when they are 3 years old.

Contrary to predictions made before the spring of 1993,the adult Pacific herring population of PWS declined morethan 75% from an estimated 11 × 107 kg in 1992 to 1.7 ×107 kg in 1994, and most of the decline occurred during thewinter of 1992–1993 (Marty et al. 1998). High prevalence ofulcers in surviving fish in spring 1993 provided strong evi-dence that the population decline was associated with a dis-ease outbreak. Although natural variation in clupeidpopulation size can be large and unpredictable (Blaxter andHunter 1982; Cole and McGlade 1998), recruitment failuredid not explain the collapse (fish from all year classes wereinvolved), and the magnitude of the 1992–1993 collapsesuggests external factors played a critical role.

Several hypotheses have been advanced to explain thepopulation collapse including both spill-related and naturalreasons, and combinations of the two. These hypotheses arediscussed in detail by Pearson et al. (1999) and Carls et al.(2001a) and have not been repeated here. Our perspective isthat important stressors responsible for the collapse of theadult herring population in 1993 included large populationsize, poor overwinter fish condition, and possibly environ-mental conditions, but links between the EVO spill anddelayed population response are tenuous. The herring popu-lation in PWS may have approached or exceeded carryingcapacity between 1988 and 1992, thus the risk of a diseaseepizootic was high. Although the major agent in the epi-zootic, VHSV, can be induced by acute exposure to oil andcan cause rapid mortality in infected adult herring, delayedreaction of herring to oil toxicity in 1993 is unlikely. The es-timated additional increase in population in 1989 resultingfrom the postspill fishery closure was small, but the precipi-tous 1993 collapse of the population near its carrying capac-ity underscores the importance of formulating resourcemanagement decisions that integrate knowledge of the his-torical fishery. Populations at carrying capacity that are ex-posed to additional stress, such as oil, may be at greater riskthan populations below carrying capacity, in addition to indi-rect effects such as postspill management decisions. Envi-ronmental conditions unrelated to the oil spill, such asprewinter prey availability and winter starvation, may havealso been important contributing factors. Probably no singlefactor can completely explain the population collapse.

Summary and conclusionsReassessment of response of Pacific herring to oil in PWS

suggests that observed impacts could not have been pre-dicted from routine toxicological monitoring methods in useat the time of the spill. This review emphasizes the need tothoroughly study pollutant exposure, adequately characterizepollutant composition, estimate toxic effects using appropri-ate laboratory models, critically examine sensitive life stagesover appropriate time intervals, and place the findings withinbroad ecological perspectives. That the PWS oil spill oc-curred in a nearly pristine environment (Karinen et al. 1993)yields much information that can be advantageously used insituations complicated by previous habitat degradation. Ad-verse biological responses were observed orders of magni-tude below expected responses in part because the spillchemistry had not been accurately predicted, in part becausesensitive early life stages were exposed during critical devel-opment periods, and because sufficient funding and contro-versy ensured intensive study. Because oil typically persistsfor at least several days after a spill, the research paradigmshould shift away from an emphasis on mechanisms of acutetoxicity (such as narcosis) to long-term toxicity (e.g., oxida-tive cellular damage). As a result of this reassessment, werecommend that safety standards for dissolved PAH shouldbe revised to reflect a new toxicity threshold of <1µg·L–1

(part per billion) TPAH to adequately protect aquatic organ-isms and habitat. Assessment of risk should also considerpopulation dynamics (e.g., a population at carrying capacitymay be at greater risk than one below capacity), season (e.g.,spill effects are more deleterious when they impact criticalearly life stages), location of the spill (such as spawninggrounds), local ecology, hydrographic conditions, and large-scale ecological processes. Results of the manyExxonValdez studies have broad applicability to other situations,such as the combined industrial and non-point-source runoffthat results in elevated aquatic PAH loads near urban areas(Rice et al. 2001).

