Research was conducted at the University of Idaho (Moscow, ID, USA) on the toxicity of weathered Exxon Valdez crude oil to embryos of pink salmon from 2001 to 2003 for the purpose of comparing these data with those from the National Oceanic and Atmospheric Administration Fisheries Laboratory at Auke Bay (AK, USA). Mortality reported at Auke Bay for embryos chronically exposed to very low concentrations of aqueous solutions of weathered oil, measured as dissolved polycyclic aromatic hydrocarbons (PAHs), was inconsistent with that in other published research. Using the Auke Bay experimental design, we found that toxicity is not evident in pink salmon embryos until chronic exposure to laboratory weathered and naturally weathered oil concentrations exceeding 1,500 and 2,250 ppm, respectively, representing a total PAH tissue burden in excess of 7,100 ppb. Effluent hydrocarbons also drop well below concentrations sufficient to cause harm over the time frame of a few weeks, regardless of oiling level. Resolution of differences with Auke Bay involved the source of contributing hydrocarbons. The experimental design did not exclude dispersed oil droplets from the aqueous solution; thus, toxicity was not limited to the dissolved hydrocarbon fraction. The implications of the present results are discussed regarding the toxic risk of weathered oil to pink salmon embryos in streams of Prince William Sound (AK, USA).
The 1989 Exxon Valdez spill of 11 million gallons of North Slope Alaska crude oil into Prince William Sound (PWS; AK, USA) posed a threat to the wild pink salmon (Oncorhynchus gorbuscha) that run in the area. Up to 70% of wild-run pink salmon use the tidally influenced areas of streams entering PWS for spawning . Eggs incubating in tidal reaches of streams in the spill path (14% of salmon streams affected, or ∼3% of total PWS pink salmon eggs ) may have been vulnerable to oil that reached the shoreline. Consequently, efforts were made to assess the impacts of the spill on incubating pink salmon [3–9]. However, no effects of oil on incubating pink salmon embryos (i.e., deformities or mortality) were found in streams immediately after the spill (spring of 1989) , and no lingering effects (genetic damage or reduced adult returns) were observed years later [8,10,11].
Spill-related research conducted at the National Oceanic and Atmospheric Agency Fisheries Laboratory in Auke Bay (ABL; AK, USA) arrived at different conclusions based on the results of laboratory studies [5–7,12]. They hypothesized that the toxicity of weathered Exxon Valdez crude oil (EVC) to salmon embryos was caused solely by dissolved polycyclic aromatic hydrocarbons (PAHs) that partitioned from oiled substrate into the water. Pink salmon eggs exposed directly to the oiled gravel or to aqueous effluent from columns containing artificially weathered oil on gravel in the laboratory had significantly higher mortality and developmental abnormalities compared to controls at concentrations of dissolved PAH in column effluents ranging from 1 to 18 ppb (μg/L), depending on the degree of weathering of the oil on gravel. A very weathered oil (VWO) that had been stored under uncontrolled conditions for at least a year apparently was the most toxic. Exposure concentrations were reported as total PAH (TPAH) concentrations dissolved in the column effluent. These concentrations are two to three orders of magnitude lower than the petroleum TPAH concentrations reported previously by ABL researchers as being lethal to pink salmon embryos . Toxicity was attributed to dissolved three- through five-ring PAHs, particularly alkyl phenanthrenes, and were extrapolated to the field under the premise that pink salmon embryos and larvae in tidally influenced stream gravels of the spill zone may have been exposed to toxic concentrations of dissolved PAHs washing into the stream from adjacent oiled shoreline sediments by tidal flushing and hydraulic gradients [6,14]. The alleged mechanism was called the interstitial toxic water hypothesis .
The purpose of the present study was to assess independently the toxicity of weathered EVC to incubating pink salmon. The inconsistency between the ABL results that have alleged continuing risk to pink salmon in PWS and the field investigations showing that only low, nontoxic concentrations of oil reached the spill-path salmon streams [8,9] requires reanalysis of weathered oil toxicity to ascertain the exposure risk to incubating pink salmon.
MATERIALS AND METHODS
This investigation was performed in the fall of 2001 and 2003 at the Aquaculture Research Institute Laboratory, University of Idaho (UI). The experimental design was based on that described by Marty et al.  and Heintz et al. . The VWO used in the ABL study was of uncertain composition, and no attempt was made to duplicate that treatment.
Pink salmon eggs were obtained from the Washington Department of Fish and Wildlife, Hoodsport Hatchery, on Hood Canal (WA, USA) and from the Bellingham Hatchery at Bel-lingham (WA, USA). On September 18, 2001, eggs were spawned from 52 females, thoroughly mixed, fertilized with sperm from 14 males, and water-hardened in Hoodsport freshwater. The fertilized eggs were transferred to the test laboratory in buckets containing 11°C freshwater. The transfer from the hatchery to the test laboratory required 8 h. The eggs were placed in the experimental incubators and exposed to test conditions within 2 h of arrival in the laboratory.
On October 9, 2003, eggs were spawned from four females, pooled, and fertilized with milt from two males. This time, eggs were water-hardened in aqueous effluents from the oiled generator columns from the UI laboratory and transferred in containers to the laboratory within 8 h.
Oil toxicity tests were performed at the UI laboratory with two sources of EVC collected in 1989 and stored in our laboratory in a sealed, refrigerated container until use in these studies. The first source was fresh EVC removed from the Exxon Valdez at the time of the spill and artificially weathered in our laboratory by heating to 70°C until the mass of oil was reduced to 80%. This was the method used by Marty et al.  and Heintz et al.  to weather EVC artificially. The preparation is referred to in the present paper as artificially weathered EVC (AWEVC).
