Fisheries and Oceans Canada (DFO) Annex 6

PDF Page

Lee, K., Wohlgeschaffen, G., Trembaly, G., Vandermeulen, J., Mossman, D., Wilson, J., 2 Doe, K., Jackman, P., Prince, R., Garrett, R., and Haith, C. 1999. Natural recovery reduces impact of the 1970 Arrow oil spill. Proceeding of the 1999 International Oil Spill Conference (pp. 1075-1078). Savannah, GA, US: IOSC.

Lessard J., Campbell, A., Zhang, Z., MacDougall, L., and Hankewich, S. 2007. Recovery 6 potential assessment for the northern abalone (Haliotis kamtschatkana) in Canada. Canadian Science Advisory Secretariat. Research Document. 2007/061. 101 pp.

Li, Z., Lee, K., King, T., Boufadel, M., and Venosa, A. 2008. Assessment of chemical 113 dispersant effectiveness in a wave tank under regular non-breaking and breaking wave conditions. Marine Pollution Bulletin, 56(5), 903-912.

Li, Z., Lee, K., King, T., Kepkay, P., Boufadel, M., and Venosa, A. 2009b. Evaluating chemical 123 dispersant efficacy in a wave tank: part 1-dispersant effectiveness as a function of energy dissipation rate. Environmental Engineering Science, 26(6), 1139-1148.

Li, Z., Lee, K., King, T., Boufadel, M., and Venosa, A. 2009c. Evaluating chemical dispersant 134 efficacy in a wave tank: part 2-significant factors determining in situ oil droplet size distribution. Environmental Engineering Science, 26(9), 1407-1418.

Li, Z., Lee, K., King, T., Boufadel, M., and Venosa, A. 2009a. Evaluating crude oil chemical 146 dispersion efficacy in a flow-through wave tank under regular non- breaking wave and breaking wave conditions. Marine Pollution Bulletin, 58(5), 735-744.

Milliman J.D. 1980. Sedimentation in the Fraser River and its estuary, Southwestern 166 British Columbia. Estuarine and Coastal Marine Science 10 609-633.

Mizroch, S.A., Rice, D.W., Zwiefelhofer, D., Waite, J., and Perryman, W.L. 2009. 191 Distribution and movements of fin whales in the North Pacific Ocean. Mammal Review. 39(3):193-227.

Morano, J.L., Rice, A.N., Tielens, J.T., Estabrook, B.J., Murray, A., Roberts, B.L., and Clark, 226 C.W. 2012. Acoustically detected year-round presence of right whales in an urbanized migration corridor. Conserv. Biol. 26(4):698-707.

National Academies of Sciences (NAS). 2016. Spills of diluted bitumen from pipelines: a 236 comparative study of environmental fate, effects, and response. Washington, DC, the National Academies Press, ISBN: 978-0-309-38010-2.

001 NATURAL RECOVERY REDUCES IMPACT OF THE 1970 ARROWOM SPILL

Kenneth Lee, Gary D. Wohlgeschaffen, and Gilles H. Tremblay Fisheries and Oceans Canada Maurice Lamontagne Institute P.O. Box 1000 Mont-Joli, Quebec G5H 3Z4, Canada

J. H. Vandermeulen and D. C. Mossman Kenneth G. Doe and P. M. Jackman Fisheries and Oceans Canada, Environmental Science Centre, Environment Canada Bedford Institute of Oceanography Moncton, New Brunswick, Canada Dartmouth, Nova Scotia, Canada R. C. Prince, R. M. Garrett, and C. E. Haith J. E. H. Wilson Exxon Research and Engineering Company Morgan State University, Department of Biology Annandale, New Jersey Baltimore, Maryland

ABSTRACT: In 1970 the tanker Arrow ran aground releasing 2,000 m3 of Bunker C crude oil along 300 km of Nova Scotia's coastline, of which only 10% was subjected to cleanup, the rest was left to degrade naturally. Sediment and interstitial water collected in 1993 and 1997 from Black Duck Cove in Chedabucto Bay, a representative untreated site, showed that the remaining residual oil has undergone substantial biodegradation. The environmental significance of this intrinsic remediation process was assessed with a battery of microscale biotests: CYP1A and mixed function oxygenäse induction in winter flounder, Amphipod Sur- vival, Echinoid Fertilization, Grass Shrimp Embryo-Larval Tox- icity, Microtox® Solid-Phase and 100% Tests. While much oil remains in the sediment (426-12,744 ppm), results of the biotests show that it is of low toxicity and habitat recovery is evident from the level ofbenthic diversity.