We do not recommend that routine monitoring of hydro-carbons in other species (such as mussels) be substituted fordirect measurements in the species of interest (e.g., herringeggs). For this particular spill we made use of the data avail-able, and the close correspondence between TPAH concen-trations in herring eggs and mussels allowed us to verify theassertions of NRDA researchers that mussels were adequateproxies for direct measurement in herring eggs. However, wehave used the mussel data with caution and based our con-clusions regarding adverse reaction of PWS herring embryosto oil principally on observed TPAH concentrations in thoseeggs, responses to similarly low TPAH concentrations in ex-perimentally exposed herring eggs, and directly observed ef-fects in PWS larvae. That TPAH concentrations in herringeggs and mussels were correlated suggests that correlationsmay be found in other spills, but the nature of such correla-tions may depend on many factors, such as mixing energyand duration, temperature, salinity, oil viscosity, and organiccontent of receiving water (Rice et al. 1977; Gustafson andDickhut 1997), and thus may not be consistent among oilspills. Mussels, which are filter feeders, preferentially accu-mulate particulate or colloidal oil (Short and Harris 1996b;Axelman et al. 1999), but the majority of exposed herringeggs in PWS likely responded to dissolved oil. Particulate

© 2002 NRC Canada




© 2002 NRC Canada

Carls et al. 169

and dissolved hydrocarbon concentrations can be related inthat the large surface area presented by small particles willaccelerate dissolution (Short and Heintz 1997). However,because petroleum hydrocarbons are sparingly soluble, con-tact time is also very important, and when contact time isshort, the correlation between particulate and dissolved com-pounds may be substantially different than when longer con-tact times are involved (Axelman et al. 1999). In general,routinely monitored “indicator” species (e.g., mussels) canidentify likely environmental problem areas, but should notsimply be substituted for direct measurements in the speciesof interest.

Differing perspectives between industry and NRDA re-searchers led to disparate conclusions concerning the conse-quences of theExxon Valdezspill to PWS herring in 1989.Because initial results were equivocal and environmental oilconcentrations (roughly 1µg·L–1 aqueous TPAH) were belowpreviously published toxicity thresholds, industry researchersstopped testing and concluded that the spill caused little dam-age. In contrast, NRDA researchers were unconvinced thatinitial results could be interpreted as harmless and continuedtheir study, experimentally determining that 1µg·L–1 aqueousTPAH concentrations can cause embryo abnormalities andmortality and ultimately concluded that short-term spill ef-fects were major.

This reassessment shows that the short-term consequencesof the spill were detrimental to herring in PWS, but that pos-sible long-term consequences of the spill are difficult todiscern. Results from these analyses have refined the nomi-nal classification of visually oiled spawn areas using site-specific chemical evidence. Because TPAH concentrations inmussels were correlated with TPAH in herring eggs (justify-ing their use as a surrogate index of exposure by NRDA re-searchers), we inferred that herring eggs from almost allspawning sites within the oil trajectory in 1989 were ex-posed to EVO in the water column. The incubation ofherring eggs approximately coincided with peak aqueoushydrocarbon concentrations in PWS and the bioaccumulatedoil was detrimental to 25–32% of the spawned biomass. Her-ring larvae collected from oiled PWS areas throughoutspring 1989 were malformed, small in size, and had highermortality rates than in reference areas, consistent with fieldmeasurements of EVO above laboratory toxicity thresholds.Embryolarval toxicity appeared to be limited to 1989 and theadult herring biomass continued at high levels through 1992.The 1993 population collapse was likely caused by a combi-nation of factors including high population size, disease, andsuboptimal nutrition, but indirect links to the spill cannot beruled out.

Acknowledgements

We thank Jeep Rice, Jeff Short, Mike Murphy, Bob Thomas,and anonymous reviewers for their constructive criticism ofthis manuscript, and theExxon ValdezOil Spill TrusteeCouncil for their support. However, the findings and conclu-sions presented by the authors are their own, and do not nec-essarily reflect the view or position of the Trustee Council.

References

Akcha, F., Izuel, C., Venier, P., Budzinski, H., Burgeot, T., and

Narbonne, J-F. 2000. Enzymatic biomarker measurement andstudy of DNA adduct formation in benzo[a]pyrene-contaminatedmussels,Mytilus galloprovincialis. Aquat. Toxicol. 49: 269–287.

Anderson, J.W., Neff, J.M., Cox, B.A., Tatem, H.E., and Hightower,G.M. 1974. Characteristics of dispersions and water-soluble ex-tracts of crude and refined oils and their toxicity to estuarine crus-taceans and fish. Mar. Biol.27: 75–88.