The second oil source tested was naturally weathered EVC (NWEVC) collected after the spill from the shoreline of PWS. The NWEVC was mousse, or water-in-oil emulsion, and was subjected to no treatments before use. Both oil sources were divided into precisely weighed aliquots for use in preparing oil-on-gravel substrates to be placed in the generator columns described below.
Incubation system for eggs and alevins
The incubation exposure system consisted of vertical columns (length, 60 cm) made from polyvinyl chloride pipe (diameter, 15 cm) after the design described by Marty et al.  and containing oil-contaminated gravel. These generator columns were equipped with inlet and outlet ports at the bottom and top, respectively. Clean, fresh, oxygenated well water at a measured flow rate of 150 ml/min moved upward through the oiled gravel, exiting the column outlet at the top. The effluent water was carried away from the column through polyethylene tubing. For reasons explained below, the UI experiments did not employ the alternating cycles of fresh water and saltwater as used in the design described by Heintz et al. .
Gravel with a grain size of 0.5 to 3 cm was washed to remove smaller grain sizes and then dried. A preweighed amount of AWEVC or NWEVC was added to 10.8 kg of gravel in a rotating mixer and mixed for 5 min to coat the gravel with oil as uniformly as possible. The mixing sequence was performed in order from lowest to highest concentration to minimize the risk of cross-contamination of exposure substrates. The oiled gravel was spread on a plastic sheet and left for 24 h in the open air, then placed in the incubation column and flushed for 4 d with clean freshwater at a flow rate of 150 ml/min before introduction of the eggs. Both AWEVC and NWEVC were tested at four nominal EVC-on-gravel loadings (200, 750, 1,500, and 2,250 ppm [mg/kg dry wt]), with a clean gravel control, and all tests run in triplicate. Effluent samples were first collected at 4 d after the water flow was initiated, immediately after the eggs were introduced (day 1). Tissue and effluent subsequently were sampled on days 11, 23, 40, and 83. An oil-on-gravel mixture without eggs also was maintained for each of the exposure concentrations. Oiled-gravel samples were collected from these columns at the same time the effluent was sampled and then analyzed to monitor PAH concentrations in the incubation substrate.
Embryo exposure regime
Approximately 250 eggs were placed at each of two locations in each incubator on September 18, 2001. The conditions included, first, eggs in direct contact with oil-contaminated gravel, duplicating the ABL direct-exposure regime, and, second, eggs suspended on a screen above the gravel in effluent flushing through oil-contaminated gravel, similar to the indirect exposure regime tested at ABL. Because each of the two lots of salmon eggs occupied incubation containers at different locations in the effluent stream and, therefore, could experience different exposure concentrations and flow patterns, each test location had its own control against which PAH exposure conditions could be compared.
The present study was performed in darkness (<0.1 lumens) except when the eggs were examined with a flashlight each day for mortality. Temperatures and flow rates were monitored daily during incubation. Temperature remained at 12°C, and flows were adjusted to 150 ml/min as needed. Dead embryos were removed when the eggs were loaded into the incubators and when they were examined daily after the embryos had finished epiboly. Dead embryos were examined for anomalies.
Effects of oil on gametes and newly fertilized eggs
We also exposed eggs to AWEVC during sperm activation and the water-hardening process to evaluate possible effects of differences in exposure conditions between the ABL protocol  and conditions in the field during the fall of 2003. Twelve lots of fertilized pink salmon eggs were removed from an egg/milt pool mixture of several females and males before activation. Each lot was placed in a separate jar, in sets of three, containing the water from the effluent stream of the columns filled with oil-on-gravel concentrations of 0, 750, 1,500, and 2,250 ppm of oil. Sperm activation and water-hardening of the eggs were initiated immediately in the jars. All 12 lots remained in the water-hardening solutions during the 8-h transport to the test laboratory and, subsequently, were placed in separate incubators irrigated only with clean freshwater. Mortality was monitored through the alevin swim-up stage.
All samples were sent to Battelle Laboratories (Duxbury, MA, USA) for chemical analysis. Water samples were sealed in precleaned glass containers and shipped to the analytical laboratory in chilled coolers. Gravel and eggs were shipped frozen, and they remained frozen until analyzed within 60 d of arrival at Battelle. Samples of NWEVC and AWEVC collected at the initiation of the present study and of water, oiled gravel, and salmon embryos collected at different times during incubation were analyzed for total and individual PAHs. Oil samples were prepared for analysis by dilution in methylene chloride to achieve a nominal concentration of approximately 5 mg/ml. The dilutions were spiked with PAH standards. One-liter water samples were extracted with liquid/liquid techniques . Fifty-gram samples of gravel were extracted by a modification of the ambient temperature solvent agitation method as described by Brown et al.  and modified by Peven and Uhler  (http://www.ntis.gov/) for the U.S. National Status and Trends Mussel Watch Program. An additional aliquot of approximately 5 g of gravel was removed, weighed, and dried to a constant weight at 105°C to determine sample dry weight.
Salmon eggs and alevins were extracted by maceration in solvent at ambient temperature by methods described by Peven and Uhler . An additional aliquot of approximately 1 to 2 g of eggs was removed, weighed, and dried to a constant weight at 105°C to determine the sample dry weight.
The salmon egg extracts were processed through an alumina clean-up column to isolate a combined saturated and aromatic hydrocarbon fraction . The resulting extracts were fractionated further by high-performance liquid chromatography/gel permeation chromatography with a Phenomonex phenogel column (22.5 × 300 mm, pore size of 100 Å pore, with a 7.8- × 50-mm precolumn; Phenomonex, Torrance, CA, USA)  to remove saturates, fats, and lipids. The PAH fraction of the eluate was collected, concentrated under a stream of nitrogen, and spiked with PAH internal standards before analysis.