Introduction

In February 1970 the tanker Arrow grounded off Nova Scotia, Figure 1. Sampling Sites at Black Duck Cove, Nova Scotia Canada. Of the 11,000 m3 of Bunker C fuel oil released into the (♦ 1993 sediment collection sites for fish-exposure studies; sea, 2,000 m3 impacted 300 km of coastline (Sergy and Owens, 1997 collection sites #1-5). 1993). Only 10% of the shoreline was subjected to cleanup procedures, the rest was left to nature and is now evident as asphalt was determined with a flame ionization detection (FID) gas pavements, stain on rocks, surface sheen, and liquid reserves Chromatograph. Each of the five sediments used in the 1997 bio- trapped within sediments at some low-energy sites, such as the tests were extracted with dichloromethane for gas chromatogra- lagoon in Black Duck Cove (Figure 1). To assess the risk of natu- phy/mass spectroscopy (GC/MS) analysis to quantify the rela- ral attenuation (no treatment) as an oil spill cleanup strategy, tively easily degradable normal alkanes and the more resistant chemical characterization and a battery of biotests were con- PAHs. ducted in 1993 and 1997 on samples of contaminated sediment Biological effects assessment of residual oil. Mixed function and interstitial water recovered from five sites in this cove. oxygenase (MFO) systems in organisms metabolize foreign molecules for excretion. MFO induction in fish has become a standard bioassay to diagnose environmental contaminants in Methods freshwater and marine ecosystems (Goks0yr and Förlin, 1992). Acclimated winter flounder (Pleuronectes americanus) from a Chemical analysis. Residual hydrocarbon concentration in the clean site were exposed to the contaminated sediment (850 ppm Black Duck Cove sediment used for the fish-exposure experiment oil residue) and a clean sand control. On days 0 (pre-exposure), 1,

