MARINE ECOLOGY PROGRESS SERIES Published August 20 Mar Ecol Prog Ser I

' Valdez' : impacts and recovery in the soft-bottom benthic community in and adjacent to eelgrass beds

Stephen C. Jewett1**,Thomas A. ~ean',Richard 0. Smith2, Arny Blanchardl

'Institute of Marine Science, School of Fisheries and Ocean Sciences, University of Fairbanks, Fairbanks, Alaska 99775, USA 'Coastal Resources Associates. Inc., 1185 Park Center Drive. Suite A, Vista, California 92083, USA

ABSTRACT We assessed impacts of the '' oil spill on bentluc communities within and adjacent to eelgrass beds in , Alaska, USA. The concentration of total polynuclear aromatic hydrocarbons (TPAHs),'benthic community composition. diversity, biomass, and abundance were compared between matched pairs of oiled and reference sites in 1990 (approx. 16 mo after the spill), andin 1991, 1993, and 1995. TPAHs in sediments were high (up to 15300 ng g.') at sites adjacent to oiled shorelines in 1990, but declined sharply thereafter. Some reference sites in 1990-91 also had elevated TPAHs in sediments and signatures matching Exxon Valdez oii, but concentrations were sig- nificantly lower than at oiled sites. Based on classification and ordination analyses, cornrnunities of infauna and small epifauna at some oiled sites in 1990 differed from communities at reference sites, and from the Same sites in subsequent years. Percent sand and mud aild concentration of total chrysenes (PAH analytes indicative of crude oil) explained significant proportions of the temporal and spatial vari- ation in benthic community structure. Total abundance and biomass of epifauna were generaliy higher at oiled sites, primarily because of higher densities of epifaunal bivalves. Otherwise, there were few consistent cornrnunity-wide responses to oiling in diversity. richness, total abundance. total biomass. or the abundances of major taxonomic groups (e.g. polychaetes or bivalves). We attnbute the lack of a stronger community-wide response to the varying sensitivities of constituent taxa to oil and organic enrichment. Over half of the dominant famiiies differed with respect to abundance at oiled versus ref- erence sites Most, including 9 families of polychaetes, were more abundant at oiled sites. Most of these were stress-tolerant or opportunistic. and their increase was likely due to organic enrichment. Negative impacts were most evident in oil-sensitive amphipods, especially the families Isaeidae and Phoxo- cephalidae. There were consistently more of these amphipods at reference sites, and abundances at oiled sites were likely reduced as a result of oil toxicity. Most of these differences between oiled and reference sites persisted through 1995,6 yr after the spill. We suspect that these differences are a result of the spill, but we rely on post-spill coinparisons to infer impacts, and our conclusions rely on the untestable assumption of equality between oiled and reference sites in the absence of a spill. Future assessments of the impacts of oil spiiis or other accidental environmental disturbances could benefit from pre-iinpact studies that provide objective critena for selection of matched pairs of sites, thereby supporting the assumption of equality in the absence of the disturbance.

KEYWORDS: 'Exxon Valdez' . Oil spill . Subtidal . Eelgrass . Benthos . Infauna - Prince Wdiiam Sound - Alaska

INTRODUCTION edge of oil impacts stems primanly from 2 sources: (1) laboratory studies of the toxicity of individual spe- Oil spills are major sources of localized disturbance cies in isolation; and (2) opportunistic field assessments in the marine environment and can have severe that examine the impacts of accidental spills. The for- impacts on nearshore biota (NRC 1985). Our knowl- mer provide detailed physiological and toxicological information about the impacts of oil, but are of limited value in predicting impacts of a given spill on biologi- cal communities. This is because (1) spills vary with

0 Inter-Research 1999 Resale of full article not permitted 60 Mar Ecol Prog Ser 185: 59-83, 1999

respect to the types an.d amount of oil released, the 'Amoco Cadiz', the benthic community moved (2) physical conditions cause variation in the dispersal through several successive penods of temporal change and weathenng of oil, (3) the toxicity of spilled oil (Jacobs 1980, Dauvin 1982, Glemarec & Hussenot varies based on the type spilled and how it is weath- 1982, Ibanez & Dauvin 1988, Warwick & Clarke 1993) ered, (4) sensitivities of anima1.s vary among species until the community approached a stable 'climax' envi- and not all species have toxicity test data so substitu- ronment nearly 20 yr after the spill (Dauvin 1998). tions are made, and (5) Systems are complex and In March 1989, the super tanker 'Exxon Valdez' ran impacts often result from indirect effects (e.g. reduc- aground on Bligh Reef in Prince William Sound, tion in predator abundance) or cleanup efforts rather Alaska, and spilled nearly 42 million 1 of crude oil, the than direct toxicological impacts. There are some data largest spill in U.S. history (Spies et al. 1996). The oil from controlled experimental spills (Cross et al. 1987) was driven by wind and currents, and much of it found and mesocosm studies (Grassle et al. 1981, Elmgren & its way into sheltered embayments within the Sound Frithsen 1982), but these are few. As a result, our most (Kelso & Kendziorek 1991, Maki 1991). About half of reliable understanding of how spills impact the envi- the oil became stranded on the shoreline, and an esti- ronment and affect natural biota Comes from examina- mated 13 % was deposited in subtidal sediments (as of tion of impacts from accidental spills. fall 1992; Wolfe et al. 1994). These shorelines were Accidental spills in manne coastal environments subsequently subjected to extensive and intensive often result in the coating of the shoreline with oil, and cleanup activities (reviewed in Mearns 1996). Shortly can lead to immediate and readily detectable impacts after the spill, plants and animals living on or in shal- to the biota (NRC 1985), but impacts are less obvious to low subtidal sediments were directly exposed to nearshore subtidal habitats because these are out of beached oil and oil-contaminated fine sediments that sight. Oil often finds its way into the subtidal via diffu- were reintroduced into nearshore tvaters by wave- sion, or more importantly from Settlement of oiled par- induced redistribution, leaching by groundwater flow- ticulates often denved from beached oil that is redis- ing through the intertidal region, and flushing seaward tributed by waves and cleanup activities (Boehm et al. by shoreline cleanup activities (Boehm et al. 1987, 1987, Owens et al. 1987, Short et al. 1996a). The oil can O'Clair et al. 1996. Short et al. 1996a). Most of this rein- be absorbed onto inorganic particles (Meyers & Quinn troduced oil, oil absorbed onto inorganic particles and 1973), or ingested by zooplankton and incorporated oil incorporated into feces, subsequently settled subti- into feces that settle onto subtidal sediments (Conover dally at depths les~than 20 m (Carlson & Kvenvolden 1971, Bassin & Ichiye 1977, Clark & MacLeod 1977). 1996, O'Clair et al. 1996). Nearshore subtidal zones, and especially soft-sedi- There was extensive injury to nearshore environ- ment environments in protected bays, often serve as ments as a joint result of the 'Exxon Valdez' spill and reservoirs of oil that can persist for long periods of time associated cleanup efforts. These included reductions (Gundlach et al. 1983, NRC 1985, Page et al. 1995). in the abundance and Cover by the dominant intertidal Impacts to epifaunal and infaunal organisms can be alga Fucus gardnen and reductions to a suite of inter- severe but are often unpredictable. The Same spill can tidal infauna and epifauna (e.g. De Vogelaere & Foster lead to enhancement of some species via organic 1994, Gilfillan et al. 1995. Driskell et al. 1996, High- ennchment, and to the local extinction of more sensi- srnith et al. 1996, Houghton et al. 1996, Stekoll et al. tive species as the result of toxicity. This dual response 1996, van Tamelen et al. 1997). Assemblages of (toxicity and organic ennchment) to pollution is inher- infauna from oiled mixed-soft beaches and epifauna ant in most studies on environmental effects of hydro- from oiled sheltered rocky shores were essentially carbon exposure (e.g. Spies & DesMarais 1983, Ols- recovered by the Summer of 1991; sites subjected to gard & Gray 1995, Peterson et al. 1996). In 3 of the most high-pressure hot-water washing remained in early well studied coastal spills, the 'Flonda' spill off the stages of recovery for both infauna and epifauna by the coast of West Falmouth, Massachusetts, USA (Sanders summer 1992 (Driskell et al. 1996, Houghton et al. et al. 1980), the 'Amoco Cadiz' spill off the Bnttany 1996). Injunes appeared to be less severe in subtidal coast of France (Jacobs 1980, Glemarec & Hussenot habitats. There were possible impacts to subtidal algae 1982, Ibanez & Dauvin 1988), and the 'Tsesis' oil spill based upon higher densities of at least 1 of the domi- off the Swedish Baltic coast (Kineman et al. 1980, Elm- nant (Agarum clathratum, Laminaria saccharina, gren et al. 1983), there were extreme increases as well Nereocystis luetkeana) at oiled sites, and generally as decreases in benthic infauna at heavily oiled sites. more smaller plants there (Dean et al. 1996a). Eelgrass Impacts varied among spills and among species within Zostera rnarina populations showed reductions in flow- a spill, but in all cases, there was a reduction in 1 or ering and shoot density and recovered within 2 yr after more amphipod species and an increase in opportunis- the spill (Dean et al. 1998). There were catastrophic tic polychaete species. In the longest studied oil spill, declines in benthic infauna in a heavily oiled fjord sev- Jewett et al.:Oil spill impacts on eelgrass beds 61

eral months after the spill that was associated with as descnbed in Dean et al. (1996a, b, 1998). There were anoxic conditions (Jewett et al. 1996). The presence of no pre-spill data, and out of necessity we relied on a oil may have exacerbated the local deletion of oxygen, comparison of oiled and reference sites sampled after but anoxia can occur naturally in these environments, the spill to evaluate impacts rather than a more ideal and it is uncertain whether the spill contnbuted to this BACI (Before-After, Control-Impact) design (Wiens & condition. There was a reduction in several large epi- Parker 1995, Stewart-Oaten 1996). Oiled areas were fauna found in subtidal and eelgrass beds, includ- identified based on the summer 1989 oil survey maps ing the leather sea Star Dermasterias jmbricata and the and the September 1989 shoreline surveys (ADEC helmet crab Telrnessus cheiragonus (Dean et al. 199613) 1989). Briefly, eelgrass beds adjacent to moderately to at depths <20 m, but there were no detectable injuries heavily oiled shorelines were stratified by geographic to Tanner crab Chionoecetes bairdi and pandalid region, and 1 site from each region was randomly Shrimp Pandalus spp. at depths of 20 to 150 m (Arm- selected. Reference sites were chosen that matched strong et al. 1995). Also, Feder & Blanchard (1998) ana- the oiled sites as closely as possible with respect to lyzed infauna from 14 bays in the Sound in 1990 at 40, aspect, proximity to frech-water run-off, slope, wave 100, and >I00 m, and found no detectable effects on exposure, and water circulation. Sampling was con- the biota 16 mo after the spill. The spill also resulted in ducted in July or August at 4 pairs of sites in 1990, 5 in the reduction in the numbers of several important 1991, 3 in 1993, and 4 in 1995. Data from only those 3 predators in the nearshore Zone, including sea otters pairs of sites sampled in each year are presented here. Enhydra lutra (Ballachey et al. 1994), that can have a The oiledheference site pairs were Bay of Isles/Drier profound influence on the structure of nearshore com- Bay, Herring Bay/Lower Hernng Bay, and Sleepy munities (Estes & Duggins 1995) including those on Bay/Moose Lips Bay, respectively (Fig. 1). soft bottoms (Kvitek & Oliver 1992a, b). Two depth Strata were sampled adjacent to each In this study, we examined the effect of the 'Exxon 200 m section of shoreline selected for sampling: 6 to Valdez' oil spill on the nearshore benthic community in 20 m and within the eelgrass bed (generally <3 rn). eelgrass beds and immediately offshore of the beds, Within each stratum, three 30 m sampling transects, and followed the recovery of the community for a run parallel to shore, were placed at randomly selected penod of 6 yr after the spill. The shallow subtidal Zone locations. Transects within the eelgrass bed were less than 20 m deep in Pnnce William Sound typically placed in what was perceived as the center of the has dense stands of kelps (on rocky shores) or eelgrass depth distnbution of eelgrass. The sites in the deeper (in soft sediment). Eelgrass beds generally occur at stratum were placed at randomly selected depths. Ran- stream mouths in protected bays and extend from the dom positions along the shore and random depths lower intertidal (about 0.5 m mean lower low water, were reassigned prior to each year. MLLW) to depths of about 5 m (McRoy 1970, Dean et Densities of infaunal invertebrates were estimated al. 1998). Soft substrata outside of the eelgrass beds from two 0.1 m2 by 10 cm deep diver-collected suction consist mainly of mud and sand that is largely devoid dredge sarnples taken from each of the 3 sampling of macrophytes except for the occasional Laminaria transects within each depth stratum. The 2 dredge saccharina (S. Jewett & T. Dean pers. obs.). Eelgrass samples were spaced 7.5 m apart, with the initial sam- communities are among the most productive habitats pling position at a random distance along the sampling in Alaska (McRoy 1970, Phillips 1984) and elsewhere transect. In the eelgrass bed, eelgrass shoots were cut (e.g. Boström & Bonsdorff 1997), and they serve as off just above the sediments' surface, removed, and important nursery and feeding grounds for a variety of counted prior to taking suction dredge sarnples. The invertebrates, fish, birds, and marine mammals dredge sampler was fitted with a collection bag with a (Thayer & Phillips 1977, McConnaughey & McRoy mesh size of 1 mm. 1979, Feder & Jewett 1987). Divers also collected 2 sedirnent samples from each transect and depth stratum. Surface sediments were collected from within 3 m of the Start of each sampling METHODS transect. The samples were collected to depths <5 cm in pre-cleaned wide mouth jars and frozen immedi- Sampling. The impacts to the benthos were assessed ately after collection. One of the samples was used to using a stratified random sampling design. Community determine sedirnent grain-size composition and the parameters (abundance, biomass, diversity, and rich- other to determine hydrocarbon concentrations. ness) of epibenthic and infaunal invertebrates, as well Laboratory analyses. Benthic samples were re- as densities of all numerically dominant invertebrate turned to the laboratory where they were sieved families, were measured at matched pairs of oiled and through a 1 mm mesh, and the anirnals were identified, unoiled reference sites beginning 13 mo after the spill counted, and weighed (blotted-wet wt). Organisms 62 Mar Ecol Prog Ser 185: 59-83, 1999