All sample extracts were analyzed for target parent and alkyl PAH by capillary column gas chromatography/mass spectrometry (U.S. Environmental Protection Agency method 8270 modified) operated in the selective ion-monitoring mode as described by Boehm et al.  and Page et al. . Target analytes included naphthalene through benzo[ghi]perylene and the C1- through C3- or C4-alkyl homologues of naphthalenes, fluorenes, phenanthrenes/anthracenes, dibenzothiophenes, fluoranthenes/pyrenes, and chrysenes [19,20]. Concentrations of PAH target analytes were quantified versus the internal standards added to the extracts before analysis, with average response factors generated from a five-point calibration curve of the parent PAH (i.e., nonalkyl-substituted PAH). The concentrations of the alkyl PAH were determined from the response factors of the corresponding parent PAH. The alkyl PAHs were quantified by a straight baseline integration of each alkyl homologue series. The selection criteria for the alkyl homologue series were based on retention time, the pattern recognition compared to an EVC standard, and the presence of primary and confirmatory ions and ion ratios. Surrogate compound recoveries were calculated to monitor the recovery of the target PAH throughout the analytical scheme.
Analytical quality-assurance/quality-control measures followed a written laboratory quality-assurance plan and standard operating procedures as well as internal data-auditing procedures. Analytical chemistry quality control consisted of analysis of procedural blanks, standard reference materials, laboratory control samples, and an EVC standard. Detailed data-quality objectives were monitored for surrogate recoveries, spiked-compound recoveries, standard reference materials, crude oil standards, five-point calibrations, and continuing calibrations. The specific data-quality objectives and procedures have been described in detail by Boehm et al.  and Page et al. .
We tested for effects of position in the column and effluent and of concentration on mortality rates (p.dead) with a two-way analysis of variance:
The grand mean proportion dead is μ. Two positions were used: In gravel and above gravel in the effluent. Nominal concentrations were control (0), 200, 750, 1,500, and 2,250 ppm of EVC on gravel. The p symbol signifies the interaction of position and concentration. Error was binomial, because eggs were recorded as either alive or dead . When blue sac disease (ascites) became apparent, the proportion of alevins infected was similarly tested with the equation. Contrasts were used to test equal mortality between control and concentrations greater than zero. Artificially weathered EVC and NWEVC were analyzed separately.
Fate of PAH
Initial TPAH concentrations were a factor of 2.5 to 8.7 higher on gravel containing AWEVC than on gravel containing NWEVC at comparable nominal column loadings of EVC (Table 1). Total PAH concentrations in gravel ranged from 460 to 12,100 ppb on AWEVC-coated gravel and from 53 to 4,400 ppb on NWEVC-coated gravel at the start of the egg incubation period. The difference in TPAH concentrations in AWEVC and NWEVC was shown to be the high water content of NWEVC (64.5% by wt).
The TPAH concentrations in gravel decreased with time during the egg incubation period at all loadings of oil on gravel. Loss of PAH from the gravel for both AWEVC and NWEVC was most rapid early in the exposure period, with relatively little additional loss during the last 60 d of the exposure. The PAH loss rate decreased with decreasing oil loading of the gravel.
Nonweathered EVC contains approximately 13,000 ppm (μg/g; 1.3% by wt) of total resolved PAHs, 96% of which are two- and three-ring PAHs and dibenzothiophenes (sulfur heterocyclics), including their alkyl homologues . Nearly 50% of the total resolved PAHs are naphthalene and C1- through C4-alkyl naphthalenes. Most high-molecular-weight PAHs, such as the carcinogens benzo[k]fluoranthene and ben-zo[a]pyrene, are present at trace concentrations, often below the method detection limits. As with all crude oils, the alkyl homologues are more abundant than the parent compounds. The low-molecular-weight PAHs remained the most abundant fraction in the gravel PAH profile for the duration of the exposure period (Fig. 1). (See Appendix1 for identification of chemical abbreviations.)
Over the time course of the present study, both AWEVC and NWEVC weathered on the gravel in the generator columns (Fig. 1), as indicated by depletion of C1- through C4-naphtha-lenes, phenanthrenes, and dibenzothiophenes. A greater loss of low-molecular-weight, two- and three-ring PAHs was found compared with that of high-molecular-weight, four- and five-ring PAHs, as shown by PAH profiles in the gravel (Fig. 1) and the effluent (Fig. 2). The concentration of total naphthalenes (two-ring PAHs) on gravel containing 2,250 ppm of NWEVC decreased by 65% during the 83-d exposure period; the concentration of total chrysenes (four-ring PAHs) decreased by 32% in the same time period.
Table Table 1.. Cumulative percentage mortality and blue sac disease in incubating pink salmon exposed to initial concentrations of total polycyclic aromatic hydrocarbons (TPAHs) from Exxon Valdez crude oil (EVC) mixed onto dry column gravel and effluent water from oil-on-gravel concentrations of artificially weathered EVC (AWEVC) or naturally weathered EVC (NWEVC)
Day 0 exposure concentration
EVC on gravel (ppm)
Gravel TPAH (ppb)a
Water TPAH (ppb)b
c M = total % mortality.
d SD = standard deviation.
e BS = blue sac deviation.
f Significantly from control mortality (p = 0.05).
The TPAH concentrations in the column effluent water at the initiation of the salmon egg exposure period were directly related to the oil loading on the gravel (Table 1). Effluent TPAH concentrations declined rapidly during the exposure period. Total PAH effluent concentrations dropped to less than 1 ppb (1,000 ng/L) in all column effluents within 20 d (Fig. 2). For the two lowest loading levels (200 and 750 ppm of oil), effluent TPAH concentration approached that in control effluent water (∼0.01–0.02 ppb) within 40 d (Tables 1 and 2). At the higher loadings (1,500 and 2,250 ppm of oil on gravel), TPAH concentrations in effluent approached the background (control) level by the end of the exposure at day 83.