Armstrong, D.A., Dinnel, P.A., Orensanz, J.M., Armstrong, J.L.,McDonald, T.L., Cusimano, R.F., Nemeth, R.S., Landolt, M.L.,Skalski, J.R., Lee, R.F., and Huggett, R.J. 1995. Status of se-lected bottomfish and crustacean species in Prince WilliamSound following theExxon Valdezoil spill. In Exxon Valdezoilspill: fate and effects in Alaskan Waters.Edited byP.G. Wells,J.N. Butler, and J.S. Hughes. American Society for Testing andMaterials, Philadelphia, Pa. ASTM STP 1219. pp. 485–547.

Atlas, R.M. 1981. Microbial degradation of petroleum hydrocar-bons: an environmental perspective. Microbiol. Rev.45: 180–209.

Atlas, R.M. 1995. Petroleum biodegradation and oil spill bio-remediation. Mar. Pollut. Bull.31: 178–182.

Axelman, J., Næs, K., Näf, C., and Broman, D. 1999. Accumula-tion of polycyclic aromatic hydrocarbons in semipermeablemembrane devices and caged mussels (Mytilus edulisL.) in rela-tion to water column phase distribution. Environ. Toxicol.Chem.18: 2454–2461.

Barron, M.G., and Ka’aihue, L. 2001. Potential for photoenhancedtoxicity of spilled oil in Prince William Sound and Gulf ofAlaska waters. Mar. Pollut. Bull.43: 86–92.

Barron, M.G., Podrabsky, T., Ogle, S., and Ricker, R.W. 1999. Arearomatic hydrocarbons the primary determinant of petroleumtoxicity to aquatic organisms? Aquat. Toxicol.46: 253–268.

Bence, A.E., and Burns, W.A. 1995. Fingerprinting hydrocarbonsin the biological resources of theExxon Valdezspill area.In Ex-xon Valdezoil spill: fate and effects in Alaskan Waters.Editedby P.G. Wells, J.N. Butler, and J.S. Hughes. American Societyfor Testing and Materials, Philadelphia, Pa. ASTM STP 1219.pp. 84–140.

Bienert, R.W., and Pearson, W.H. 1995. Application of ecologicalrisk assessment principles to questions of oil exposure in Pacificherring following theExxon Valdezspill. Can. Tech. Rep. Fish.Aquat. Sci. No. 2060. pp. 219–225.

Black, J.A., Birge, W.J., Westerman, A.G., and Francis, P.C. 1983.Comparative aquatic toxicology of aromatic hydrocarbons.Fundam. Appl. Toxicol.3: 353–358.

Blaxter, J.H.S., and Hunter, J.R. 1982. The biology of the clupeoidfishes. Adv. Mar. Biol.20: 1–223.

Brady, J., Morstad, S., Simpson, E., and Biggs, E. 1991. PrinceWilliam Sound area annual finfish management report 1989. Re-gional Information Report No. 2C90-07. Alaska Department ofFish and Game, Cordova, Alaska.

Bragg, J.R., Prince, R.C., Harner, E.J., and Atlas, R.M. 1994. Ef-fectiveness of bioremediation for theExxon Valdezoil spill. Na-ture (London),368: 413–418.

Brown, E.D., Baker, T.T., Hose, J.E., Kocan, R.M., Marty, G.D.,McGurk, M.D., Norcross, B.L., and Short, J. 1996a. Injury tothe early life history stages of Pacific herring in Prince WilliamSound after theExxon Valdezoil spill. Am. Fish. Soc. Symp.18:448–462.

Brown, E.D., Norcross, B.L., and Short, J.W. 1996b. An introduc-tion to studies on the effects of theExxon Valdezoil spill onearly life history stages of Pacific herring,Clupea pallasi, inPrince William Sound, Alaska. Can. J. Fish. Aquat. Sci.53:2337–2342.

Carls, M.G., Marty, G.D., Meyers, T.R., Thomas, R.E., and Rice,



S.D. 1998. Expression of viral hemorrhagic septicemia virus inpre-spawning Pacific herring (Clupea pallasi) exposed to weath-ered crude oil. Can. J. Fish. Aquat. Sci.55: 2300–2309.