1075

002 1076 1999 INTERNATIONAL OIL SPILL CONFERENCE

7, and 14, the livers of 7-10 fish (3-5 of each sex from exposure phenanthrene, and chrysene (Figure 2) with the parent compound and control tanks) were excised and assayed for microsomal pro- more depleted than the methyl (Cl), dimethyl and ethyl (C2), and tein, the P450 protein CYP1A, ethoxyresorufin-O-deethylase trimethyl, methyl-ethyl and propyl (C3) forms are consistent with (EROD), benzo(a)pyrene hydroxylase (BAPH), and 7-ethoxy- biodegradation (Douglas et al., 1996). Phenanthrene and diben- coumarin deethylase (ECOD) (Vignier et al., 1994). zothiophene are almost eliminated. The pattern and rate of C1-C3 Each of the five sediments and an "uncontaminated" control chrysene loss suggest that these compounds are almost unaffected were used in the 10-day regulatory Amphipod (Eohaustorius even after 27 years. The residual hydrocarbon concentration of estuarius) Survival Test (Environment Canada, 1992a). The white the composite Black Duck Cove sediment sample used in the sea urchin (Lytechinus pictus) was used to assess the toxicity of fish-exposure experiment was 850 ppm. The total extractable interstitial water (Environment Canada, 1992b): four replicates of organic materials (TEOM) in the 1997 sediments used in the a dilution series and control (laboratory seawater) were prepared microscale biotests ranged from 426 to 12,744 ppm. for each porewater sample. Eggs (100) were examined from each Biological effects of residual oil. Hepatic microsomal protein, replicate to determine the concentration of porewater causing x% CYP1A protein and enzymatic activities for the winter flounder inhibition of fertilization (ICJ. In the Grass Shrimp Embryo-Lar- exposed to contaminated sediment and clean sand (Figure 3) re- val Toxicity Test (Wilson, 1997), 6 and 10-day old embryos were vealed insignificant differences in microsomal protein levels be- exposed to clean seawater and 100% sediment elutriate for 4 days tween fish exposed for 0, 1, 7, or 14 days, and between control then transferred to clean seawater for further development. Tox- and exposed fish within the same period. Fish exposed to the icity end points were embryo mortality, hatchability, and larval contaminated sediment for 14 days had higher levels of CYP1A viability to 4-day post hatch. Controls were clean seawater and protein than fish at days 0, 1 and 7. Elevated CYP1A levels were sediment elutriate from a reference site. The Microtox® 100% and only statistically significant from controls on day 14 (p<0.029: solid-phase tests measured the change in bioluminescence of Vibrio non-parametric Student's t-test). These 14-day exposed fishha d fisheri in serial dilutions of porewater and sediment (Microbics Cor- higher (p<0.05; one-way ANOVA) EROD and BAPH activities poration, 1992; Mearns etaL, 1995). than those measured on days 1 and 7; however, differences were undetectable between fish from the control and exposure groups on the same day (1, 7 or 14). Higher (p<0.05) CN-ECOD activi- Field observations and results ties in the exposure group compared to the day 14 control group At the lagoon in Black Duck Cove, stratified layers of buried and the day 0 group did not appear until the fourteenth day due to Bunker C residues were located at a 10-16 cm depth within the data variability. tidal flat area located just above the low water mark. On-site The average survival of control animals in the 10-day E. es- analysis suggested that most of the residual hydrocarbons were tuarius survival test was 99% (Table 1). Sites 1, 2 and 4 (12744, located within an oxygen limited/anaerobic zone (mean values: 4418 and 3008 ppm TEOM) have survival rates less than 70% of dissolved oxygen <0.2 ppm; sulfide 290 ìÌ; redox potential -298 the control, and are considered toxic under Canadian regulatory mV. Evidence of residual oil mobility included the observation of guidelines (Environment Canada, 1996). Calculated IC50 and IC25 oil sheens on the surface water in the vicinity of worm and clam values for the Echinoid Fertilization Test were greater than 100% holes. Dark brown/black oily droplets and iridescent sheens were for all the interstitial water samples. According to this test, pore- also observed when interstitial water filled holes dug by shovel. water from all of the sites is nontoxic. Survival and hatchability The range of interstitial water nutrient concentrations at Sites 1-5 of Grass Shrimp embryos were between 80 and 100% for all elu- were determined as nitrate and nitrite: 0.09-0.53 ìÌ; phosphate: triate samples and the seawater control. Larval viability was 92- 0.82^.36 uJvl; and ammonia: 72.3-176.1 ìÌ. At the time of 98%, except for tests with 6-day old embryos, in which Sites 1 sampling (September 30, 1997), the mean surface water tem- and 2 exhibited decreases to 73% and 71% values. Hence, water- perature and porewater salinity were 16 °C and 29.0 ppt respec- soluble components in the residual oil were of very low toxicity tively. to shrimp embryos and larvae. Light emission by V. fisheri was Oil chemistry. A known highly recalcitrant molecule in the unaffected by porewater or sediments from Sites 1-5, and therefore toxicity was not detected. oil, C30-17a(H),21ß(H) hopane, was used to quantify the extent of biodegradation. Patterns of depletion for dibenzothiophene,

100 c ▼ liSBiS; o iif!!:$i: pilfcl CD 75 III III! lllll t 111 Q. Ill Uli III 50 mm Uli IIII ■»—» 111! III 111 : iiil i llll Asse |S;ge| Ulla c IIII mm® lilill 25 I III! |lill| llllll ! Ü I * 1 l""l·' 1 CD 111! Hill pill §l!ll| CD mm* MI IfiiiiiSAi *â:£| CL 0 ill lllll lllllll lllll i Ð H DBT C1D C2D C3D Phen C1P C2PC3P Chrys C1C C2C C3C

Figure 2. Biodegradation of residual ARROW cargo oil from Black Duck Cove. DBT, Phen, and Chrys refer to dibenzothiophene, phenanthrene, and chrysene respectively. Error bars = one standard deviation.