The TPAH concentrations at the start of the exposure were approximately twice as high in effluents from columns containing AWEVC as in those containing NWEVC at all loading levels tested (Table 1). The highest day 1 effluent TPAH concentrations were 16.4 ppb from the column containing 2,250 ppm of AWEVC and 8.27 ppb from the column containing 2,250 ppm of NWEVC. By day 11, TPAH concentrations had declined to 2.24 and 1.62 ppb, respectively, in effluents from these two columns.
The PAH profiles in effluent water showed that the dominant PAH in the water collected during the first 15 (4 + 11) days of column irrigation were low-molecular-weight PAHs, which is consistent with a predominantly dissolved PAH fraction (Fig. 2). Although the concentrations were low, the relative contributions of low-solubility, high-molecular-weight PAHs increased in the effluent water after day 15. These high-molecular-weight PAHs were the dominant components of the PAH assemblage in effluent water samples collected on days 40 and 83 of egg exposure.
At the highest concentration (2,250 ppm) of AWEVC oil on gravel, we observed oil residue on the inside walls of the incubator columns above the gravel and on the walls of the tubing downstream in the effluent, which indicated that oil droplets were entrained in the water passing through the oil-contaminated gravel. The presence of oil droplets was confirmed by observation of small sheen spots on the incubator water surface, which indicated that the droplets were minute and remained in the effluent stream. If oil droplets were the source of the toxicity in these short-term column experiments, then they would not be expected to be introduced into PWS pink salmon streams through tidal flushing of adjacent gravels in the years following the spill.
Tissue TPAH concentration in salmon embryos
Salmon embryo tissue TPAH concentration increased to maximum before 40 d of exposure in all exposure scenarios and then declined to near pre-exposure concentrations by day 83. The time to maximum tissue TPAH concentration varied depending on the initial TPAH concentration in the effluent.
Embryos exposed directly to the gravel with the lowest oil loading (200 ppm) accumulated PAH slowly for up to 40 d. Embryos located in the effluent water in all oil loadings showed increased PAH accumulation for 11 to 23 d and declined thereafter. Embryos exposed to the effluent from the columns containing AWEVC accumulated higher concentrations of PAH than did those exposed to NWEVC, reflecting the higher TPAH concentrations in AWEVC.
Maximum tissue TPAH concentrations increased with increasing oil loading on the column gravel and TPAH concentrations in the column effluents (Table 2). In general, for loadings of AWEVC or NWEVC on gravel, embryos positioned in direct contact with oiled gravel in the columns accumulated slightly more TPAH than did embryos exposed to the effluent above the gravel. Tissue PAH concentrations were similar at both locations, because oil droplets contaminated the effluent and coated the eggs. Consequently, with these data, it is not possible to differentiate PAH from residual oil coating the egg surfaces from the soluble fraction assimilated in embryo tissues.
The pattern of maximum relative concentrations of different PAHs in embryo tissues was roughly similar to the PAH pattern in the gravel, but not to the PAH composition in the effluent at the times tissue samples were collected (Figs. 1–3). At day 23 of egg exposure, the effluent was depleted of parent and less-alkylated, low-molecular-weight PAHs consisting of the naphthalenes, fluorenes, phenanthrenes, and dibenzothio-phenes relative to the four- to six-ring, high-molecular-weight PAHs. All low-molecular-weight PAHs were depleted by day 83, with primarily the fluoranthenes/pyrenes and chrysenes remaining (Fig. 2). In the case of the PAH profiles in tissue (Fig. 3), the low-molecular-weight PAHs remained the most abundant PAHs for the duration of the exposure period. These findings are consistent with those reported by ABL .
Table Table 2.. Cumulative percentage mortality and comparative bioaccumulation of total polycyclic aromatic hydrocarbons (TPAHs) from Exxon Valdez crude oil (EVC) in incubating pink salmon during exposure to gravel and aqueous effluents from gravel columns containing weathered EVC and different concentrations of TPAHs in Heintz et al.  and the present study
Gravel EVC (ppm)
Gravel TPAH (ppb)
Water TPAH (ppb)
Tissue TPAH (ppb)
a AWEVC = artificially weathered EVC.
c VWO = very weathered oil.
dNWEVC = naturally weathered EVC (collected in 1989).
Alkyl naphthalenes, phenanthrenes, and dibenzothiophenes were the most abundant PAHs in tissues of embryos exposed to the column effluents at both day 23 and day 83 (Fig. 3). Alkyl fluorenes were about as abundant as alkyl phenanthrenes and dibenzothiophenes in embryo tissues on day 83. Small amounts of high-molecular-weight PAHs also were present in the embryo tissues at days 23 and 83.
Dose effects of AWEVC on pink salmon embryos
Low mortality was observed during the first portion of the incubation period among embryos exposed to the control, hydrocarbon-free effluent and to effluents from columns with the four AWEVC loadings, even at the highest oil-on-gravel concentrations. However, cumulative mortality increased during the exposure period to an overall mean of 23.1% of the embryos and alevins exposed directly to 1,500 ppm of AWEVC on gravel and 59.7% mortality at 2,250 ppm of oil on gravel, compared with 16.9% mortality among controls (Table 1).