Carls, M.G., Rice, S.D., and Hose, J.E. 1999. Sensitivity of fish em-bryos to weathered crude oil. Part 1. Low level exposure during in-cubation causes malformations, genetic damage and mortality inlarval Pacific herring (Clupea pallasi). Environ. Toxicol. Chem.18:481–493.

Carls, M.G., Hose, J.E., Thomas, R.E., and Rice, S.D. 2000. Expo-sure of Pacific herring to weathered crude oil: assessing effectson ova. Environ. Toxicol. Chem.19: 1649–1659.

Carls, M.G., Marty, G.D., and Hose, J.E. 2001a. Synthesis of thetoxicological and epidemiological impacts of theExxon Valdezoilspill on Pacific herring in Prince William Sound, Alaska.ExxonValdezOil Spill Restoration Project 99328 Final Report. NationalOceanic and Atmospheric Administration, National Marine Fish-eries Service, Auke Bay Laboratory, Juneau, AK 99801, U.S.A.

Carls, M.G., Babcock, M.M., Harris, P.M., Irvine, G.V., Cusick,J.A., and Rice, S.D. 2001b. Persistence of oiling in mussel bedsafter theExxon Valdezoil spill. Mar. Environ. Res.51: 167–190.

Cole, J., and McGlade, J. 1998. Clupeoid population variability,the environment, and satellite imagery in coastal upwelling sys-tems. Rev. Fish. Biol. Fish.8: 445–471.

Floodgate, G.D. 1984. The fate of petroleum in marine ecosystems.In Petroleum microbiology.Edited by R.M. Atlas. MacmillanPublishing Company, New York. pp. 355–396.

Gibbs, C.F., and Davis, S.J. 1976. The rate of microbial degrada-tion of oil in a beach gravel column. Microb. Ecol.3: 55–64.

Grimmer, G. 1983. Environmental carcinogens: polycyclic aromatichydrocarbons. Chemistry, occurrence, biochemistry, carcinogenic-ity. CRC Press, Inc., Boca Raton, Fla.

Gundlach, E.R., Bauer, J., Bayliss, R., Provant, S., and Kendziorek,M. 1990. Response to theExxon Valdezoil spill by the AlaskaDepartment of Environmental Conservation.In Oil spills man-agement and legislative implications: proceedings of the confer-ence, 15–18 May 1990, Newport, R.I., U.S.A.Edited byM.L.Spaulding and M. Reed. American Society of Civil Engineers,New York. pp. 471–478.

Gustafson, K.E., and Dickhut, R.M. 1997. Distribution of polycyclicaromatic hydrocarbons in southern Chesapeake Bay surface water:evaluation of three methods for determining freely dissolved waterconcentrations. Environ. Toxicol. Chem.16: 452–461.

Hay, D.E., McCarter, P.B., and Grant, D. 1995. Nestucca oil spilland impacts on herring eggs.In Proceedings of the Seventh Pa-cific Coast Herring Workshop, 27–28 January 1994.Edited byD.E. Hay and P.B. McCarter. Can. Tech. Rep. Fish. Aquat. Sci.No. 2060. pp. 95–99.

Heintz, R., Short, J.W., and Rice, S.D. 1999. Sensitivity of fish em-bryos to weathered crude oil. Part II. Increased mortality of pinksalmon (Oncorhynchus gorbuscha) embryos incubating down-stream from weatheredExxon Valdezcrude oil. Environ. Toxicol.Chem.18: 494–503.

Heintz, R.A., Rice, S.D., Wertheimer, A.C., Bradshaw, R.F., Thrower,F.P., Joyce, J.E., and Short, J.W. 2000. Delayed effects on growthand marine survival of pink salmonOncorhynchus gorbuschaafterexposure to crude oil during embryonic development. Mar. Ecol.Prog. Ser.208: 205–216.

Hose, J.E., McGurk, M.D., Marty, G.D., Hinton, D.E., Brown, E.D.,and Baker, T.T. 1996. Sublethal effects of theExxon Valdezoil spillon herring embryos and larvae: morphologic, cytogenetic, andhistopathological assessments, 1989–1991. Can. J. Fish. Aquat.Sci. 53: 2355–2365.