148W 147W R Chemical analysis of sedinients yielded values ,' ! < ,,L. for several component hydrocarbon analytes. We ('' ." ,' ' ., / -.,.*y;-.'I$ , ,, report the concentrations of total polynuclear aro- 61' 00' matic hydrocarbon (TPAH) fractions minus pery- e. F . lene and sum of the chrysenes as indicators of the contribution of 'Exxon Valdez' oil. Perylene was excluded because it is a naturally occurring PAH . i compound that ic produced diagenetically in , . , marine sediments (Venkatesan 1988) and is not

; 1,p L. #. Prince -,P--- -derived. While concentrations of TPAH $2 ,L -. , r,... . reflect the distribution of petroleum-derived hydro- -. William Sound carbons, a Portion of the TPAHs observed may 60'30' - have been denved from petroleum hydrocarbons J- J- other than 'Exxon Valdez' oil, especially diese1 fuel. ,A ,i Chrysene concentrations were used as rnore spe- '--F~~cific indicators of 'Exxon Valdez' oil, because chry- senes are present in crude oil (including 'Exxon -1 Valdez' oil) but not in diese1 fuel (Bence & Burns 1995). Both TPAH and chrysene values may include some PAHs from non-anthropogenic sources and from sources other than 'Exxon Valdez' oil (e.g.fuel spilled from cleanup and research ves- N sels), and thus, do not necessarily represent absolute concentrations attributable to the oil spill. -1 ' .,, However, there is no reason to suspect biases with ,/-- -, ., ,C-V km i 1-' -1 respect to the distnbution of non-spill related oil at .- - , 0 5 10 <- .-~,,' oiled versus reference sites. Therefore, differences

in concentrations OETPAHs and chrysenes at oiled versus reference sites should reflect relative Fig. 1. Eelgrasc sarnpling sitec in Prince William Sound. Three oiled im~acts0f the s~ill. (H,0, A)/reference (U, 0, A) site pairs are shown. Also depicted Results of the chemical an.alysis cvere screened is the site of the 'Exxon Valdez' grounding on Bligh Reef and the On the basis of Surrogate recoveries arid minimum trajectory of the spill detection lirnits (MDLs). Individual analytes and the sumrnary statistics affected by them were were generally identified to at least the family level, excluded frorn the analysis if the recoveries of corre- but there were a few instances when only higher taxo- sponding analyte Surrogates fell outside the range 30 nomic levels were assigned to an individual. Many of to 150%. For example, if the Surrogate of 1 PAH the more common organisms were identified to spe- analyte fell outside the acceptable range, the TPAH cies. Meiofauna (50.5mm) and highly motile, non-ben- concentration for that sample was excluded from thic forms, such as mysids and fishes were excluded the an.alysis. Dry weight concentrations of indivi.dua1 frorn the analyses. analytes reported belocv MDL were replaced by 'Os' for Sediment granulometric analysis was conducted our analyses. The MDL for aromatic hydrocarbons using the pipette-sieve method, providing conven- was 1 ng g-'. tional grain-size Parameters (Folk 1980). Hydrocarbon Criteria were established for companng hydrocar- analyses were performed by the Geochemical and bon concentrations in sedirnents with those in 'Exxon Environmental Research Group at Texas A&M Univer- Valdez' oil (J. Short, NOAA/NMFS, pers. comm.). The sity (in 1990 and 1991) and the Technical Services Task Pattern of PAH concentrations in the sediment samples Force, Analytical Chemistry Group (TSTF-ACG), was judged similar to 'Exxon Valdez' oil if it rnet 3 cri- NOAA/NhIFS, Auke Bay, Alaska (in 1993 and 1995). tena: (1)the ratio of alkyl dibenzothiophenes (summed) Hydrocarbons were extracted from the sediment Sam- to alkyl phenanthrenes (summed)exceeded 0.3; (2)the ples and analyzed for the concenlration of vanous ratio of alkyl chrysenes (siimmed) to alkyl phenan- hydrocarbon fractions using gas chromatography corn- threnes (summed) exceeded 0.05; and (3) the concen- bined with a mass spectrometer detector (Short et al. tration of alkyl phenanthrenes exceeded 25 ng g-'. 1996b). All hydrocarbon data are reported on a dry This test is similar to, but more conservative than, the weight basis (ng g-'1. method used by O'Clair et al. (1996). Jewett et al.: Oil spill impacts on eelgrass beds 63

Data analyses. The data were analyzed using both culation of richness and diversity. Analyses were also multivanate and univariate techniques. Multivanate perforined for dominant families and for some major analyses were used to examine spatial and temporal higher taxonomic groupings. Dominant families were pattei-ns in community composition, and to test for the those that occurred in one-sixth or more of the samples correspondence of these Patterns with changes in and ranked among the 15 most abundant taxa in any environmental variables. The Bray-Curtis dissimilarity given year and depth stratum. coefficient (Bray & Curtis 1957) was calculated using Two-way randomization ANOVAs (Manly 1991) ln(x + 1)-transformed family-level abundance data were used to test for differences between oiled and ref- (ind. 0.1 m-2) for dissimilanty matrices. Separate matri- erence sites, between years, and for the interaction of ces were produced for each depth stratuin and for oiling category and year, as described in Dean et al. infauna and epifauna. Agglomerative unweighted (199613). Oiling category (oiled or reference) and year group-averaging cluster analyses (Clifford & Stephen- were treated as fixed mean effects with pair as a ran- son 1975) and nonmetric multidimensional scaling dom blocking factor Sites were treated as experimen- (MDS; Field et al. 1982) were applied to the dissimilar- tal units and site means were used as replicates in all ity matrices to provide the initial classification of sites analyses. Separate analyses were performed for each and sampling times into site/year groups. Site/year depth stratum, and for biological data, for infaunal and groupings were determined based on general agree- epifaunal species. Individual families were assigned to ment of the clustering and ordination techniques. Dis- either infaunal or epifaunal groups according to the criminant analysis was used to examine the degree to habitat of the dominant family within the group. In which key environmental variables could discriminate cases where the 2-way randomization ANOVA indi- the site/year groups. The absolute magnitude of the cated a significant effect of oiling category or a signifi- standardized coefficients of the variables along the dis- Cant interaction between oiling category and year, cnminant axes represent the contribution (i.e. impor- contrasts were made between oiled and reference sites tance) of each variable to an axis. Environmental vari- within years. In the majority of the cases (e.g. 23 of 45 ables used in the analysis were: percentage sand, families foi- which analyses were conducted) there was percentage rnud, TPAH, and chrysene concentrations. a significant effect of pair (p < 0.10), the random block- Percentage Sand and mud were used because sedi- ing factor. We do not report these because while they ment grain-size characteristics are major determinants generally Support our use of a matched-pair design, of community cornposition in marine benthic commu- they do not influence our interpretation of the data nities (e.g. Rhoads 1974, Grebmeier et al. 1989).Other with respect to the impacts of oil. environmental variables, including percentage gravel The percentage of mud was used as a covariate in and various other hydrocarbon analytes (e.g. non- some analyses of infaunal data. A preliminary 2-way aromatics, dibenzothiophenes, fluorenes, naphthalenes, ANOVA was performed using standard normal theory phenanthrenes, and alkanes), were excluded because methods (without randomization) with percentage (in the case of gravel) they were linearly dependent rnud as a covariate. In cases where percentage rnud with percentage sand and rnud or (in the case of other contributed significantly to the model, with p < 0.20, hydrocarbon analytes) were highly correlated (r > 0.9) we used mud as a covariate in the final randomization with either TPAH or chrysene concentrations. Percent- test. While probabilities associated with covariates age sand and rnud variables were arcsine[square root were not exactly the Same for the normal theory and (X)]-transformed and chrysene and TPAH concentra- randomization models, preliminary tests indicated that tions were ln(x)-transformed. All 4 variables were con- they were generally similar (usually within 0.02) and sidered in the discriminant analysis model, as opposed were deemed a reasonable and efficient means of to stepwise procedures, so that the contnbution of each evaluating whether to include mud as a covanate, variable to the discriminant axes could be assessed. and to avoid over parameterization in cases where Univariate analyses were used to test for differences the percentage rnud was unlikely to contribute sig- between oiled and reference sites, between years, and nificantly. to examine possible interactions between oiling cate- We measured eelgrass shoot density in each sam- gory and year. Metncs analyzed included TPAH and pling quadrat, but did not use this as a covariate for chrysene concentrations, and various community and the following reasons. First, there were differences in individual taxon measures for the biological data. eelgrass shoot density at oiled and reference sites and Community measures included total abundance, total differences in the possible impact of the spill on eel- biomass, Margalef's species richness index, SR (Green grass (Dean et al. 1998). Therefore, implications of 1979), and Shannon's diversity index, H' (Shannon & spill impacts on benthic invertebrates could poten- Weaver 1963). Taxa identified to the family level (or tially be masked in analyses using eelgrass shoot den- higher, i.e. class, order, phylum) were used in the cal- sity as a covariate. Second, using data froin only refer- 64 Mar Ecol Prog Ser 185: 59-83, 1999

ence sites, we correlated the density of shoots from RESULTS within each quadrat with the density of dominant families of benthic invertebrates and found no signifi- Hydrocarbons in sediments cant relationships. Squared correlation coefficients never exceeded 0.17 (n = 35, p > 0.13 in all cases). In 1990, over a year after the oil spill, mean TPAH Also, we did not use percentage sand as a covariate concentrations were 4 to 8 times higher in sediments because preliminary analyses suggested that when collected from sites adjacent to heavily oiled shorelines percentage mud was also used as a covanate, per- than at reference sites (Table I), and the majority of centage sand did not contribute significantly to the samples from oiled sites had analyte compositions model, and its inclusion could lead to over-parameter- indicative of 'Exxon Valdez' oil (Table 2). The average ization. TPAH concentration at oiled sites was approx. 5000 ng A Type I error rate of p I0.10 was used for all analy- g-' and was significantly higher than at reference sites ses. This was done in an effort to balance Type I (a) at each Stratum (p < 0.02) (Tables 2 & 3). There was and Type I1 (ß) error rates, and to insure that possible however considerable variability between sites, with differences were not overlooked because of a lack of concentrations ranging from only 45 ng g-' (at Herring statistical power (high Type I1 error) (Eberhardt & Bay, 6 to 20 m depth) to over 15000 ng g-' (at Bay of Thomas 1991, Mapstone 1995). Type I1 errors are those Isles, 6 to 20 m depth). associatrd with the failure to correctly identify signifi- TPAH concentrations declined significantly over cant differences when they exist. High Type I1 error time at both oiled and reference sites (Tables 1 & 3). A rates result from high vanance and small sample sizes. significant interaction between oiling category and We have not calculated power for our tests, but the year for both depth strata resulted from the higher con- largs variation between sites, small sample size (n = 3), centrations at oiled sites in 1990, and decline of TPAH and the generally large differences between means in concentrations at both oiled and reference sites to sim- cases where statistically significant differences were ilarly low concentrations in 1995. In 1991, the majority observed suggest relatively low power. In all instances of samples from oiled sites still had signatures indica- we report at least relative probabilities (p 5 0.10, p i tive of 'Exxon Valdez' oil (Table 2), but mean concen- 0.05, or p 5 0.01). trations had declined by more than 90%. By 1993.