Blue sac disease was observed in the present study, with a higher incidence among embryos exposed to effluents from oiled columns than among controls and a significantly higher incidence at 1,500 and 2,250 ppm (p < 0.05), respectively, for both on- and above-gravel samples (Table 1). The alevins in direct contact with the oil on gravel showed the highest incidence, which implies that direct contact with the oil concentration causes stress, at least at the higher concentrations . The alevins above gravel showed less incidence of the disease, but that was true of the controls as well, which implies that the disease was accentuated by location in the incubation stream. Blue sac disease was still present in some control and experimental alevins at the end of the study.
The temporal pattern and final cumulative mortality of embryos and alevins suspended in the effluent stream above the gravel and at all oil-on-gravel loadings in the columns were not significantly different from those among embryos and alevins in direct contact with the oil on gravel (p = 0.922). This indicates that the compositions and concentrations of the toxic fractions of dissolved and dispersed AWEVC were approximately the same at the different exposure locations (in gravel and above gravel). At both positions, eggs exposed to effluent containing 2,250 ppm of AWEVC experienced more than 40% mortality, which is significantly higher than mortality among controls (p < 0.05). Mortality (12.6–23.1%) among embryos and alevins exposed continuously to effluents from columns containing the three lower loadings of AWEVC was not statistically different from control mortality (16.9%) (Table 1).
Dose effects of NWEVC on pink salmon embryos
Initial TPAH exposure concentrations from NWEVC columns ranged from 53 to 4,400 ppb in gravel and from 0.84 to 8.27 ppb in water (Table 1). No significant differences were found in mean mortality among embryos located at the different positions in the effluent stream (p = 0.797). Percentage mortality among embryos, even those exposed to effluents from the highest concentration of NWEVC on gravel (2,250 ppm; 14.8%), was not significantly different from that of controls (16.9%). Blue sac disease was observed in alevins exposed to NWEVC (Table 1) but was significantly higher only at the highest oil concentration (2,250 ppm).
Effects of oil exposure on pink salmon eggs during fertilization
Pink salmon eggs exposed to effluent from columns containing three concentrations of AWEVC for only 8 h during activation, fertilization, and water-hardening experienced mortality that was slightly higher than that of control eggs. Mortality of eggs exposed to effluent from AWEVC concentrations of 750, 1,500, and 2,250 ppm on gravel ranged from 16.2 to 18.5%. These percentages were not statistically different from their control (14.8%).
The current UI research, which was undertaken to evaluate the toxicity of weathered oil from the Exxon Valdez oil spill to pink salmon embryos under laboratory conditions, provides important results for assessing the effects of oil on pink salmon eggs and alevins in PWS spill-path streams. It also provides a comparative basis to elucidate the differences in the ABL study results from those reported in the field at the time of the spill and immediately thereafter [3,8,23].
Effects of weathered oil on pink salmon embryos in the UI study
In the present investigation, pink salmon embryo mortality significantly greater than that for controls only occurred during direct exposure to oiled gravel and effluents from columns containing 2,250 ppm of AWEVC on gravel (Table 1). Initial TPAH exposure concentration resulting in mortality was 12,100 ppb on gravel. In contrast, the NWEVC removed from the shore of PWS did not result in salmon embryo mortality significantly greater than that for controls at any NWEVC loading on gravel (Table 1). Initial TPAH concentrations at the highest initial NWEVC loading were 4,400 ppb on gravel and 8.27 ppb in column effluents. The NWEVC used in the present study, like much of the weathered EVC on the shores of PWS [24,25], was a mousse containing more than 60% water by weight. Mousse is a water-in-oil emulsion containing water droplets that are small enough (5–20 μm) to remain mixed in the oil indefinitely . The dispersion of water in the oil increases its viscosity, thus decreasing the partitioning of petroleum hydrocarbons into ambient water . These results demonstrate that oil reaching the shores of PWS in the form of mousse rendered it less toxic per unit mass than would be estimated from laboratory-weathered oil.
The relative PAH composition in embryo tissues remained fairly constant over the exposure period, despite the marked changes in the profile and concentrations of the aqueous PAH over the same time period (Figs. 2 and 3). The relationships among TPAH profiles in gravel, effluent, and tissue suggest that the initial high TPAH concentrations associated with eggs is from direct contact with oil, oil films, or droplets adsorbed on the chorion and may have partitioned directly from the oil phase into the hydrophobic chorion and yolk. Toxicity of dissolved TPAH alone could not be determined, because dispersed oil droplets were present in the effluents. The TPAH profile suggests that the dissolved fraction played a lesser role in oil toxicity to the embryos compared to the dispersed oil. If the PAH residues in salmon embryos were derived primarily or exclusively from dissolved PAHs in the column effluents, and given the lag in tissue uptake, tissue concentrations should have been in quasiequilibrium with concentrations in solution in the water [28,29], which was not the case. The data suggest, therefore, that the toxicity of weathered EVC in the present study was caused primarily by direct contact with oil or droplets that adhere to the eggs during the initial 11 to 20 d of exposure, with mortality gradually increasing in the developing embryo population over successive weeks.
Blue sac disease was observed in the present study, including the control lots, but the incidence was greatest at the higher concentrations of oil. The presence of the disease among the control lots indicates that oil only exacerbated the condition, but the mechanism could not be discerned. Pink salmon eggs and alevins incubated in hatcheries often show blue sac disease when exposed to poor water circulation, and in the present study, oil may have inhibited respiration across the yolk membrane.