Karinen, J.F., Babcock, M.M., Brown, D.W., MacLeod, W.D., Jr.,Ramos, L.S., and Short, J.W. 1993. Hydrocarbons in intertidal

sediments and mussels from Prince William Sound, Alaska,1977–1980: characterization and probable sources. NOAA Tech.Mem. NMFS-AFSC-9.

Karrow, N.A., Boermans, H.J., Dixon, D.G., Hontella, A., Solomon,K.R., Whyte, J.J., and Bols, N.C. 1999. Characterizing theimmunotoxicity of creosote to rainbow trout (Oncorhynchusmykiss): a microcosm study. Aquat. Toxicol.45: 223–239.

Kocan, R.M., and Hose, J.E. 1997. Laboratory and field observa-tions of sublethal damage in marine fish larvae: lessons from theExxon Valdezoil spill. In Chemically induced alterations infunctional development and reproduction of fishes. Proceedingsfrom a session at the Wingspread Conference Center, 21–23 July1995, Racine, Wis.Edited byR.M. Rolland, M. Gilbertson, andR.E. Peterson. Society of Environmental Toxicology and Chem-istry (SETAC), Pensacola, Fla. pp. 167–176.

Kocan, R.M., Hose, J.E., Brown, E.D., and Baker, T.T. 1996. Pa-cific herring (Clupea pallasi) embryo sensitivity to Prudhoe Baypetroleum hydrocarbons: laboratory evaluation and in situ expo-sure at oiled and unoiled sites in Prince William Sound. Can. J.Fish. Aquat. Sci.53: 2366–2375.

Kuhnhold, W.W. 1974. Investigations on the toxicity of seawater-extracts of three crude oils on eggs of cod (Gadus morhuaL.).Ber. Dtsch. Wiss. Komm. Meeresforsch.23: 165–180.

Livingstone, D.R., Kirchin, M.A., and Wiseman, A. 1989. Cyto-chrome P450 and oxidative metabolism in molluscs. Xenobiotica,19: 1041–62.

Livingstone, D.R., Garcia Martinez, P., Michel, X., Narbonne, J.F.,O’Hara, S., Ribera, D., and Winston, G.W. 1990. Oxyradicalproduction as a pollution-mediated mechanism of toxicity in thecommon mussel,Mytilus edulisL., and other molluscs. Funct.Ecol. 4: 415–424.

Mackay, D., Buist, I., Mascarenhas, R., and Paterson, S. 1980. Oilspill processes and models. Report to Research and Develop-ment Division, Environment Emergency Branch, EnvironmentalImpact Control Directorate, Environmental Protection Service,Environment Canada, Ottawa, Ont.

Marty, G.D., Hose, J.E., McGurk, M.D., and Brown, E.D. 1997.Histopathology and cytogenetic evaluation of Pacific herring lar-vae exposed to petroleum hydrocarbons in the laboratory or inPrince William Sound, Alaska, after theExxon Valdezoil spill.Can. J. Fish. Aquat. Sci.54: 1846–1857.

Marty, G.D., Freiberg, E.F., Meyers, T.R., Wilcock, J., Farver, T.B.,and Hinton, D.E. 1998. Viral hemorrhagic septicemia virus,Ichthyophonus hoferi, and other causes of morbidity in Pacificherring Clupea pallasi spawning in Prince William Sound,Alaska, U.S.A. Dis. Aquat. Org.32: 15–40.

Marty, G.D., Okihiro, M.S., Brown, E.D., Hanes, D., and Hinton,D.E. 1999. Histopathology of adult Pacific herring in PrinceWilliam Sound, Alaska, after theExxon Valdezoil spill. Can. J.Fish. Aquat. Sci.56: 419–426.

McGurk, M.D. 1992. Petroleum effects on herring eggs and larvaein Prince William Sound, Alaska, 1989. Conference proceedingsof the Fourth Information Transfer Meeting, U.S. Department ofthe Interior, Minerals Management Service, Alaska Outer Conti-nental Shelf (OCS) Region, Anchorage, Alaska. OCS StudyMMS 92-0046. pp. 119–124 and 128–130.