Table 1. Mean concentrat~onsof (A) TPAHs and (B) chrysenes in shallow subtidal sediments in oiled (0) and reference (R)sites, Prince Wiiliam Sound, 1990-95. BI: Bay of Isles, DB: Drier Bay. HB: Herring Bay, LHB: Lower Hernng Bay, SB: Sleepy Bay, MB: Moose Lips Bay

Year BI (0) DB (R) HB (0) LHB (R) SB (0) MB (R) Oiled Reference

(Al TPAH (ng g-I) at 6-20 m 1990 15253 1991 1122 1993 612 1995 256 TPAH (ng g-I) at c3 m 1990 14309 1991 790 1993 443 1995 313

(B) Chrysene (ng g-I) at 6-20 m 1990 16 3 3 54 3 4 1991 21 2 0 14 2 3 1993 32 3 0 13 2 2 1995 5 5 7 23 5 4 Chrysene (ng g-') at <3 m 1990 42 5 1 9 2 2 1991 9 2 0 7 1 1 1993 17 U 3 0 1 2 1 1995 16 3 1 0 1 1 2 Jewett et al.: Oil spill iinpacts on eelgrass beds 65

Table 2. Classification of sedirnent hydrocarbon samples, appeared to decline over time (Table 1). Concentra- based on the presence of an 'Exxon Valdez' oil signature, tions were significantly higher at oiled sites, but differ- from 2 depth Strata in and adjacent to eelgrass beds ences between years and interactions between years and oiling categories were not significant (Table 3). Year Oil/ Deep (6-20 m) Bed (<3 m) Within year contrasts showed that chrysene concentra- Reference No. of No. with No. of No. with tions were significantly higher at oiled sites than at ref- samples 'Exxon samples 'Exxon Valdez' oil Valdez' oil erence sites from 1990 through 1993, but not in 1995. There was moderate hydrocarbon contamination at 1990 0 7 2 9 8 reference sites. Relatively high concentrations of both R 6 2 9 3 TPAH and chrysenes were observed at reference sites in 1990 and 1991, and nearly one-third or more of the samples had signatures indicative of 'Exxon Valdez' oil (Table 2). At some reference sites, contamination was apparently from sources other than 'Exxon Valdez' oil. For example, the average concentration of TPAHs at 6 to 20 m depth at the Lower Hernng Bay in 1990 was 1908 ng g-', but none of the samples from there had analyte signatures characteristic of weathered 'Exxon Table 3. Two-way ANOVA and contrasts of oiled versus refer- Valdez' crude oil. The TPAH Patterns were more evi- ence sites within years for mean concentrations of (A)TPAHs dent of gasoline (possibly contaminated during collec- and (B) chrysenes in shallow subtidal Sediments in oiled and reference sites, Prince Wiiliam Sound, 1990-95 tion), diese1 oil (possibly from cleanup activities), sub- marine oil seeps (Page et al. 1995), or coal (Short et al. 1999). p-values 6-20 m <3 m

(A) TPAH (ngg-I) The benthic community-multivariate analyses Source of variation Oilcode 0.01 0.02 Over 400 taxa (genera and species) of infaunal and Year <0.01 <0.01 Interaction <0.01 <0.01 small epifaunal invertebrates with representatives from over 151 families were collected. The infaunal Contrasts of oil vs reference within years 1990 0.02 0.02 community was numencally dominated by a vanety of 1991 0.08 O.O? polychaetes, bivalve and gastropod mollusks, and 1993 <0.01 < 0 01 crustaceans (mostly amphipods), an assemblage typi- 1995 0.24 0.23 cal of infaunal communities from shallow mud and (B) Chrysene (ngg-') sand bottoms in subarctic waters (Feder & Jewett Source of variation 1987). Mytilid bivalves (Musculus spp.) and spirorbid Oilcode <0.01 <0.01 polychaetes dominated the epifaunal community. Year 0.15 0.15 Cluster and MDS analyses (Fig. 2) indicated that, for Interaction 0.13 0.13 both infauna and epifauna, there was greater spatial Contrasts of oil vs reference within years than temporal vanability. In most instances samples 1990 <0.01 <0.01 1991 <0.01 ~0.01 collected from the Same sites in different years tended 1993 <0.01 <0.01 to be more similar than samples collected from differ- 1995 0.16 0.18 ent sites in a given year. Also, there was a tendency for sites of a matched pair (e.g. Herring Bay and Lower Hernng Bay) to cluster. Clusters of site/years were not average TPAH concentrations dropped to below 300 as well defined for epifaunal communities in the 6 to ng g-' at oiled sites, and by 1995, averaged less than 20 m depth Stratum compared with infaunal communi- 230 ng g-' at both oiled and reference sites and did not ties and compared with the epifauna within the eel- differ significantly between sites. However, there were grass bed. still indications of slight exposure to oil in 1995, as 2 Infauna in the deeper Stratum at 1 oiled site, Bay of samples (1 from Bay of Isles and 1 from Sleepy Bay) Isles, was remote in ordinate space from the majority of had TPAH concentrations exceeding 300 ng g-', and the other site/years, and the sample from 1990 was had signatures matching 'Exxon Valdez' oil. extremely remote compared with other years (Fig. 2). A similar Pattern was observed in total chrysenes, as This site generally had higher hydrocarbon concentra- concentrations were generally higher at oiled sites and tions than other sites, and the mean concentration of WSUYKgk- W Y

RY3,ML OYISB 095BI RYSML OY5SB 1 W3Bl 09398 OYlBl R9OML 09SB RYSLH R95LH R93LH OYJSB W R93LH 095HB OYISB RYILH 091HB 090S8 R9IML IO9lHB OYJHB 090H B WSHB RYILH 093HB RYOLll OWllB WISB RWLH 09SB 09581 091SB 0 09181 O%B R95ML R95ML 09381 R93ML RWML R91ML R95DB R90ML F R93ML R95DB R93DB R93DB RWDB R91DB R91DB RWDB A 09081 OWBl W 3

Axis 2 Axis 2 Axis 2 C 3 W Jewett et al.: Oil spill impacts on eelgrass beds 67

TPAH at Bav of Isles in 1990 (qreater.- than Table 4. Percentage dissimilanty between pairs of oiled and reference 15000 ng g-l) was the highest concentration sites over time and the change in percentage dissimilarity between 1990 and 1995. BI: Bay of Isles, DB: Drier Bay, HB: Herring Bay, LH: Lower of TPAH observed (Table 1). Herring Bay, SB: Sleepy Bay. ML: Moose Lips Bay For infauna in the eelgrass bed (<3 m depth), 1990 samples from Bay of Isles, Drier Depth Year Infauna Epifauna Bay, and Lower Herring Bay were remote BI/DB HB/LH SB/ML BI/DB HB/LH SB/ML from other site/years (Fig. 2). Two of these were reference sites, but sediments from al13 6-20 m 1990 68 30 41 65 46 61 1991 44 26 39 82 40 56 had high TPAH concentrations (14309 ng g-' 1993 49 26 35 71 38 62 at Bay of Isles, 1418 ng g-' at Drier Bay, and 1995 36 33 34 72 47 75 297 ng g-' at Lower Herring Bay), and 2 of % Change -32 3 -7 7 1 14 the 3 (Bay of Isles, Drier Bay) had the highest <3m 1990 35 42 31 36 46 46 concentrations. 1991 33 29 33 33 50 52 There was little evidence of convergence 1993 33 30 41 41 60 59 between communities at matched pairs of 1995 37 34 30 71 46 67 oiled and reference sites over time. Levels of % Change 2 -8 -1 3 5 0 21 dissimilarity decreased markedly between 1990 and 1995 for the infaunal community at Table 5. Results from discriminant analyses of environmental vanables the 6 to 20 m depth at Bay of Isles and Drier for shallow subtidal sites in and adjacent to eelgrass beds in Prince William Sound, 1990, 1991, 1993, and 1995. F: Fstatistic, p-value: signifi- Bay. Otherwise, there were no substantial cance of F statistic, Sig. Ax.: significance of denved axis. % Var.: cumu- decreases in dissimilarity over time (Table 4). lative percent of variance accounted for by each axis, % Corr.: percent of In fact, some pairs of sites, and especially epi- correct classifications of original data by the discriminant functions fauna at <3 in at Bay of Isles and Drier Bay, became more dissimilar with time. F p-value Standardized coefficients Discriminant analyses suggest that sedi- Axis 1 Axis 2 Axis 3 Axis 4 -- ment characteristics and chrysene concen- Deep (6-20 rn)infauna trations were important in distinguishing Sand 7.39 0.003 0.38 -0.76 -0.65 site/year groupings indicated by cluster and Mud 16.18 <0.001 -0.84 -0.55 -0.13 MDS analyses. In analyses using percentage Chrysene 6.93 0.003 0.86 -1 08 1.3 sand, percentage mud, TPAH concentra- TPAH 2.29 0 118 0.73 0.85 -0.82 tions, and chrysene concentrations as poten- Sig. Ax. <0.001 <0.001 0.004 tial explanatory vanables, at least 91 % the % Var. 73% 91% 100% % Corr 91% variance in the community composition was explained by the first 2 discnminant axes Deep (6-20 m) epifauna (Table 5). The percentage of correct classifi- Sand 38 <0.001 Mud 32.88 <0.001 cations ranged from 91 to 100%, indicating a Chrysene 6.14 0.005 strong relationship between the distributions TPAH 2.12 0 132 of the variables and the site/year groupings Sig. Ax. for each data Set. In all but the epifaunal O:, Var community within the eelgrass bed, percent- 'X,Corr. 100 % age mud and sand were highly significant Bed (e3 rn)infauna contnbutors to the discriminant model (p 5 Sand 14.58 <0.001 0.01). Chrysenes were also important con- Mud 13.76 <0.001 Chrysene 2.31 0.118 tributors (i.e. p 5 0.1) in all but the infaunal TPAH 1.58 0.235 community at <3 m and had the highest Sig. Ax. standardized coefficients for most analyses. % Var TPAH was never a significant contributor to :L Corr. 100% the discrirninant model. Bed(e3 rn)epifauna The relationships between environmental Sand 0.68 0.586 vanables and community structure are also Mud 3.07 0.084 evident in overlays of environmental vari- Chrysene 2.98 0.089 ables on the MDS ordinations (Fig. 3). TPAH 1.65 0.246 Site/year groupings generally exhibited val- Sig. Ax. ues of percentage sand and mud of similar % Var. % Corr. 94% magnitudes, with values increasing from left % Mud % Sand 'TPAH Chrysene

R

Axis L Axis I Aris 1

Axis 1

Axia 1 Aris I Axis 1

Axia 1 Axis 1 Aris 1 Axis I

Fig. 3. Overlays of environrnental variables (% Mud, O/o Sand, TPAH, Chrysene) on blDS ordinations of sites. Circles and letters denote site groupings from cluster analysis. Size ol symbol indicates relative value 01 the variable Jewett et al.: Oil spill impacts on eelgrass beds 69

to nght, particularly for percentage mud. There was a The abundance of individual families less obvious correspondence of site/year groupings with magnitudes of chrysenes or TPAHs, but samples Of the 25 dominant families in the 6 to 20 m depth within a site/year group or within adjacent groups Stratum, and the 22 dominant families within the eel- generally had similar TPAH and chrysene concen- grass bed, over half had densities that differed signifi- trations. cantly between oiled and reference sites, or displayed a significant interaction between oiling category and year (Tables 10 & 11). Some were more abundant at Community parameters and the abundance of higher reference sites, some were more abundant at oiled taxonomic groups

The most consistent Pattern among community Table 6. Mean values of community parameters for benthic parameters was the significantly higher total abun- invertebrates at oiled (0) and reference (R) sites in and adja- Cent to eelgrass beds in Pnnce Wdiam Sound, 1990, 1991 dance and biomass of epifauna at oiled than at refer- 1993, and 1995 ence sites at both depth strata (Tables 6 & 7). Also, there were significant interactions between oiling Community O/R 1990 1991 1993 1995 category and year for epifaunal total abundance in parameters both depth strata, and for epifaunal total biomass at 6 to 20 m. Biomass of epifauna at <3 m was signifi- 6-20 m infauna cantly higher at oiled sites in 1990 and 1991, but not Shannon 0 2.14 2.36 2.58 2.52 diversity R 2.42 2.53 2.44 2.65 thereafter. Otherwise. there were no consistent tem- poral trends in epifaunal abundance or biomass. Total Species 0 5.45 5.35 5.76 6.22 richness R 4 65 5.84 6.07 6.63 abundance and biomass of infauna also differed sig- nificantly between oiled and reference sites, but the Total abund. 0 472 958 1018 1042 (no. 0.1 m-2) R 655 1185 1579 1021 direction differed with depth. Both abundance and biomass were higher at reference sites at 6 to 20 m, Total biom. 0 6.22 6.17 7.38 7.79 (g dw 0.1 m-2] R 6 57 13.95 13.63 10.70 and higher at oiled sites at <3 m. Total abundance of infauna at the shallow depth was higher at oiled sites 6-20 m epifauna in 1993 and 1995, but not pnor. Otherwise, there Shannon were no consistent temporal trends and there were diversity no significant interactions between oiling category Species and year for infauna. Shannon diversity (H') of epi- richness fauna was higher at oiled sites within the eelgrass Total abund. bed, but otherwise, there were no significant effects (no. 0.1 m-2) of oiling category for diversity measures (H' or spe- Total biom. cies richness) . (g dw 0.1 m-2) The greater overall abundance of epifauna at oiled <3 m infauna sites was mostly attributable to a greater abundance of Shannon bivalves (Tables 8 & 9). There were significantly more diversity epifaunal bivalves at oiled sites at both depth strata Species (p 1 0.01 for both). At <3 m there were also signifi- richness cantly higher densities of epifaunal gastropods and Total abund echinoderms at oiled sites. For infauna, there were (no. 0.1 n1r2) higher densities of polychaetes and gastropods at oiled Total biom. sites in both depth strata. Significant interactions (g dw 0.1 m-2) between oiling category and year were evident for <3 m epifauna other crustacean infauna (i.e. cumaceans, isopods) at Shannon both depth strata, and epifauna at <3 m, as well as for diversity epifaunal polychaetes, amphipods, and echinoderms Species in the eelgrass bed. However, none showed consistent richness temporal trends. Only infaunal amphipods within the Total abund eelgrass bed were more abundant at reference sites (no. 0.1 m-') (p 5 0.05). Densities of infaunal amphipods at <3 m Total biom. were significantly higher at reference sites in 1990, but (g dw 0.1 m-') not in subsequent years. 7 0 Mar Ecol Prog Ser 185: 59-83, 1999