Embryos exposed to the only toxic dose of AWEVC (2,250 ppm on gravel) bioaccumulated PAHs to a maximum concentration of 13,200 ppb (Table 2). Dominant PAHs in the salmon embryos were C2- and C3-naphthalenes, C1-phenanthrene, and C1-dibenzothiophene. Only traces of four- and five-ring PAHs were present in the embryo tissues. This suggests that low-molecular-weight PAHs were the main cause of mortality. Such a conclusion is consistent with the observation of Page et al.  that the toxicity threshold (based on laboratory bioassays with benthic amphipods) for NWEVC in intertidal sediments from oiled shorelines in PWS is 2,600 ppb of TPAH. Sediments containing greater than 2,600 ppb of TPAH and high relative concentrations of naphthalenes were more toxic compared with sediments containing greater than 2,600 ppb of TPAH and high relative concentrations of chrysenes. The high-molecular-weight PAHs, although more toxic than low-molecular-weight PAHs , were not present in bioavailable forms at sufficient concentrations in either the column effluents in the present study or in oiled sediments on the shores of PWS to cause chronic toxicity. Barron et al.  evaluated the results of the herring and pink salmon embryo toxicity studies performed at ABL and concluded that the toxicity of weathered EVC was caused primarily by alkyl phenanthrenes. Our results and those of Page et al.  indicate that alkyl naphthalenes also contributed to weathered EVC toxicity.
Comparative effects of AWEVC on pink salmon embryos
The results of the UI studies reported here and of the studies by Heintz et al.  differed in several ways. The cumulative mortalities of eggs and embryos in the UI study were much lower than those reported by Heintz et al. , even for the controls (16.9% vs 33.9%). Cumulative embryo mortality reported by Heintz et al.  was 51.4% at 281 ppm and 73.4% at 2,450 ppm on gravel with AWEVC weathered at ABL. In the present study, mortality significantly greater than that among controls occurred only at a concentration of 2,250 ppm oil on gravel with AWEVC weathered at UI (Table 1).
Initial AWEVC threshold tissue TPAH concentrations at which mortality was significantly greater than that among controls in the two laboratory studies were similar (ABL, =6,000 ppb; UI, >7.100 ppb) (Table 2), but not when based on TPAH loading on gravel (ABL, 3,800 ppb; UI, 12,100 ppb). Initial gravel TPAH concentrations in the ABL columns were well over twofold those in the UI columns at approximately the same oil-on-gravel loadings, and initial TPAH concentrations in the effluent were more than 2- to 12-fold higher in the ABL columns (Table 2). The most likely explanation for these differences is that the AWEVC used by ABL was fresher and less weathered than the AWEVC used by UI. As discussed below, the lightly weathered oil used by ABL contained substantially more naphthalenes than would be expected for a crude oil weathered to 80% of its original mass.
Total PAH represents 1.3% by weight of the AWEVC weathered at ABL at the 2,450 ppm loading on gravel reported by Heintz et al. , compared to 0.5% by weight of AWEVC in the 2,250 ppm loading on gravel in the present study. Fresh, nonweathered EVC contains approximately 1.2% TPAH by weight , and the TPAH concentration decreases with weathering. For example, EVC that had weathered in PWS for one year contained 0.47% TPAH by weight . The 2,450 ppm of AWEVC oil on gravel used by Heintz et al.  had a weathering parameter (w) of zero , indicating that the oil was essentially unweathered by that criterion. The weathering parameter increased progressively with decreasing AWEVC loadings on gravel in the ABL study, indicating that these lower doses were subject to water-washing in the 4 d between the initiation of column irrigation and the start of egg exposure.
Heintz et al.  stated that naphthalenes represented approximately 80% by weight of the TPAH fraction in the AWEVC used in the ABL studies, “in agreement with the composition of weathered EVC from PWS as reported by Bence and Burns” . However, Bence and Burns  reported that the naphthalenes comprise 83% by weight of the calculated equilibrium water-soluble fraction of fresh EVC. The PAH fraction of fresh EVC contains approximately 50% naphthalenes by weight, and the PAH fraction for the average of 71 shoreline oils collected from PWS one year after the spill is 16.6% naphthalenes by weight . Total naphthalenes represent nearly 48% of TPAH by weight in the AWEVC at an oil-on-gravel loading of 2,450 ppm in the ABL study, which is a value similar to that of fresh oil. Consequently, the PAH composition of the AWEVC employed by Heintz et al.  is not consistent with either NWEVC or EVC topped to 80% of its mass by heating to 70°C. In the present study, total naphthalenes represented 31% of TPAH by weight in the AWEVC at an oil-on-gravel loading of 2,250 ppm. The high concentration of naphthalenes in the ABL study may be part of the reason why their toxicity findings are inconsistent with the present results and with those of previous toxicity studies. Consequently, the ABL results should not be considered as indicative of the toxicity of weathered crude oil in the laboratory or in the field.
Comparative effects of VWO and NWEVC on pink salmon embryos
The VWO used by Heintz et al.  and the NWEVC used in the present study also are markedly different from one another. Although both contained approximately 0.2% TPAH by weight, the VWO contained no C0- through C3-naphthalenes, and the most abundant PAH group was the C3-phenanthrenes. The NWEVC contained substantial amounts of alkyl naphthalenes, and the C2-phenanthrenes were the most abundant group of alkyl phenanthrenes. Effluent from the NWEVC columns contained an eightfold-higher concentration of TPAH compared with the effluent from the VWO columns. The chemistry and concentrations indicate that the NWEVC was less weathered than the VWO. However, the VWO was reported to be toxic to salmon embryos, and the NWEVC was not.