McGurk, M.D., and Brown, E.D. 1996. Egg–larval mortality of Pa-cific herring in Prince William Sound, Alaska, after theExxonValdezoil spill. Can. J. Fish. Aquat. Sci.53: 2343–2354.

McGurk, M.D., Warburton, D., Parker, T., and Litke, M. 1990.Early life history of Pacific herring: 1989 Prince WilliamSound herring egg incubation experiment. Final report, pre-pared for NOAA/NOS/OMA/OAD, Anchorage, Alaska. Con-tract No. 50ABNC-7-00141.

© 2002 NRC Canada




Middaugh, D.P., Chapman, P.J., and Shelton, M.E. 1996. Responsesof embryonic and larval inland silversides,Menidia beryllina, to awater-soluble fraction formed during biodegradation of artificiallyweathered Alaska North Slope crude oil. Arch. Environ. Contam.Toxicol. 31: 410–419.

Middaugh, D.P., Shelton, M.E., McKenney, C.L., Jr., Cherr, G.,Chapman, P.J., and Courtney, L.A. 1998. Preliminary observa-tions on responses of embryonic and larval Pacific herring,Clupea pallasi, to neutral fraction biodegradation products ofweathered Alaska North Slope crude oil. Arch. Environ. Contam.Toxicol. 34: 188–196.

Mironov, O.G. 1967. The effect of small concentrations of oil andoil products on developing eggs of the Black Sea turbot. Trans-lation by Division of Foreign Fisheries, NMFS, NOAA, fromAkademiya Nauk SSSR. Vopr. Ikhtiol.7: 44.

Moles, A.D., Rice, S.D., and Okihiro, M.S. 1993. Herring parasiteand tissue alterations following theExxon Valdezoil spill. InProceedings of the 1993 Oil Spill Conference, American Petro-leum Institute (API), Washington, D.C. 20005. API PublicationNo. 4580. pp. 325–328.

Moore, S.F., and Dwyer, R.L. 1974. Effects of oil on marine or-ganisms: a critical assessment of published data. Water Res.8:819–827.

NRC (National Research Council). 1980. The international musselwatch. National Research Council, Washington, D.C.

Neff, J.M. 1979. Polycyclic aromatic hydrocarbons in the aquaticenvironment. Sources, fates, and biological effects. Applied Sci-ence Publishers Ltd., London.

Neff, J.M., and Anderson, J.W. 1981. Response of marine animals topetroleum and specific petroleum hydrocarbons. Halsted PressDivision, John Wiley and Sons, New York.

Neff, J.M., and Burns, W.A. 1996. Estimation of polycyclic aro-matic hydrocarbon concentrations in the water column based ontissue residues in mussels and salmon: an equilibrium partition-ing approach. Environ. Toxicol. Chem.15: 2240–2253.

Neff, J.M., and Stubblefield, W.A. 1995. Chemical and toxicologi-cal evaluation of water quality following theExxon Valdezoilspill. In Exxon Valdezoil spill: fate and effects in Alaskan Wa-ters.Edited byP.G. Wells, J.N. Butler, and J.S. Hughes. Ameri-can Society for Testing and Materials, Philadelphia, Pa. ASTMSTP 1219. pp. 141–177.

Neff, J.M., Owens, E.H., Stoker, S.W., and McCormick, D.M. 1995.Shoreline oiling conditions in Prince William Sound followingthe Exxon Valdezoil spill. In Exxon Valdezoil spill: fate and ef-fects in Alaskan Waters.Edited byP.G. Wells, J.N. Butler, andJ.S. Hughes. American Society for Testing and Materials, Phila-delphia, Pa. ASTM STP 1219. pp. 312–346.

Neff, J.M., Ostanzeski, S., Gardiner, W., and Stejskal, I. 2000. Ef-fects of weathering on the toxicity of three offshore Australiancrude oils and a diesel fuel to marine animals. Environ. Toxicol.Chem.19: 1809–1821.

Norcross, B.L., Hose, J.E., Frandsen, M., and Brown, E.D. 1996.Distribution, abundance, morphological condition, and cyto-genetic abnormalities of larval herring in Prince William Sound,Alaska, following the Exxon Valdezoil spill. Can. J. Fish.Aquat. Sci.53: 2376–2387.