Table 7 Summary of randomization ANOVAs for ben.thic invertebrate community Parameters. Covariates used in the analyses are listed. Also given are contrasts between oiled and reference sites within years for those instances where there was a signifi- Cant effect of oll category or a significant interaction between oll category and year

Covanate 2-way randomization ANOVA Within year contrasts Oil category Year lnteraction 1990 1991 1993 1995

Deep (6-20 m) Infauna Shannon diversity None Total abundance Mud Total biomass None Species richness Mud Epifauna Shannon diversity None Total abundance None Total biomass None Species nchness None

Eelgrass bed (<3 m) iriiduiia Shannon diversity None Total abundance Mud Total biomass None Species richness None Epifauna Shannon diversity None - - - Total abundance None XXX m.. m.. .. - Total biomass None - - - Species richness None .. -

= greater at oiled sites @@@,000, MX = p < 0.01 0 = greater at control sites @@,00, M = 0.01 < p < 0.05 X = significant difference sites @, 0,X = 0.05 < p < 0.1 - =p>0.1

sites, and some had a significant oiling/year interac- sites (generally >20 ind. 0.1 m-') in both depth strata tion. The most well represented group was composed (Fig. 5, Tables 10 & ll), and there were very few of families that tended to be more abundant at oiled isaeids or phoxocephalids at oiled sites, especially in sites. This included 1 epifaunal bivalve (Mytilidae 1990. Both families showed some marginal increase in within both depth strata), 2 infaunal gastropods (Cae- abundance at oiled sites over time, but the interaction cidae at 6 to 20 m and Lacunidae at <3 rn), 2 infaunal between the oiling category and year was significant bivalves (Montacutidae and Turtoniidae at 13 rn), and only for phoxocephalids at 6 to 20 m. By 1995. only 9 infaunal polychaete families (Spionidae at both phoxocephalids were more abundant at reference sites depths; Maldanidae, Nephtyidae, and Syllidae at 6 to within the eelgrass bed (p 5 0.01), but both families 20 m; Amphictenidae, Nereidae, Opheliidae, Sabelli- were still about 10 times more abundant at reference dae, and Siga11.onidae at <3 m) (Tables 10 & 11). As sites at both depth strata. Infaunal bivalve families indicated in abundance Patterns for mytilids, and rep- (including Lucinidae, Montacutidae, and Thyasiridae resentative families of dominant polychaetes and gas- in the 6 to 20 m stratum, and Tellinidae within the eel- tropods (Fig. 4),all were consistently more abundant at grass bed) and polychaetes (Lumbrineridae and Ophe- oiled sites from 1990 through 1995. There was little or liedae at 6 to 20 m) were also more abundant at refer- no indication of declining abundances at oiled sites, or ence sites (Fig. 6, Tables 10 & 11).Some of these (e.g. increases at reference sites over time. Mon.tacutidae and Thyasindae) showed marginal The second group, with greater abundance at refer- increases in abundance at oiled sites between 1990 ences sites, was represented by 2 infaunal amphipod and 1995, but none had significant interactions families, 3 families of infaunal bivalves, and 2 infaunal between oiling category and year. polychaete families. Isaeid and phoxocephalid The remaining group consisted of those families for amphipods were more abundant (p < 0.1) at reference which there was a significant oiling category by year Jewett et al.: Oil spill impacts on eelgrass beds 7 1.

Table 8. Mean densities (no. 0.1 m-2) for rnajor taxonomic interaction, but no significant effect of oiling- category.-. groups of benthic invertebrates at oiled (0) and reference (R) This included 2 polychaetes (Syllidae and Spirorbidae) sites in and adjacent to eelgrass beds in Pnnce William and 2 amphipods (Ischyrocendae and Caprellidae) at Sound, 1990. 1991. 1992, and 1995 <3 m (Fig. 7, Table 11). Both polychaetes showed rela- Taxon O/R 1990 1991 1993 1995 tive increases at reference sites over time. For the 2 groups amphipod families, there was relatively few individu- als at either oiled or reference sites in 1990, a peak in 6-20 m infauna abundance at reference sites in 1991, and a peak at Polychaeta 0 263 606 688 692 R 297 590 930 349 oiled sites in 1993. Gastropoda 0 73 171 75 105 R 3 1 79 38 20 Bivalvia 0 87 74 129 117 DISCUSSION R 260 412 541 499 Amphipoda 0 21 40 64 61 General design considerations R 52 54 28 101 Other Crustacea 0 22 23 28 24 There are no pre-spill data for hydrocarbons in sedi- R 10 31 20 33 ments or for benthic invertebrate communities with Echinodermata 0 5 19 13 20 which to compare our post-spill survey data in a BACI R 3 11 16 7 design (Stewart-Oaten 1996). As a result, we rely 6-20 m epifauna largely on post-spill comparisons of oiled versus refer- Polychaeta 0 27 28 39 36 R 21 78 25 23 ence sites to infer impacts. This After, Control-Impact Gastropoda 0 1 11 4 8 design suffers in that making an inference with respect R 1 13 6 2 to impacts, based on differences between oiled and Bivalvia 0 50 39 171 290 reference sites, rests on the assumption that hydrocar- R 3 3 1 1 bon concentrations and benthic communities would Amphipoda 0 6 22 14 10 have been similar in the absence of a spill (Wiens & R 8 14 5 1 Parker 1995, Stewart-Oaten 1996). Obviously, this Other Crustacea 0 5 1 4 10 assumption is untestable, and we can not be certain R 5 12 3 2 that differences between oiled and reference sites Echinodermata 0 1 1 0 <1 R < 1 1 <1 1 were a result of the spill. The best that we can do is to point out when patterns that we observed are consis- e3 rn infauna tent with the hypothesis of spill related impacts, evalu- Polychaeta 0 503 771 810 650 R 200 981 474 192 ate other sources of information (e.g. toxicity studies Gastropoda 0 305 465 263 367 and prior oil spill studies), and make reasonable judg- R 106 344 111 177 ments based on the weight of evidence. A Pattern in Bivalvia 0 174 450 448 333 which oiled and reference sites differed is consistent R 103 163 119 100 with the hypothesis of an oil spill impact. A significant Amphipoda 0 25 37 116 68 interaction between oiling category and year, with ini- R 81 226 96 133 tial differences followed by convergence, is consistent Other Crustacea 0 16 13 49 37 with the hypothesis of initial impacts followed by R 4 39 12 35 recovery. Echinoderrnata 0 1 2 15 6 R 2 3 5 5

<3 m epifauna Hydrocarbons in sediments Polychaeta 0 370 290 729 288 R 80 157 71 925 Gastropoda 0 7 6 14 8 The proportion of spilled oil from the 'Exxon Valdez' R 2 2 3 5 found in the shallow subtidal region (8 to 16% mainly at Bivalvia 0 163 447 729 340 depths <20 m) was similar to that found subtidally from R 13 67 16 4 other major Spills (i.e. 'Amoco Cadiz', 'Tsesis', 'Ixtoc', Amphipoda 0 52 118 644 93 'Argo Merchant') (O'Clair et al. 1996). Our hydrocarbon R 94 233 172 76 data, plus those of Boehm et al. (1995) and O'Clair et al. Other Crustacea 0 3 <1 5 2 (1996), substantiate that the nearshore benthos was R 2 1 2 3 exposed to moderate levels of oiling and that TPAH Echinodermata 0 4 <1 2 7 R 0 0 1 <1 concentrations were generally higher in shallow sub- tidal sediments near oiled shorelines than at ref- 72 Mar Ecol Prog Ser 185: 59-83,1999

Table 9. Summary of randomization ANOVAs for the density of major taxonomic groups of benthic invertebrates. Covariates used in the analyses are listed. Also given are contrasts between oil and reference sites within years for those instances tvhere there was a significant effect of oil category or a significant interaction between oil category and year

Covanate 2-way randomization ANOVA Within year contrasts

Oil category Year Interaction 1990 1991 1993 1995 1

Deep (6-20 m) Infauna Polychaeta Mud • X Gastropoda None 0.0 XX Bivalvia None - - Amphipoda Mud - - Other Crustacea None M Echinodermata None X Epifauna Polychaeta None Gastropoda None Bivalvia None Amphipoda None Other Crustacea Norie Echinodermata None

Eelgrass bed (c3m) lnfauna Polychaeta Mud Gastropoda None Bivalvia None Amphipoda Mud Other Crustacea None Echinodermata None Epifauna Polychaeta None Gastropoda None Bivalvia None Amphipoda None Other Crustacea None Echinodermata None

= greater at oiled sites 0.0, 000, XXX = p < 0 01 0 = greater at control sites 00, 00, XX = 0.01 < p 5 0.05 X = significant difference sites 0, 0, X = 0.05 < p 5 0.1 - =p> 0.1

erence sites. However, relatively high concentrations of It is likely that benthic populations were exposed to TPAHs were also observed at some reference sites and much higher concentrations of hydrocarbons than Lve at least some of these samples had a signature match- measured in the sediments. Our samples were not col- ing 'Exxon Valdez' oil. This suggests that oil from the lected until Summer 1990, approx. 16 mo after the spill, spill may have reached subtidal sediments even at sites and TPAH concentrations in subtidal sediments were relatively far removed from heavily oiled beaches and generally higher in 1989 than in subsequent years, at that inferences made with respect to effects of oil based least in the shallow (<3 m) subtidal (O'Clair et al. on compansons at oiled and reference sites are perhaps 1996). There was considerable vanability between conservative estimates of injury. This is not unreason- sites and depths with respect to temporal trends, but at able given that the mechanism by which oil reaches the 3 m at the oiled sites sampled (Bay of Isles, Hernng subtidal is largely transported from the surface waters Bay, and Sleepy Bay), O'Clair et al. (1996) found an or oiled intertidal sediment on particulates by hydrody- average 64 % decline in TPAH concentrations bettveen narnic forces. Larger differences in biological responses 1989 and 1990. At 6 m depths, TPAH concentrations at oiled and reference sites might be expected, if refer- decreased at 2 sites (71 and 78% at Bay of Isles and ence sites had been totally free of oil from the spill. Hernng Bay), but increased sharply at a third (300% Jewett et al . Oil spill impacts on eelgrass beds 7 3

Table 10. Summary of randomization ANOVAs for the density of dominant benthic invertebrate families at 6 to 20 m. Covanates used in the analyses are listed. Also given are contrasts between oiled and reference sites within years for those instances where there was a significant effect of oil category or a significant interaction between oil category and year

Covariate 2-way randoinization ANOVA Within year contrasts Oil category Year Interaction 1990 1991 1993 1995

lnfauna Polychaeta Ampharetidae Mud Amphictenidae None Capitellidae Mud Lumbrineridae None - - - 0 Maldanidae Mud @@ • - - Nephtyidae h4ud - - - - Opheliidae None 0 - 00 - Orbiniidae Mud Sabellidae h4ud Sigalionidae None Spionidae h4ud - @@ - Syllidae Mud - - • - Gastropoda Caecidae None @@@ @@@ @@@ Olividae Mud Rissoidae None Bivalvia Lucinidae None - - 0 Montacutidae Mud - - - - Tellinidae hlud Thyasiridae None - 00 000 000 Crustacea Oedicerotidae Mud - - - Isaeidae Mud - - - Phoxocephalidae Mud 00 000 - Echinodermata Ophiuroidea None

Epifauna Polychaeta Spirorbidae None - - Bivalvia Mytilidae None @@@ - - - - @@@ @@