Because the VWO was left over from an earlier ABL experiment and stored under uncontrolled conditions for at least a year before its use , uncertainty exists regarding both its composition and the toxicants that are present. The composition of the PAH assemblage in the VWO does not resemble that of NWEVC in PWS shoreline sediments. Embryos exposed directly to gravel and effluent from the column containing 2,860 ppm of VWO on gravel experienced elevated mortality when tissue TPAH reached 470 ppb . Embryos exposed directly to gravel in the column containing 2,250 ppm of NWEVC in the UI study contained 7,800 ppb of TPAH in their tissues but did not experience a mortality rate significantly greater than that of controls. Heintz et al.  attributed the high toxicity of the VWO to alkyl phenanthrenes. However, the embryos bioaccumulated a maximum of only approximately 118 ppb of alkyl phenanthrenes during exposure to 2,860 ppm of VWO on gravel, whereas embryos exposed to 2,250 ppm of NWEVC on gravel in the UI study bioaccumulated a maximum of 1,575 ppb of alkyl phenanthrenes. Therefore, bioaccumulated alkyl phenanthrenes could not have been the cause of elevated mortality among embryos exposed to VWO. Other factors associated with the VWO, such as inorganic by-products of oil biodegradation (ammonia and sulfide)  and oxidation products of microbial hydrocarbon degradation , appear likely to have contributed more than the PAHs to the observed toxicity.
This interpretation is supported by the relationship between tissue TPAH concentrations and mortality. If PAHs are causing the toxicity to salmon embryos, there should be good correlation between tissue TPAH concentration and the observed toxic effects. However, percentage mortality of salmon embryos that accumulated 470 ppb of TPAH during exposure to effluents from 2,860 ppm of VWO on gravel was similar to that among embryos that accumulated 71,000 ppb of TPAH during exposure to effluents from 2,450 ppm of AWEVC on gravel, and concentrations of alkyl phenanthrenes and chry-senes were lower in embryos exposed to VWO than in embryos exposed to AWEVC in the study by Heintz et al. .
Resolution of differences in the ABL and UI studies
The UI study showed by observation and hydrocarbon chemistry that eggs positioned above the oiled gravel, downstream in the effluent stream of the columns, were exposed to minute oil droplets as well as to dissolved PAHs. Chemical analysis of PAHs in effluent water for those experiments in which elevated embryo mortality was observed indicates that the effluent contained a mixture of dissolved and particulate oil, being particularly obvious during the later part of the exposure period. This is consistent with the results of Payne and Driskell , who determined that the sources of PAH (dissolved or oil droplet) could be determined by the tissue TPAH signatures. The presence of minute oil droplets in column effluents of similar study design also was disclosed through visual and chemical analysis by Pearson . Small oil droplets adsorb readily to the hydrophobic chorion of fish eggs  and contribute to the PAH body burdens in the embryos.
These results indicate that the 4 d of irrigation of the oil/gravel mixture in the columns before starting the UI study were insufficient to eliminate oil droplets from appearing in the column flow thereafter. In essence, the pink salmon eggs in the effluent were exposed to a mixture of both dispersed and dissolved oil rather than just to the dissolved fraction alone.
The problem of oil-droplet contamination in the water column is viewed differently by the ABL and UI researchers. We suggest that hydrocarbon concentrations diffusing across the egg membrane when in direct contact with oil are much higher than those of the dissolved fraction in the water around eggs not in contact with oil. Heintz et al.  assumed that even if the eggs were in direct contact with oil, the bioavailability and, thus, the toxicity was only in the form of the dissolved fraction measured in the water. That viewpoint was based on two oil loadings on gravel (74 and 717 ppm), for which bio-accumulation of TPAH was similar whether embryos were exposed just to dissolved PAH or were in direct contact with oil. Because the prestudy flushing of the gravel columns in the UI study was ineffective in preventing subsequent dispersal of oil, it cannot be assumed that the effluent in the Heintz et al.  study was free of minute oil droplets. Thus, the dissolved hydrocarbon levels measured in that study may not represent the actual TPAH concentrations experienced by the eggs.
Moreover, at all other oil concentrations tested by Heintz et al. , the eggs were not exposed just to the dissolved TPAH fraction but were actually in direct contact with oil-on-gravel mixtures. Therefore, rather than reporting dissolved TPAH concentrations, the more meaningful criteria is to use the total oil-on-gravel PAH concentrations with which the eggs were actually in contact. In the case of the Heintz et al.  study, the TPAH exposure was not just the dissolved 18 ppb (AW-EVC) and 1 ppb (VWO) but, rather, the oil-on-gravel TPAH concentrations in comparable units of 3,800 and 4,600 ppb, respectively.
Application of laboratory results to field conditions
Although egg mortality in PWS streams following the oil spill was reported by the Alaska Department of Fish and Game [3,4], the mortality subsequently was found to have been caused by physical shock of the eggs from sampling too early rather than from oil [9,39]. However, having overlooked those findings, Peterson et al.  maintained that oil was the cause of the reported long-term mortality. Because oil concentrations were too low in PWS streams to be toxic [8,14], it was proposed that the hypothetical source was the highly weathered oil deposits on the shorelines adjacent to pink salmon streams away from the flushing freshwater flows , a hypothesis that was felt to be supported in the laboratory by the allegedly highly toxic VWO at 1 ppb . The hypothetical transport mechanism was interstitial flow carrying the reputedly highly toxic dissolved hydrocarbons into the incubation substrate  and was referred to as the interstitial toxic water hypothesis .
Several conditions are apparent in the ABL and UI laboratory results, however, that would be inconsistent with the interstitial toxic water hypothesis being the process responsible for any egg mortality in the field. First, embryo death observed in the laboratory accumulated over the length of the study period, whereas the Alaska Department of Fish and Game reported that egg mortality in the stream occurred early in incubation, often within the first few days postspawning . Therefore, the pattern of mortality in the field did not follow the laboratory mortality time frame but was associated with a much more acute event.