O’Clair, C.E., Short, J.W., and Rice, S.D. 1996. Contamination ofintertidal and subtidal sediments by oil from theExxon ValdezinPrince William Sound. Am. Fish. Soc. Symp.18: 61–93.

Page, D.S., Boehm, P.D., Douglas, G.S., and Bence, A.E. 1995.Identification of hydrocarbon sources in the benthic sedimentsof Prince William Sound and the Gulf of Alaska following theExxon Valdezoil spill. In Exxon Valdezoil spill: fate and effectsin Alaskan Waters.Edited byP.G. Wells, J.N. Butler, and J.S.

Hughes. American Society for Testing and Materials, Philadel-phia, Pa. ASTM STP 1219. pp. 41–83.

Payne, J.R., Clayton, J.R., Jr., McNabb, G.D., Jr., and Kirstein, B.E.1991.Exxon Valdezoil weathering fate and behavior: model pre-dictions and field observations.In Proceedings of the 1991 Inter-national Oil Spill Conference (prevention, behavior, control,cleanup), 4–7 March 1991. American Petroleum Institute Publica-tion No. 4529, Washington, D.C. pp. 641–654.

Pearson, W.H., Woodruff, D.L., Kiesser, S.L., Fellingham, G.W.,and Elston, R.A. 1985. Oil effects on spawning behavior and re-production in Pacific herring(Clupea harengus pallasi). Finalreport to the American Petroleum Institute, Environmental Af-fairs Department, Washington, D.C.

Pearson, W.H., Moksness, E., and Skalski, J.R. 1995. A field andlaboratory assessment of oil spill effects on survival and repro-duction of Pacific herring following theExxon Valdezspill. InExxon Valdezoil spill: fate and effects in Alaskan Waters.Edited byP.G. Wells, J.N. Butler, and J.S. Hughes. AmericanSociety for Testing and Materials, Philadelphia, Pa. ASTM STP1219. pp. 626–661.

Pearson, W.H., Elston, R.A., Bienert, R.W., Drum, A.S., andAntrim, L.D. 1999. Why did the Prince William Sound,Alaska, Pacific herring (Clupea pallasi) fisheries collapse in1993 and 1994? Review of hypotheses. Can. J. Fish. Aquat.Sci. 56: 711–737.

Pelletier, M.C., Burgess, R.M., Ho, K.T., Kuhn, A., McKinney,R.A., and Ryba, S.A. 1997. Phototoxicity of individual polycyclicaromatic hydrocarbons and petroleum to marine invertebrate lar-vae and juveniles. Environ. Toxicol. Chem.16: 2190–2199.

Rice, S.D., Short, J.W., and Karinen, J.F. 1977. Comparative oiltoxicity and comparative animal sensitivity.In Fate and effectsof petroleum hydrocarbons in marine organisms and ecosystems.Edited byD.A. Wolfe. Pergamon Press, New York. pp. 78–94.

Rice, S.D., Moles, A., Taylor, T.L, and Karinen, J.F. 1979. Sensi-tivity of 39 Alaskan marine species to Cook Inlet crude oil andNo. 2 fuel oil. In Proceedings of the 1979 Oil Spill Conference(prevention, behavior, control, cleanup), 19–22 March 1979.American Petroleum Institute Publication No. 4308, Washing-ton, D.C. pp. 549–554.

Rice, S.D., Babcock, M.M., Brodersen, C.C., Gharrett, J.A., andKorn, S. 1987. Uptake and depuration of aromatic hydrocarbonsby reproductively ripe Pacific herring and the subsequent effect ofresidues on egg hatching and survival.In Pollution physiology ofestuarine organisms.Edited byW.B. Vernberg, A. Calabrese, F.P.Thurberg, and F.J. Vernberg. Belle W. Baruch Libr. Mar. Sci. 17,University of South Carolina Press, Columbia, S.C. pp. 139–154.

Rice, S.D., Thomas, R.E., Carls, M.G., Heintz, R.A., Wertheimer,A.C., Murphy, M.L., Short, J.W., and Moles, A. 2001. Impactsto pink salmon following theExxon Valdezoil spill: persistence,toxicity, sensitivity, and controversy. Rev. Fish. Sci.9: 165–211.