= greater at oiled sites @@@, 000, XXX = p < 0.01 0 = greater at control sites @@,00, XX = 0.01 < p 10.05 X = significant difference sites @, 0, X = 0.05 < p 2 0.1 - =p > 0.1

increase at Sleepy Bay). Also, sediments collected from the effort (Carpenter et al. 1991, Mearns 1996) and small sediment traps at Sleepy Bay in 1989 had TPAH con- accidental releases from these boats likely contributed to centrations as high as 28 000 ng g-' and were generally higher hydrocarbon concentrations. Naturally occurring higher than in nearby sediments on the bottom (Short TPAH in subtidal sediments is likely from coal deposits et al. 1996a), suggesting that exposure to oil in sus- east of the Sound; contributions from oil seeps are neg- pended sediments may have been higher than the ligible (Short et al. 1999). There is no way of knowing exposure to deposited sediments. what proportion of the TPAHs observed were from the Some of the hydrocarbon contarnination observed was spiil itself, from spill-related contamination, or from probably the result of spill-related activities as well as naturally occurring sources. However, the fact that coal rather than spilled 'Exxon Valdez' oil (O'Clair et al. 'Exxon Valdez' oil signatures and TPAH levels de- 1996, Short et al. 1999). At the peak of the cleanup and creased substantially over time suggests that most of the monitonng activities, over 1400 vessels were involved in contamination observed was spill or cleanup related. 74 Mar Ecol Prog Ser 185: 59-83, 1999

Table 11 Summary of randomization ANOVAs for the density of dominant benthic invertebrate families within eelgrass beds (<3 m). Covariates used in the analyses are listed. Also given are contrasts between oiled and reference sites within years for those instances where there was a significant effect of oil category or a significant interaction between oil category and year

Covariate 2-way randornization ANOVA Within year contrasts Oil category Year Interaction 1990 1991 1993 1995

lnfauna Polychaeta Amphictenidae None Capitellidae Mud Nereidae Mud Opheliidae Mud Polynoidae Mud Sabellidae None Sigalionidae None Spion.i.dae None Syilidae None Gastropoda Lacunidae None Rissoidae None Trochidae Mud Bivalvia Montacutidae None Tellinidae None Turtoniidae None Crustacea Corophiidae Mud Ischyroceridae None Isaeidae Phoxocephalidae None None Epifauna Polychaeta P XXX 0.. - Spirorbidae - ... Bivalvia None Mytilidae None 0.0 P - •• • • •• •• • • •• Crustacea Caprellidae None - XXX XXX - 0.

= greater at oiled sites 0.0, 000,XXX =p<0.01. 0 = greater at control sites 00, 00, XX = 0.01 < p 2 0.05 X = significant difference sites 0, 0,X = 0.05 < p 2 0.1 - =p>0.1

Effects on the benthos successful in matching sites with similar characteris- tics. While factors other than oil were clearly the most Similarity dendrograms and ordination analyses inlportant determinants of differences in community demonstrated that cornmunities of infaunal and srnall structure, the similarity dendrograms and ordination epifaunal invertebrates were generally Inore similar at techniques indicate sorne effects of the Spill. Bay of the Same sites sampled in different years than at differ- Isles in 1990 had the highest concentrations of TPAHs, ent sites sarnpled in the Same year. In addition, and had a clearly different infaunal cornmunity at 6 to selected pa.irs of sites were generally rnore similar to 20 rn, but not at <3 m, than other sites or years. Several one another than sites from different pairs. These other sites at <3 rn also appeared to be different in analyses suggest that, while all sites were within simi- infauna in 1990 than in subsequent years, and these lar habitats in Western Prince William Sound, there sites were those with exceptionally high concentra- was considerable variability among sites, and that the tions of TPAH and chrysenes in 1990. subtle physical differences between sites were impor- The patterns in community measures (total abun- tant in determining community structure. Analyses dance, total biomass, diversity, and richness) were not also suggest that selection of site pairs was generally indicative of a strong comrnunity-wide response to oil- Jewett et al.: Oil spill impacts on eelgrass beds 75

Fig. 4. Mean density (11 SE) of Mytilidae (bi- Caeudae ~e~th6 to 20 m valves), Caecidae and Lacunidae (gastropods), and Spionidae. Syiiidae. and Nereidae (poly- chaetes) that were representatives of families that were significantly more abundant at oiled sites. (0) Oiied, (0) reference. Two-way ANOVA results are summarized as p-values for compar- isons of oil category (OC), year (Yr), and OC-Yr interaction (Int)

Lacunidae De~lh 3 m ing. This 1s consistent with observations of Dauvin (1982) following the 'Amoco Cadiz' spill. In nearly 3 yr after the spill he found increases in abundance of several benthic groups and declines in amphipods in a muddy-fine-sand community exposed to the 'Amoco Cadiz' oil spill, but no trends in diversity and evenness. In contrast, Sanders et al. (1980) found that there was a Spionidae SMlidae DeDh6 Q 20 m Deplh 6 Q 20 rn reduction in both diversity and total faunal OC4.04. YFO.82. lnM.89 abundance, as well as a clear reduction in the abundance of several major taxonomic groups following the 'Florida' spill (No. 2 fuel oil). The lack of a consistent response with respect to community Parameters and higher order taxonomic groups following the 'Exxon Valdez' and 'Amoco Cadiz' Spills probably reflects the different sensi- Spionidae Nereidae Depth < 3 m tivities of families within the particular DeDih < 3 m faunal group, and the lack of exposure to high enough concentrations of oil to cause acute toxicity among all taxa. Patterns among individual families por- tray a clearer picture of changes that pre- sumably resulted from the spill. Two

Phommphalidae lsaeidae amphipod families, Isaeidae (primarily Deplh 6 Q 20 m Depth 6 1o 20 rn Photis and Protomedeia spp.) and Phoxo- OC=0.01, Yr=0.04, lnt0.08 OC=O 05, YFO 98, lnF0.76 60 lWT cephalidae (mostly Phoxocephalus and Paraphoxus spp.), were consistently more abundant at reference than at oiled sites. Differences were greatest in 1990 through 1993, but persisted to some extent through 1995. This Pattern is consistent with the hypothesis of an initial impact of the spill, followed by partial recovery. Two other lsaeidae amphipod families, Ischyroceridae and Depih c 3 rn OC4.01, YFO.52.lnt0.48 lso~

Fig. 5. Mean density (+I SE) of Phoxocephalidae and Isaeidae (amphipods) that were significantly more abundant at reference sites. (0) Oiled; (0) reference. Two-way ANOVA results are summa- nzed as p-values for comparisons of oii category (OC),year (Yr),and OC-Yr interaction (Int) 76 Mar Ecol Prog Ser 185: 59-83, 1999

Luanidae Lumbnnendae Svlidae Depth=6io20rn Depih=6to20 m Depth 3 m 400 OC=0.05. YM.31. lnb025 40 - OCcO.01. Yd.69. lW.55 OC=0.17. Yd.12. lni=0.07 300 . -"E- - T T

Spirorbidae Depth 3 rn Opheliidae OC=0.52,Yr=0.30, lnN0.01 Depth = 6 to 20 rn 2000- _ OC=0.01. YrQ.01.lnt-023

lschpceridae Depm < 3 rn OC=0.43, Yr-0.21, lnt<0.01 150 - I Thyasiridae Tellinidae iE' Depth=6to20m Depih 3 rn 300 _ üCc0.01. Yd.41. lnH.49 150 - OCQ.O1. Yd.87. Int=0.76 - P a OL- 1989 1990 1991 1992 1993 1994 1995 Z .$ loa - Caprellidae d Depth < 3 rn OC=0.31. Yrc0.01. lntc0.01 0 1 - -, C f 1m

0 Fig. 6. Mean density (t1 SE) of Lucinidae, hlontacutidae, Thyasiridae, and E Tellinidae (bivalves) and Lumbrinendae and Opheliidae (polychaetes) that 3m - were significantly more abundant at reference sites. (0) Oiled; (0) reference. 2 Two-way ANOVA results are summarized as p-values for comparisons of oil 0 o .- - P --P* -P category (OC), year (Yr),and OC-Yr interaction (Int) 1989 1990 1991 1992 1993 1994 1995

Caprellidae, showed Patterns suggestive of an initial Fig. 7. Mean density (+I SE) &of Syllidae and Spirorbidae impact of oiling followed by recovery. These responses (polychaetes) and Ischyroceridae and Capreilidae (amphi- are similar to that observed following other oil spills. pods) for which there were significant interactions. (0) Oiled; (0) reference. Two-way ANOVA results are sumrnarized as p- Massive declines in benthic amphipods were observed values for compansons of oil category (OC), year (Yr), and following the 'Amoco Cadiz' oil spill: 5 amphipod spe- OC-Yr interaction (Int) cies almost totally disappeared from heavily oiled areas subsequent to the spill (Chasse 1978, den Hartog dence from toxicity tests is only mildly supportive of & Jacobs 1980, Jacobs 1980, Dauvin & Gentil 1990) and this hypothesis. Boehm et al. (1995) and Wolfe et al. only showed partial recovery after 2 yr (Dauvin 1982). (1996) tested for toxicity of sediments from the spill Similarly, massive mortality of amphipods were noted area using amphipods in standard laboratory toxicity following other spills (the 'Tsesis' spill: Linden et al. tests in both 1990 and 1991. In 1990, Wolfe et al. (1996), 1979, Kineman et al. 1980, Notini 1980, Elmgren et al. using Ampelisca abdita, found that most intertidal sed- 1983; the 'Irini' spill: Notini 1978), and reductions in irnent samples collected from oiled beaches caused amphipods were noted at the most heavily oiled sites significant toxicity, and mortalities were significantly following the 'Braer' spill off the coast of Scotland greater in sediments from oiled than from reference (Kingston et al. 1997). sites. Similarly, Boehm et al. (1995), using the phoxo- The relative scarcity of am.phipods at oiled sites may cephalid Rhepoxynius abronius,found significant have been the result of acute toxicity of oil, but evi- amphipod mortality in intertidal sediments from sev- Jewett et al.: Oil spill irnpacts on eelgrass beds 77

eral oiled sites in both 1990 and 1991. There was less differences in abundance were due solely to the acute evidence for toxicity in subtidal sediments. Wolfe et al. toxicity of oil. Two representative species, Thyasira (1996) noted that sediments from several subtidal sites, and Macorna, are known to be tolerant of hydrocarbon including all of our oiled sites (Sleepy Bay, Hernng contamination and Thyasira has been shown to reach Bay, and Bay of Isles) were toxic in 1991. However, extremely high densities in grossly oil-contaminated there was also significant toxicity in sediments from conditions (Dando & Southward 1986, Kingston et al. several reference sites, and there were no significant 1995). After the 'Tsesis' fuel oil spill, Elmgren et al. differences in amphipod mortality at oiled versus refer- (1983) found that subtidal Macoma balthica (Tel- ence sites. Boehm et al. (1995) also found little evi- linidae) were highly contaminated with oil (tissue con- dence of toxicity in subtidal sediments, and also found centrations of 2 X 106 ng g-'), but there were no indica- no significant difference in toxicity between oiled and tions of reductions in population abundance. Studies reference sites. Furthermore, reviews of contaminant (in vitro and in situ) have shown that intertidal M. effects on benthic communities (Long & Morgan 1990, balthica were adversely affected by exposure to Prud- Long 1992) suggest that toxicity occurs at TPAH con- hoe Bay crude oil in sediments, but only at much centrations of approx. 4000 ng g-', higher than concen- higher concentrations (>5 X 105 ng g-': Shaw et al. trations we observed in all but a few cases. 1976, Taylor & Kannen 1977, Stekoll et al. 1980) than In spite of the lack of stronger evidence from these we observed approx. 16 mo after the spill. toxicity studies, we suspect that the lower amphipod By far the more prevalent pattern was significantly abundance at oiled sites was at least in part due to greater abundance at oiled sites. Nine polychaete fam- acute toxicity to oil. Several lines of evidence suggest ilies were among those more abundant at oiled sites. this. First, sediments collected from many (but not all) Representatives of most of these, including amph- oiled sites in 1990 were toxic to amphipods (Wolfe et al. ictenids (Gray 1979), nereids (Boesch 1977, Gray 1996). Second, sediments were not tested until nearly 1979), spionids (Grassle & Grassle 1974, Pearson & 16 mo after the spill. TPAH concentrations were gener- Rosenberg 1978, Gray 1979), and syllids (Pearson & ally higher in sediments in 1989 (O'Clair et al. 1996), Rosenberg 1978) have been shown to be tolerant of and the oil in 1989 was less weathered and likely had a organic enrichment and additionally are classified as higher proportion of lighter fractions (Bence & Burns opportunists. More specifically, densities of both 1995) that are generally more toxic (NRC 1985).Third, nereids (Kasymov & Aliev 1973, Hyland et al. 1989) at least some sites in this study had TPAH concentra- and spionids (Sanders et al. 1980, Dauvin 1998) have tions up to 15 000 ng g-', perhaps high enough to cause increased after oil spills elsewhere. Furthermore, Spies mortality in amphipods according to some acute labo- & DesMarais (1983) showed that petroleum was uti- ratory tests. For example, Lee et al. (1977) estimated lized (via bacteria) by the maldanid Praxillella as a car- that the amphipods Gammarus mucronatus and Ampi- bon source in a natural petroleum seep in the Santa thoe valida began to die when hydrocarbon concentra- Barbara Channel, CA, and suggested the petroleum tions in aqueous solutions of crude oil reached 2400 ng was used in sufficient amounts to account for greater g-'. Fourth, laboratory toxicity studies did not use the density of organisms observed in the seep than in a Same species or physical conditions that occurred nat- similar nonseep environment. Peterson et al. (1996) urally in Prince William Sound, and may not reflect reviewed relevant studies from the literature on the actual in situ toxicity levels. Finally, there is strong evi- dual response of toxicity and organic enrichment to dence for mortality to amphipods from other nearshore soft-sediment benthic communities and noted a consis- spills of comparable magnitude (Cabioch et al. 1980, tent pattern of depression of amphipods (as well as den Hartog & Jacobs 1980, Dauvin 1982, Elmgren et al. echinoderms) and an enhancement of polychaetes, 1983, Kingston et al. 1997). It took more than 10 yr for mainly non-selective deposit feeders. Although oil the ampeliscid amphipod population to rebuild to pre- likely was responsible for greater abundance of oppor- 'Amoco cadiz' spill densities (Dauvin 1998). tunists at oiled sites in the present study, caution must We also observed fewer individuals from 3 bivalve be used in attributing oil as the only cause. Numerous families at oiled sites: Thyasiridae and Montacutidae at investigations have shown benthic opportunists to flu- 6 to 20 m, and Tellinidae within the eelgrass bed, orish from organic ennchment of non-petroleum although not nearly as consistently as the amphipods. sources (e.g. Lizarraga-Partida 1974, Pearson & Rosen- The principal genera within these families were the berg 1978, Anderlini & Wear 1992). suspension- or deposit-feeding clams Thyasira, Mys- Some of the faunal increases in oiled habitats were eUa, and Macoma, respectively. Montacutidae showed attributable to the small epifaunal, suspension-feeding a pattern consistent with initial impact followed by mytilid mussel, Musculus spp. Little information is recovery. Other taxa were consistently more abundant available on the response of these organisms to oil, at reference sites through 1995. It is unlikely that the although considerable information is available on the 7 8 Mar Ecol Prog Ser