Second, the embryos reportedly exposed to a lethal aqueous TPAH dose of 18 ppb for AWEVC and 1 ppb for VWO, as reported in the ABL study , were, in fact, incubated in direct contact with 280,000 and 2,860,000 ppb of oil on gravel, respectively, and exposed to gravel TPAH concentrations of 3,800 and 4,600 ppb, respectively. In contrast, following the spill in 1989, gravel TPAH concentrations in PWS streams did not exceed a mean of 267 ppb , or a level 14- to 17-fold less than the lethal doses reported at ABL.
Third, the VWO, which was the most toxic oil in the ABL study, did not resemble the NWEVC removed from the PWS shoreline and also was unlike the AWEVC used in either the ABL or the present study. Application of VWO data cannot apply to the conditions in oiled PWS streams.
Fourth, the naturally weathered oil removed from the PWS shoreline was nontoxic at laboratory-tested levels as high as 2,250 ppm of oil on gravel. The solubility of the toxic, high-molecular-weight PAH was very limited in the UI study, which implies a similar lack of bioavailability in the field.
Fifth, naturally weathered oil from the shore of PWS is not lethal to pink salmon embryos until tissue TPAH concentrations are in excess of 7,800 ppb, as reported here. Tissue PAH concentrations of embryos sampled from PWS oiled streams in 1990 and 1991 were 63 and 94 ppb, respectively . These levels are at least 80-fold lower than what is associated with toxic concentrations, indicating that interstitial toxic water was not experienced by eggs that were incubating at the height of oil presence on PWS beaches.
Sixth, using the ABL experimental design [5,6], the UI study demonstrated that oil droplets were not eliminated from the water that eggs were exposed to downstream of the oil-on-gravel substrate in the incubator columns. Therefore, attributing embryo mortality to the dissolved TPAH concentration in an effluent with oil droplets present would result in misinterpretation of the dissolved toxic dose.
In addition to the laboratory results, the biological and physical conditions in PWS pink salmon streams also eliminate the likelihood of interstitial toxic water. Biologically, spawning occurs in gravel flushed with freshwater from the surface stream during most stages of the tidal cycle . Physically, tidal dilution from the tons of marine water flushing oil deposits twice a day quickly exhaust any toxic potential of the deposits.
As shown by these experiments, as well as by those of Heintz et al. , effluent hydrocarbon toxicity levels drop well below those sufficient to cause harm to salmon embryos over the time frame of a few weeks, regardless of oiling level or extent of oil weathering. These data do not support claims of toxic exposure by salmon to hydrocarbons carried by tidal flushing years after the spill.
From the biological perspective of natural spawning in the intertidal reaches of the streams, the fertilization and water-hardening of the eggs, as well as the initial blastula formation, occur primarily in flowing freshwater, and exposure to aqueous PAHs is unlikely before tidal flooding. Therefore, to simulate accurately the field conditions during deposition, fertilization, and filling of the perivitelline space in the water-hardening process, the eggs should not be exposed to test concentrations of oil and saltwater until after initial water-hardening. Exposing eggs to high PAH concentrations during this early developmental phase would be atypical of field conditions and may exaggerate the effects of chronic oil exposure during the later stages of embryonic development.
Another laboratory scenario that is atypical of the field is the continuous exposure of eggs to PAH-contaminated effluent. Under field conditions, cycling between brackish water and freshwater occurs in the incubation environment. Dissolved or dispersed hydrocarbons in pore water, hypothetically seeping into salmon streams, would be associated primarily with the seawater phase. The oil-free freshwater phase, flushing the redds during receding tides and diluting saline pore water in the incubation environment, would create alternating cycles in the presence and absence of hydrocarbons and the leaching of hydrocarbons that may have adsorbed to redd substrates during saline water exposure. Therefore, toxic thresholds determined by chronic exposure to laboratory TPAH concentrations should not be applied to conditions of exposure in the field when the temporal variation and high diluting effects of tidal and freshwater flows under natural conditions were not represented in the laboratory.
The laboratory investigations undertaken at UI on weathered EVC, modeled after the ABL experimental design, helped to resolve inconsistencies between laboratory results on weathered oil toxicity and field investigations on pink salmon eggs and alevins conducted in PWS following the 1989 oil spill. Using AWEVC, we found that mortality of pink salmon eggs was not apparent until TPAH concentrations of oil on gravel exceeded 1,500 ppb or tissue TPAH levels exceeded 7,100 ppb. The incidence of blue sac disease was higher among alevins exposed to the higher oil concentration, which at least implies that the physical presence of oil may reduce respiration efficiency, apart from any toxicity. Analysis of the aqueous phase of the oil suggests that dispersed oil was not eliminated from the water assumed to carry only dissolved TPAH; thus, exposure to the column effluent water was not representative of only dissolved hydrocarbons. Consequently, determinations of aqueous TPAH toxicity made using this experimental design should be reconsidered.
The critical observation in the UI study was that chronic exposure of pink salmon embryos to NWEVC at concentrations as high as 2,250 ppm on gravel resulted in mortality no greater than that of laboratory controls. Nonlethal tissue burdens were as high as 7,800 ppb, and at least 80-fold higher than what was observed in embryos in 1990 and 1991 from PWS streams in the spill path.
The implications of these results in field conditions are clear. The demonstrated lack of toxicity of NWEVC, the limited solubility of the high-molecular-weight PAH, the high level of tidal and freshwater flushing of streams, and the extremely low tissue TPAH concentration in embryos actually removed from streams on oiled beaches following the spill clearly render the hypothesis of toxic interstitial waters un-tenable—and any claim of long-term effects years after the spill totally implausible. Laboratory results suggest that pink salmon eggs in spill-path salmon streams of PWS were not exposed to toxic concentrations of oil, whether dissolved or in direct contact, in sediments or in pore water infiltrating stream gravel from the adjacent shoreline.
Table . Polycyclic aromatic hydrocarbons and symbols