Shelton, M.E., Chapman, P.J., Foss, S.S., and Fisher, W.S. 1999.Degradation of weathered oil by mixed marine bacteria and thetoxicity of accumulated water-soluble material to two marinecrustacea. Arch. Environ. Contam. Toxicol.36: 13–20.

Short, J.W., and Babcock, M.M. 1996. Prespill and postspill con-centrations of hydrocarbons in mussels and sediments in PrinceWilliam Sound. Am. Fish. Soc. Symp.18: 149–66.

Short, J.W., and Harris, P.M. 1996a. Chemical sampling and analy-sis of petroleum hydrocarbons in near-surface seawater of PrinceWilliam Sound after theExxon Valdezoil spill. Am. Fish. Soc.Symp.18: 17–28.

Short, J.W., and Harris, P.M. 1996b. Petroleum hydrocarbons incaged mussels deployed in Prince William Sound after theEx-xon Valdezoil spill. Am. Fish. Soc. Symp.18: 29–39.

© 2002 NRC Canada

Carls et al. 171



© 2002 NRC Canada


Short, J., and Heintz, R. 1997. Identification ofExxon Valdezoil insediments and tissues of PWS. Environ. Sci. Technol.31: 2375–2384.

Short, J.W., Heintz, R.A., Nelson, B.D., Maselko, J.M., Kendziorek,M.F., Carls, M.G., and Korn, S. 1996.Exxon Valdezoil spill of1989: State/Federal Trustee Council hydrocarbon database 1989–1995. National Oceanic and Atmospheric Administration, Na-tional Marine Fisheries Service, Auke Bay, Alaska.

Smith, R.L, and Cameron, J.A. 1979. Effect of water soluble frac-tion of Prudhoe Bay crude oil on embryonic development of Pa-cific herring. Trans. Am. Fish. Soc.108: 70–75.

Stegeman, J.J. 1985. Benzo[a]pyrene oxidation and microsomal en-zyme activity in the mussel (Mytilus edulis) and other bivalvemollusc species from the western north Atlantic. Mar. Biol.89:21–30.

Thomas, R.E., Carls, M.G., Rice, S.D., and Shagrun, L. 1997. Mixedfunction oxygenase induction in pre- and post-spawn herring(Clupea pallasi) by petroleum hydrocarbons. Comp. Biochem.Physiol. C,116: 141–147.

Vandermeulen, J.H., and Penrose, W.R. 1978. Absence of aryl hy-drocarbon hydroxylase (AHH) in three marine bivalves. J. Fish.Res. Board Can.35: 543–647.

Vines, C.A., Robbins, T., Griffin, F.J., and Cherr, G.N. 2000. Theeffects of diffusible creosote-derived compounds on develop-ment in Pacific herring (Clupea pallasi). Aquat. Toxicol. 51:225–239.

Weis, J.S., and Weis, P. 1989. Effects of environmental pollutantson early fish development. Aquat. Sci.1: 45–73.

Wolfe, D.A., Clark, R.C., Jr., Foster, C.A., Hawkes, J.W., andMacLeod, W.D. 1981. Hydrocarbon accumulation and histo-pathology in bivalve mollusks transplanted to the Baie de Morlaixand the Rade de Brest.In Proceedings, 1979 international sympo-sium centre oceanologique de Bretagne, Brest, France. CentreNational Pour L’Exploitation des Oceans, Paris. pp. 599–616.

Wolfe, D.A., Hameedi, M.J., Galt, J.A., Watabayashi, G., Short, J.,O’Clair, C., Rice, S., Michel, J., Payne, J.R., Braddock, J., Hanna,S., and Sale, D. 1994. The fate of the oil spilled from theExxonValdez. Environ. Sci. Technol.28: 561A–8A.

ZoBell, C.E. 1969. Microbial modification of crude oil in the sea.In Proceedings of joint conference on prevention and control ofoil spills. American Petroleum Institute, Washington, D.C.pp. 317–326.



Documents

Synthesis of the toxicological impacts of the Exxon Valdez ... · M.G. Carls, G.D. Marty, and J.E. Hose Abstract: Pacific herring (Clupea pallasi) in Prince William Sound (PWS) were