response of other mytilids. Mytilus edulis and Modio- noted there was an unusually heavy recruitment of Jus demissus have been observed to increase respira- Macoma balthica in areas where subtidal amphipods tion and decrease feeding and assimilation when were virtually eliminated following the 'Tsesis' spill, exposed to crude oil concentrations of 1000 ng g-' and they attnbuted the increases in Macoma to a (Hyland & Schneider 1976). However, M. edulis has reduction in competition with the amphipods. Simi- generally been shown to be extremely hardy and resis- larly, van Bernem (1982) attributed a substantial tant to pollutants, including oil (e.g. Lee et al. 1973, increase in the abundance of oligochaetes on an oiled Farnngton et al. 1982, Ganning et al. 1983, Viarengo & rnudflat to a reduction of corophiid amphipod com- Canesi 1991). petitors. It is likely that many of the tolerant species were able Of all the families that showed a strong indication of to utilize increased carbon sources made available possible enhancement due to oil, only 1, spirorbid from the degradation of oil and shoreline application of polychaetes, showed a Pattern of a reduction in the dif- chemical fertilizers to stirnulate growth of hydrocar- ferences between oiled and reference sites over time bon-utilizing bacteria (). Significantly that corresponded with a sirnilar reduction in differ- higher numbers of hydrocarbon-degrading microor- ences in TPAH concentrations. This suggests that dif- ganisms were found in intertidal and shallow (<20 m) ferences in abundances at oiled and reference sites subtidal sediments foiiowing the 'Exxon Valdez' spill, rnay have been unrelated to the impacts of oil, and and declined between 1990 and 1993 (Braddock et al. instead mag be attributec! to inherent differencec be- 1995, 1996).Increased microbial activity has also been tween sites. It is possible that those processes responsi- observed following other oil Spills (e.g. Colwell et al. ble for heavy oiling of some shorelines are also respon- 1978, Ward et al. 1980, Lizarraga-Partida et al. 1991). sible for the concentration of planktonic larvae and Spies (1987) found that benthic communities were food sources at those sites. Laur & Haldorson (1996) stimulated through enhancement of microbial activity suggested that oiling levels rnay be correlated with when hydrocarbon concentrations in Sediment intersti- high food or larval abundances because sites with high tial waters were in the range of 20 to 500 ng g-'. While rates of water movement were more likely to be these values are not directly comparable to our mea- encountered by the food and larvae, as well as oil, surements of TPAHs in sedirnents, they suggest values transported by currents. Dissolution rates of calcium we might expect based on our measured TPAH con- Sulfate cylinders at oiled and reference intertidal sites centrations (45 to 15 253 ng g-') nearly 16 mo after the in Herring Bay were higher on the oiled site, indicative spill. of greater water movement (R.C. Highsmith, Univer- The increases in the abundance of some infauna, sity of Alaska, pers. comm.). This tends to support the and especially the epifauna, rnay also have resulted hypothesis that some sites rnay have been heavily oiled from an indirect effect of oiling or cleanup on their and had higher densities of selected fauna sirnply predators. Significant reductions of both the crab because of greater water transport. However, such a Telmessus cheiragonus and the sea Star Dermasterias hypothesis cannot easily explain why we observed imbricata tvere noted at oiled sites in 1990 (Dean et al. lower abundances of sorne taxa at oiled sites, and why 1996b). These species feed on a variety of inverte- there were significant site by year interactions. Fur- brates including spirorbids and Musculus spp. (S. C. thermore, this implies that these inherent differences Jewett & T. A. Dean pers. obs.). There was also a sig- between sites rnay have caused us to underestimate nificant reduction in the population of sea otters in impacts. For example, reductions in phoxocephalid heavily oiled parts of the Sound (Ballachey et al. 1994). ampbipods caused by the spill rnay have been even Otters eat a variety of benthic fauna and are consid- greater than indicated by oiled versus reference site ered keystone species that can have a profound effect comparisons. on the structure of nearshore benthic Systems (Kvitek Studies of the effects of oiling and cleanup con- et al. 1992a,b, Estes & Duggins 1995). We did not See ducted in the intertidal Zone have indicated that an increase in the abundance of most prey at cleanup activities, and especially the use of hot-water, oiled sites. However, the relatively small size of our high-pressure washing of oiled shorelines, was detn- sampling units was ill suited for estimating the abun- mental to intertidal communities, and often slowed dance of larger clams such as Saxidomus giganteus recovery of intertidal populations (e.g. De Vogelaere & and Protothaca staminea that are the principal foods of Foster 1994, Driskell et al. 1996, Houghton et al. 1996, sea otters within the Sound (Calkins 1978, Doroff & Lees et al. 1996). It is extrernely difficult to make sirni- Bodhn 1994). lar statements that discrirninate among the effects of It is also possible that some of the increases in abun- oiling, subsequent shoreline cleanup, or damage asso- dance at oiled sites were due to a reduction in compe- ciated with cleanup and monitonng efforts on subtidal tition, especially with amphipods. Elmgren et al. (1983) communities. While cleanup was often localized on a Jewett et al.: 011 spill impacts on eelgrass beds 79

particular shoreline segment, the effects, including The recovery process within benthic infaunal com- nlobilization of oiled sediments and increased boat munities following oil spills generally follows a se- traffic, were often dispersed over a broader subtidal quence of events which is as follows: (I) a toxic effect area, including reference sites. It seems likely that with considerable mortality; (2) an organically en- high-pressure washing of oiled shorelines was respon- riched period in which opportunistic taxa become ex- sible for the redistnbution of oil and oiled sediments tremely abundant; and (3) a period in which oppor- that contributed to high oil concentrations in the subti- tunists decrease in importance and fauna begin to da1 Zone (Mearns 1996). return to conditions similar to adjacent unoiled areas Our experimental design focused on examining and/or to a community characteristic of relatively un- effects of the spill in and adjacent to eelgrass beds disturbed conditions (Pearson & Rosenberg 1978, Gle- throughout the western Sound. Therefore, we used marec & Hussenot 1982, Ibanez & Dauvin 1988, Spies sampling sites (i.e. bays) as experimental units, and et al. 1988). In 1990, 16 mo following the spill, the eel- made statistical inferences to oiled versus reference grass community displayed patterns that were some- sites and not to differences between individual sites. what intermediate between the toxic phase and the However, distribution patterns of impacts among oiled enrichment phase. Since then, we have Seen some pat- sites are of interest because they suggest attributes terns suggestive of recovery. For example, in 1995, that can be used to predict where the most severe TPAH concentrations were low (averaging 1222 ng impacts might occur and might suggest which kinds of g-') and did not differ between oiled and reference sites should be given special consideration when sites. The total abundance and biomass of epifauna at responding to spills. Even within our small sample size <3 m was significantly higher at oiled sites in 1990 and and restricted sampling universe (eelgrass beds near 1991, but not in 1995; and phoxocephalid amphipods moderately to heavily oiled shorelines), there appeared at 6 to 20 m were more abundant at reference sites in to be a high degree of vanability between sites with 1990 and 1991, but not in subsequent years. However, respect to both exposure and impacts. Of the oiled sites for the most part, differences between oiled and refer- we sampled, Bay of Isles had the highest TPAH con- ence sites persisted through 1995, and there was little centrations and appeared to be the most severely indication of convergence. This is consistent with impacted benthic community. This was most evident in observations following the 'Florida' and 'Amoco Cadiz' ordination and clustenng analysis which showed the spills which suggested that communities in muddy- community at Bay of Isles in 1990 to be quite different fine-sand sediments took between 1 (Sanders et al. from the communities at other sites (both oiled and ref- 1980, Ibanez & Dauvin 1988, Dauvin & Gentil 1990) erence sites). Armstrong et al. (1995) also found that and 2 (Dauvin 1998) decades to fully recover. the tissues of bivalves from Bay of Isles had higher TPAH concentrations than those from elsewhere, and Acknowledgements. We thank the more than 20 divers who that flathead sole frorn this site had among the highest helped collect the data, often under adverse conditions; those concentrations of fluorescent aromatic compounds who assisted in the sorting and taxonomy of benthic inverte- (indicators of hydrocarbon exposure). Furthermore, brates, including Kathy Omura and Max Hoberg; and those Armstrong et al. (1995) found dead Tanner crabs at this who helped with statistical issues, data management, and and one other oiled site (Snug Harbor, western Prince analyses, especially Lyman McDonald and Dennis Jung. Financial Support was provided by the U.S.Forest Service, William Sound) in Spring 1990, but not at other oiled or the Alaska Department of Fish & Game, and the 'Exxon reference sites, and speculated that the deaths may Valdez' Oil Spill Trustee Council. The findings and conclu- have been caused by interactive effects of low salinity sions presented, however, are ours, and do not necessarily and oiling. The extent of shoreline oiling was no reflect the views or positions of the Trustee Council. greater at Bay of Isles than at other oiled sites we sam- pled. However, Bay of Isles, and especially the extreme LITERATURE CITED western portion of the Bay where we sampled, was particularly well protected from waves as evidenced ADEC (Alaska Department of Environmental Conservation) by finer sediments observed there. Oil appears to be (1989) Impact maps and Summary reports of shoreline sur- most persistent in these habitats (Page et al. 1995), and veys of the Exxon Valdez oil spill site-Prince William Sound. Report to the Exxon Valdez Oil Spill Trustee Coun- we suspect that the combination of heavy oiling and a cil. Anchorage. AK low energy environment made this location particu- Anderlini VC, Wear RG (1992) The effects of sewage and nat- larly susceptible to adverse irnpacts of oil. This sug- ural seasonal disturbances on benthic macrofaunal com- gests that responders should focus on keeping oil from munities in Fitzroy Bay. Wellington, New Zealand. Mar these protected habitats, and cleanup efforts should be Pollut BuU 24:21-26 Armstrong DA, Dinnel PA, Orensanz JM, Armstrong JL, conducted with an awareness of the special sensitivity McDonald TL, Cusimano RF, Nemeth RS, Landolt ML, of these environments. Skalski JR, Lee RF, Huggett RJ (1995) Status of selected 80 Mar Ecol Prog Ser

bottomfish and crustacean species in Pnnce William the Amoco Cadiz oil spa. Mar Poilut Bull 11:298-301 Sound following the Exxon Valdez oil spill. In: Welis PG, Clark RC Jr, MacLeod WD (1977) Inputs, transport mecha- Butler JN, Hughes JS (eds) Exxon Valdez oil spill: fate and nisms, and observed concentrations of petroleum in the effects in Alaskan waters, ASTM STP 1219. Amencan manne environment. In. Malins DC (ed) Effects of petro- Society for Testing and Materials, Philadelphia, PA, leum on arctic and subarctic manne environments and P 485-547 organisms. Academic Press, New York, p 91-224 Ballachey BE, Bodkin JL, DeGange AR (1994) An overview of Clifford HT. Stephenson W (1975) An introduction to nurnen- sea otter studies in Pnnce Wiiliam Sound. In: Loughlin TR cal classification. Academic Press, New York (ed) Marine mammals and the Exxon Valdez. Academic Colweii RR. Mills AL, Walker JD, Garcia-Tello P, Campose PV Press. San Diego, p 47-59 (1978) Microbial ecology studies of the Metrlla spill in the Bassin NJ, Ichiye T (1977) Flocculation behavior of sediment Straits of Magellan. J Fish Res Board Can 35.573-580 and oil emulsions. J Sed Petrol 47:671-677 Conover RJ (1971) Some relations between zooplankton and Bence AE, Burns WA (1995) Fingerpnnting hydrocarbons in bunker C oil in Chedabucto Bay following the wreck of the biological resources of the Exxon Valdez spiil area. In: the tanker Arrow. J Fish Res Board Can 28:1327-1330 Wells PG, Butler JN. Hughes JS (eds) Exxon Valdez oil Cross WE. Martin CM, Thomson DH (1987) Effects of expen- spill- fate and effects in Alaskan waters, ASTM STP 1219. mental releases of oil and dispersed oil on Arctic American Society for Testing and Materials, Philadelphia, nearshore macrobenthos. 2. Epibenthos. Arctic 40 (Suppl PA, P 84-140 1):201-210 Boehrn PD, Steinhauer MS, Green DR, Fowler B, Humphrey Dando PR, Southward AJ (1986) Chromatography in bivalve B, Fiest DL, Cretney WS (1987)Comparative fate of chem- molluscs of the genus Thyasira. J Mar Bio1 Assoc UK 66: ically dispersed and beached crude oil in subtidal sedi- 915-929 xcxts of the arctic nearshore. .4rctic 00 (Suppl 1),13.1-148 Dauvin JC 119821 Impact of Amoco Cadiz oil spiil on the Boehm PD, Page DS, Gilfillan ES, Stubblefield WA, Harner EJ muddy fine sand Abra alba and Meiinna palmata commu- (1995) Shoreline ecology program for Pnnce William nity from the Bay of Morlaix. Estuar Coast Shelf Sci 14: Sound, Alaska, following the Exxon Valdez oil spill: part 517-531 2-chemistry and toxicology. In: Wells PG, Butler JN, Dauvin JC (1998) The fine sand Abra alba community of the Hughes JS (eds) Exxon Valdaz oil spill: fate and effects in Bav of Morlaix twenty years after the Amoco Cadiz oil Alaskan waters, ASTM STP 1219. American Society for spiil. Mar Pollut Buil36:669-676 Testing and Materials, Philadelphia, PA, p 347-397 Dauvin JC, Gentii F (1990) Conditions of the peracand popu- Boesch DF (1977) A new look at the zonation of benthos along lations of subtidal comrnunities in northern Bnttany ten the estuanne gradient. In: Coull BC (ed)Ecology of manne years atter the Amoco Cadiz oil spffl Mar Pollut Bull 21 benthos. University of South Carolina Press, Columbia, 123-130 p 245-266 Dean TA, Stekoll MS, Smith RO (1996a) Kelps and oil: the Boström C, Bonsdorff E (1997) Community structure and spa- effects of the Exxon Valdez oil spill on subtidal algae. In: tial vanation of benthic invertebrates associated with Rice SD, Spies RB, Wolfe DA, Wnght BA (eds) Proceedings Zostera marina (L) in the northern Baltic Sea. J Sea Res of the Exxon Valdez oil spill symposium. Am Fish Soc 37:153-166 Symp 18 412-423 Braddock JF, Lindstrom JE, Brown EJ (1995) Distribution of Dean TA, Jewett SC, Laur DR, Smith RO (1996b) Injury to hydrocarbon-degrading microorganisms in sediments epibenthic invertebrates resulting from the Exxon Valdez from Prince Wiiham Sound, Alaska following the Exxon oil spill. In: Rice SD, Spies RB, Wolfe DA. Wnght BA (eds) Valdez oil spiil. Mar Pollut BuU30:125-132 Proceedings of the Exxon Valdez oil spill symposium. Am Braddock JF, Lindstrom JE. Yeager TR, Rasley BT, Brown EJ Fish Soc Symp 18:424-439 (1996) Patterns of microbial activity in oiled and unoiled Dean TA, StekoU MS, Jewett SC, Smith RO. Hose JE (1998) sediments in Pnnce William Sound In: Rice SD, Spies RB, Eelgrass (Zostera marina L.) in Pnnce William Sound. Wolfe DA, Wnght BA (eds) Proceedings of the Exxon Alaska. effects of the Exxon Valdez oll spill. Mar Poiiut Valdez oil spill symposium. Am Fish Soc Symp 18:94-108 Bull 36.201-210 Bray JR, Curtis JT (1957) An ordination of the upland forest den Hartog C, Jacobs RPWM (1980) Effects of the Amoco communities of southern Wisconsin. Ecol Monogr Cadiz oil spill on an eelgrass community at Roscoff 27:325-349 (France) with special reference to the mobile benthic Cabioch L, Dauvin JC, Mora Bermudez J, Rodnguez Babio C fauna Helgol Meeresunters 33:182-191 (1980) Effects of the 'Amoco Cadiz' oil spill on the sublit- De Vogelaere AP, Foster MS (1994) Damage and recovery in toral benthos, north of Brittany. Helgol Meeresunters intertidal Fucus gardneri assemblages following the 33:192-208 Exxon Valdez oil spill. Mar Ecol Prog Ser 106:263-271 Calkins, DG (1978) Feeding behavior and rnajor prey species Doroff AM. Bodkin JL (1994) Sea otter foraging behavior and of the sea otter, Enhydra lutris, in Montague Strait, Pnnce hydrocarbon levels in prey. In: Loughlin T (ed) Manne William Sound, Alaska. Fish Bull 76125-131 mammals and the Exxon Valdez. Academic Press, San Carlson PR, Kvenvolden KA (1996)Tracking Exxon Valdez oil Diego, p 193-208 from beach to deep water sediments of Pnnce William Dnskeii WB, Fukuyama AK, Houghton JP, Lees DC, Mearns Sound, Alaska. In: Rice SD, Spies RB, Wolfe DA, Wnght AJ, Shigenaka G (1996) Recovery of Pnnce William Sound BA (eds) Proceedings of the Exxon Valdez oil spill sympo- intertidal infauna from Exxon Valdez oiling and shoreline sium. Am Fish Soc Symp 18:109-120 treatments. 1989 through 1992. In: Rice SD, Spies RB, Carpenter AD, Dragnich RG, Srnith MT (1991) Manne opera- Wolfe DA, Wnght BA (eds) Proceedings of the Exxon tions and logistics dunng the Exxon Valdez spill cleanup. Valdezoil spill symposium. Am Fish Soc Symp 18:362-378 In. Proceedings of the 1991 International oil spiil confer- Eberhardt LL, Thomas JM (1991) Designing environmental ence prevention, behavior, control, cleanup. Amencan field studies. Ecol Monogr 61:53-73 Petroleum Institute. Washington, DC, p 325-331 Elmgren R, Frithsen JB (1982) The use of experimental Chasse C (1978) The ecological impact on and near shores by ecosystems for evaluating the environmental impact of Jewett et al.. Oll spill impacts on eelgrass beds 81

pollutants: a comparison of an oil spill in the Baltic Sea and epibiota from Exxon Valdez oiling and shoreline treat- 2 long-term, low-level oil addition experirnents in meso- ments. 1989 through 1992. In. Rice SD, Spies RB, Wolfe cosms. In: Gnce GD, Reeve MR (eds) Marine rnesocosins: DA, Wright BA (eds) Proceedings of the Exxon Valdez oil biological and chemical research in experimental ecosys- spill symposium Am Fish Soc Symp 18:379-411 tems. Springer Verlag, Inc, New York, p 153-165 Hyland JL, Schneider ED (1976) Petroleun? hydrocarbons and Elmgren R, Hansson S, Larsson U, Sundelin B, Boehm PD their effects on marine organisms, populations, coinmuni- (1983) The Tsesis oil spill: acute and long-term impact on ties and ecosystems In: Sources, effects and sinks of the benthos. Mar Biol 73:51-65 hydrocarbons in the aquatic environment. Symposium Estes JA, Duggins D0 (1995) Sea otters and kelp forests in proceedings. American Institute of Biological Sciences. Alaska: generality and variation in a community ecologi- American University, Washington, DC, p 464-506 cal paradigm. Ecol Monogr 65:75-100 Hyland JL, Kennedy J, Campbell J, Williains S, Boehm P, Farrington JW, Davis AC, Frew NM, Rabin KS (1982) No 2 Uhler A. Steinhauer W (1989)Environmental effects of the fuel oil compounds in A4ytilus edulis: retention and release Pac Baroness oil and copper spill. In: Proceedings of the after an oil spill. Mar Biol 66:15-26 1989 oil spill conference. API, EPA, and USCG (sponsors), Feder HM, Blanchard A (1998) The deep benthos of Prince San Antonio. TX. p 413-419 William Sound, Alaska. sixteen months after the Exxon Ibanez F, Dauvin JC (1988) Long-term changes (1977-1987) in Valdezoil spill. Mar Pollut Bull36:118-130 a muddy fine sand Abra alba-Mehnna palmata commu- Feder HM, Jewett SC (1987) The subtidal benthos. In: Hood nity from the Western English Channel: multivanate time- DW, Zimmerman ST (eds) The Gulf of Alaska: physical series analysis. Mar Ecol Prog Ser 49:65-81 environment and biological resources, Ocean Assessment Jacobs RPWM (1980) Effects of the Amoco Cadiz oil spill on Div., Alaska Office, U. S. Minerals Management Service, the seagrass community at Roscoff with special reference Alaska OCS Region, MMS 86-0095, U.S. Govt. Printing to the benthic infauna. Mar Ecol Prog Ser 2:207-212 Office, Washington, DC, p 347-396 Jewett SC, Dean TA, Laur DR (1996) The effects of the Exxon Field JG. Clarke KR,Warwick RM (1982) A practical strategy Valdez oil spill on benthic invertebrates in an oxygen- for analyzing multi-species distnbution Patterns. Mar Ecol deficient embayment in Pnnce William Sound. Alaska. In: Prog Ser 8:37-52 Rice SD, Spies RB, Wolfe DA. Wnght BA (eds) Proceedings Folk RL (1980) Petrology of sedimentary rocks. Hemphil Pub- of the Exson Valdez oil spili symposium. Am Fish Soc lishing Co, Austin, TX Symp 18:440-447 Ganning B. Broman D, Lindblad C (1983) Uptake of petro- Kasymov AG, Aliev AD (1973) Experimental study of the leum hydrocarbons by the blue mussel (Mytilus edulis L.) effect of oll on some representatives of benthos in the after experimental oiling and high pressure, hot water Caspian Sea. Water Air Soil Pollut 2:235-245 shore cleaning. Mar Environ Res 10:245-254 Kelso DD, Kendziorek M (1991) Alaska's response to the Gilfillan ES, Page DS, Harner EJ, Boehm PD (1995) Shoreline Exxon Valdez oil spill. Environ Sci Techno1 25:16-23 ecology program for Prince William Sound, Alaska, fol- Kineman JJ, Elmgren R, Hansson S (1980) The Tsesis oil lowing the Exxon Valdez oil spiii: part 3-Biology. In: spill U.S. Dept. of Commerce, Office of Weils PG, Butler JN, Hughes JS (eds) Exxon Valdez oil Assessment, National Oceanic and Atmospheric Adminis- spiil: fate and effects in Alaskan waters, ASTM STP 1219. tration, Boulder, CO American Society for Testing and Materials, Philadelphia, Kingston PF, Dixon IMT, Harnilton S, Moore DC (19951 The PA, p 398-443 impact of the Braeroil spill on the macrobenthic infauna of Glemarec M, Hussenot E (1982) A three-year ecological sur- the Sediments off the Shetland Islands. Mar Pollut BuU vey in Benoit and Wrac'h Abers following the Amoco 30(7):445-459 Cadiz oil spill. Neth J Sea Res 16:483-490 Kingston PF, Dixon IMT, Hamilton S, Moore CG, Moore DC Grassle JF, Grassle JP (1974) Opportunistic life histones and (1997) Studies on the response of intertidal and subtidal genetic systerns in marine benthic polychaetes. J Mar Res manne benthic communities to the Braer oil spill In: 32:253-284 Davies JM, Topping G (eds) The impact of an oil spill in Grassle JF, Elmgren R, Grassle JP (1981) Response of benthic turbulent waters: the Rraer. The Stationery Office, Edin- communities in MERL experimental ecosystems to low burgh. p 209-233 level, chronic additions of No. 2 fuel oil. Mar Environ Res Kvitek RG, Oliver JS (1992a) Influence of sea otters on soft- 4:279-297 bottom prey comnlunities in southeast Alaska. Mar Ecol Gray JS (1979)Pollution-induced changes in populations. Phi1 Prog Ser 82:103-113 Trans Royal Soc Lond Bull286:545-561 Kvitek RG, Oliver JS (1992b) Changes in Alaska soft-bottom Grebmeier JM, Feder HM, McRoy CP (1989) Pelagic-benthic prey cornmunities along a gradient in sea otter predation. coupling on the shelf of the northern Benng and Chukchi Ecology 73:4 13-428 Seas. 11. Benthic community structure. Mar Ecol Prog Ser Laur D, Haldorson L (1996) Coastal habitat studies: the effect 51:253-268 of the Exxon Valdez oil spiii on shallow subtidal fishes in Green RH (1979) Sampling design and statistical methods for Prince William Sound. In: Rice SD. Spies RB, Wolfe DA, environmental biologists. John Wiley & Sons, New York Wnght BA (eds) Proceedings of the Exxon Valdez oil spill Gundlach ER, Boehm PD. Marchand M, Atlas RM, Ward DM, symposium. Am Fish Soc Symp 18:659-670 Wolfe DA (1983) The fate of the Amoco Cadizoil spill. Sci- Lee RF, Sauerheber R, Benson AA (1973) Petroleum hydro- ence 221:122-131 carbons: uptake and discharge by the manne mussel Highsmth RC, Rucker TL, Stekoll MS, Saupe SM. Lindeberg Mytilus edulis. Science 177:344-346 MR, Jenne RN, Enckson WP (1996) Impact of the Exxon Lee WY, Welch MF, Nicol JAC (1977) Suwival of two species Valdez oil spill on intertidal biota. In: Rice SD, Spies RB, of amphipods in aqueous extracts of petroleum oils. Mar Wolfe DA, Wnght BA (eds) Proceedings of the Exxon Pollut Bull 8:92-94 Valdezoil spill symposium. Am Fish Soc Symp 18:212-237 Lees DC, Houghton JP. Driskell WB (1996) Short-term effects Houghton JP, Lees DC, Driskell WB, Lindstrom SC, Mearns of several types of shoreline treatment on rocky intertidal AJ (1996) Recovery of Prince William Sound intertidal biota in Prince William Sound. In: Rice SD, Spies RB. Wolfe 82 Mar Ecol Prog Cer

DA, Wright BA (eds) Proceedings of the Exxon Valdez oil relation to organic ennchment and pollution of the marine spill symposium. Am Fish Soc Symp 18:329-348 environment. Oceanogr Mar Rio1 Annu Rev 16:229-311 Linden 0, Elmgren R, Boehm PD (1979) The Tsesisoil spill- Peterson CH, Kennicutt MC 11, Green RH, Montagna P, its irnpact on the coastal ecosystem of the Baltic Sea Harper DE Jr, Powell EN, Roscigno PF (1996) Ecological Ambio 8:244-253 consequences of environmental perturbations associated Lizarraga-Partida ML (1974) Organic pollution in Ensenada with offshore hydrocarbon product.ion: a perspective on Bay, Mexico. Mar Pollut Bull 5:109-112 long-term exposures in the Gulf of Mexico. Can J Fish Lizarraga-Partida ML, Izquierdo-Vicuna FB, Wong-Chang I Aquat Sci 53:263?-2654 (1991) Manne bactena on the Campeche Bank oil field. Phiiiips RC (1984) The ecology of eelgrass meadows in the Mar Poilut Bull22:401-405 Pacific Northwest: a cornrnunity profile. U.S. Fish & Long ER (1992) Ranges in chemical concentrations in sedi- Wildlife Service OBS 84/24 ments associated with adverse biological effects. Mar Pol- Rhoads DC (1974) Organism-sediment relations on the lut BUH 24:38-45 muddy sea floor. Oceanogr Mar Biol Annu Rev Long ER, Morgan LG (1990) The potential for biological 12:263-300 effects of sediment-sorbed contaminants tested in the Sanders HL, Grassle JF, Hampson GR, Morse LS, Garner- National Status and Trends Program. NOAA Tech Memo Pnce S, Jones CC (1980) Anatomy of an oil spill. long-term NOS OMA 52, NOAA, Seattle effects from the grounding of the barge Flonda off West- Maki AW (1991) The Exxon Valdez oil spiil: initial environ- Faimouth, Massachusetts J Mar Res 38:265-380 mental impact assessment. Environ Sci Technol 25:24-29 Shannon CE, Weaver W (1963) The mathematical theory of Manly B (1991) Randornization and Monte Carlo methods in communication. University of IUinois Press, Urbana biology. Chapman and Haii, London Shaw DG, Paul AJ, Cheek LM, Feder HM (1976) Macoma Mapstone BD (1995) Sca!eab!e decision ru!es for environmen- halfhica: an indicator of nil polliition. Mar Pollut Bull tal impact studies: effect size, Type I and Type I1 errors. 7(2):29-31 Ecol Appl5:401-410 Short JW, Sale DM, Gibeaut JC (1996a) Nearshore transport McConnaughey T, McRoy CP (1979) I3C labe1 identifies eel- of hydrocarbons and sediments after the Exxon Valdez oil grass (Zostera marina) carbon in an Alaskan estuanne spiil. In: Rice SD, Spies RB, Wolfe DA, Wright BA (eds) food weh. Mar Biol53:263-269 Proceedings of the Exxon Valdez oil spill symposium. Am McRoy CP (1970) Standing stocks and other features of eel- Fish Soc Symp 18.40-60 grass (Zostera marina) populations on the coast of Alaska. Short JW, Jackson TJ. Larsen ML, Wade TL (1996b) Analyti- J Fish Res Board Can 27:1811-1821 cal methods used for the analysis of hydrocarbons in crude Mearns AJ (1996) Exxon Valdez shoreline treatment and oil, tissues, sediments, and seawater collected for the Nat- operations: implications for response, assessment. rnoni- ural Resources Damage Assessment of the Exxon Valdez toring. and research. In: Rice CD, Spies RB, Wolfe DA, Oil Spiii. In: Rice SD, Spies RB, Wolfe DA, Wright BA (eds) Wright BA (eds) Proceedings of the Exxon Valdez oil spüi Proceedings of the Exxon Valdez oil spill symposium. Am symposium. Am Fish Soc Symp 18:309-328 Fish Soc Symp 18:140-148 Meyers PA, Quinn JG (1973) Association of hydrocarbons and Short JW, Kvenvolden KA, Carlson PR, Hostettler FD, Rosen- mineral particles in saline solution. Nature 244:23-24 bauer RJ, Wright BA (1999) Natural hydrocarbon back- NRC (National Research Council) (1985) Oil in the sea: inputs, ground in benthic sedirnents of Pnnce William Sound, fates, and effects. National Academy Press, Washington, Alaska: oil vs coal Environ Sci Technol 33:34-42 DC Spies RB (1987) The biological effects of petroleum hydro- Notini M (1978) Long-term effects of an oil spill on Fucus carbons in the sea: assessments from the field and mi- macrofauna in a smaii Baltic bay. J Fish Res Board Can crocosms. Chapter 9. In: Boesch DF. Rabalais NN (eds) 35:?45-753 Long-term environmental effects of offshore oil and Notini M (1980) Impact of oil on the littoral ecosystem- gas developrnent. Elsevier Applied Science, London, effects on Fucus macrofauna. In: Kineman JJ, Eimgren R, P 41 1-467 Hansson S (eds) The Tsesis oil spiil. National Oceanic and Spies RB, DesMarais DJ (1983) Natural isotope study of Atmospheric Administration, Boulder, CO, p 129-145 trophic ennchment of marine benthic communities by O'Clair CE, Short JW, Rice SD (1996) Contamination of inter- petroleum seepage. Mar Biol 73:67-71 tidal and subtidal sediments by oil from the Exuon Valdez Spies RB, Hardin DD, Toal JP (1988) Organic ennchment or in Pnnce William Sound. In: Rice CD, Spies RB, Wolfe DA, toxicity: a comparison of the effects of kelp and crude oil in Wnght BA (eds) Proceedings of the Exxon Valdez oil spill sediments on the colonization and growth of benthic symposium. Am Fish Soc Symp 18:61-93 infauna. J Exp Mar Biol Ecol 124:261-282 Olsgard F, Gray JS (1995) A cornprehensive analysis of the Spies RB, Rice SD. Wolfe DA, Wright BA (1996) The effects of effects of offshore oil and gas exploration and production the Exxon Valdez Oil Spill on the Alaskan coastal environ- on the benthic communities of the Norwegian continental ment. In: Rice CD, Spies RB, Wolfe DA. Wright BA (eds) shelf. Mar Ecol Prog Ser 122:277-306 Proceedings of the Exxon Valdez oil spiü symposium. Am Owens EH, Harper JR, Robson W, Boehm PD (1987) Fate and Fish Soc Symp 18:l-16 persistente of crude oil stranded on a sheltered beach. Stekoll MS, Clement LE, Shaw DG (1980) Sublethal effects of Arctic 40 (Suppl 1):109-123 chronic exposure on the intertidal clam Macoma balthica. Page DS, Boehm PD, Douglas GS, Bence AE (1995) Identifica- Mar Biol 57:51-60 tion of hydrocarbon sources in the benthic sedunents of Stekoll MS, Deysher L, Highsmith RC, Saupe SM, Guo Z, Prince Wiüiam Sound and the Gulf of Alaska foilowing the Erickson IW, McDonald L, Snckland D (1996) Coastal Exxon Valdez oil spüi. In: Wells PG, Butler JN, Hughes JS habitat injury assessment: intertidal comrnunities and the (eds) Exvon Valdez oil spiü: fate and effects in Alaskan Exxon Valdez oil spill. In: Rice SD, Spies RB. Wolfe DA, waters, ASTM STP 1219. American Society for Testing Wnght BA (eds) Proceedings of the Exxon Valdez oil spill and Materials, Philadelphia, PA, p 41-83 symposium. Am Fish Soc Symp 18:177-192 Pearson TH, Rosenberg R (1978) Macrobenthic succession in Stewart-Oaten A (1996) Problems in the analysis of environ- Jewett et al. Oil spill impacts on eelgrass beds 83

mental monitonng data. In: Schmitt RJ, Osenberg CW perylene in marine sediments-a review. Mar Chem (eds) Detecting ecological impacts. Academic Press, San 25:l-27 Diego, p 109-131 Viarengo A, Canesi L (1991) Mussels as biological indicators Taylor TL, Kannen JF (1977) Response of the clam. Macoma of pollution. Aquaculture 94:225-243 balthica (Linnaeus), exposed to Prudhoe Bay crude oil as Ward DM, Atlas RM, Boehm PD, Calder JA (1980) Microbial unmixed oil, water-soluble fraction, and oil-contaminated biodegradation and chernical evolution of oil from the sediment in the laboratory. In: Wolfe DA (ed) Fate and Amoco spill. Ambio 9:277-283 effects of petroleum hydrocarbons in the marine eco- Warwick RM, Clarke KR (1993) Comparing the seventy of Systems and organisms. Pergamon Press, New York. disturbance: a meta-analysis of marine macrobenthic p 229-237 community data. blar Ecol Prog Ser 92:221-231 Thayer JK, Phillips RC (1977) Importance of eelgrass beds in Wiens JA, Parker KR (1995) Analyzing the effects of acciden- Puget Sound. Mar Fish Rev 36:27-39 tal environmental impacts: approaches and assumptions. van Bernem KH (1982) Effects of experimental crude oil con- Ecol Appl5:1069-1083 tamination on abundance, mortality and resettlement of Wolfe DA, and 11 CO-authors(1994) The fate of oil spilled from representative mudflat organisms in the mesohaline area the Exxon Valdez. Environ Sci Techno1 28:561A-568A of the Elbe Estuary. Neth J Sea Res 16:538-546 Wolfe DA, Krahn MM, Casiilas E. Sol S, Thomas TA, Lunz J, van Tamelen PG, Stekoll MS, Deysher L (1997) Recovery pro- Scott KJ (1996) Toxicity of intertidal and subtidal sedi- cesses of the brown alga Fucus gardnerj following the ments in contaminated by the Exxon Valdez oil spili. In: 'Exxon Valdez' oil spiil: settlement and recruitment. Mar Rice SD, Spies RB, Wolfe DA, Wnght BA (eds) Proceedings Ecol Prog Ser 160:265-277 of the Exxon Valdez oil spiil symposium. Am Fish Soc Venkatesan MI (1988) Occurrence and possible sources of Symp 18:121-139

Editorial responsibility: Charles Peterson (Contnbuting Submitted: February 18, 1998; Accepted: December 14, 1998 Editor), Morehead City, North Carolina, USA Proofs received from author(s). July 30, 1999