003 BIOREMEDIATION 1077

5.0 may be biologically available (Vandermeulen and Gordon, 1976), Control BI diffusing within the sediments and to the surface via interstitial 4.5 -| Exposed| water (Thomas, 1973) to cause chronic toxicity. 4.0 A Strong evidence of significant biodegradation of Arrow oil lies o 3.5 -j in the observed changes in distribution of polycyclic aromatic cc CL 3.0 -I compounds, and Iatroscan analysis which showed that the linear alkanes, the most prominent peaks in total ion scan chroma- 2.5 ]ðÃÇÐ tograms of the original oil, are essentially absent from most of the 0.6 -i sediments used in the biotests (Lee et al., 1998). Wang et al. (1994) reported that after 22 years of weathering, most paraffins 0.4 A and PAHs in oiled sediments from Chedabucto Bay were lost, but high-ring alkyl PAH homologues such as chrysenes and bio- 0.2 A marker triterpanes and steranes remain. Biodegradation is further supported by evidence of site specific metabolic adaptations. For 0.0 PL[| example, 4-methyldibenzothiophene in the residual oil was de- 2.0 ai graded more extensively than 2/3-methyldibenzothiophene, 1.5 -j which was more degraded than the 1-methyldibenzothiophene (Prince et al., 1998). This pattern is different from that obtained 1.0 A in laboratory bioremediation efficacy tests where 2/3-methyldibenzothiophene in Alberta Sweet Blend crude oil degraded faster 0.5 A than other forms (Wang and Fingas, 1995). 0.0 ip-cp. Induction of MFO by polyaromatic hydrocarbons is generally 1.5 -i rapid, and the effects can persist for extended periods. For example, fish at fuel oil impacted sites have shown elevated MFO 1.0 H activity after 8-20 years, indicating that biological recovery was incomplete (Teal et al., 1992). Environmental impact studies < 0.5 H following the Exxon Valdez oil spill revealed a correlation be- m tween CYP1A induction in subtidal fish and the levels of fluores- 0.0 cent aromatic compounds in their bile (Collier et al., 1996). The 0.6 3 MFO response of winter flounder to sediments contaminated with 850 ppm residual Arrow Bunker C was weak. Although some individuals seemed to show induction responses throughout the o exposure, the majority of exposed fish did not have CYP1A or o MFO activities above that of the control groups. ^ 0.2 -\ o Among the regulatory biotests, significant toxicity was only 0.0 Æâ-Ðâ- observed in the Amphipod Survival Test with Black Duck Cove 7 14 samples from Sites 1, 2, and 4. The Grass Shrimp Embryo-Larval Exposure Time (days) Toxicity Test is sensitive to hydrocarbons (Mearns et al., 1995), Figure 3. Microsomal (mg/g) and CYP1A (nmol/mg) pro- yet it detected little toxicity in the sediment elutriates. Interstitial teins and MFO (nmol/min/mg) enzymes. * Denotes signifi- water was determined non-toxic by the Echinoid Fertilization cant difference between exposure group and control. Test. The Microtox® Solid-Phase Test, which has been used successfully to assess the efficacy of oil spill remediation strategies (Lee et. al., 1995), did not detect toxicity in the sediments or Table 1. Ten-day test with Eohaustorius estuarius. porewater. Presumably oil residues from the Arrow, buried in beach sediments, are capable of retaining even highly mobile, low-molecu- % Survival lar weight components for up to 20 years (Vandermeulen and Sample Mean Standard deviation Singh, 1994), and microbial degradation of oil typically occurs at Control 99 2.2 low rates under oxygen limitation (Anderson and Malins, 1978). Considering the high residual hydrocarbon concentrations in the Sitel 14* 9.6 sediments used in the biotests, changes in the physical state and Site 2 41* 17.8 extensive biodegradation has undoubtedly reduced the toxicity of Site 3 72 9.7 the residual Arrow Bunker C oil. While the bioavailability of the Site 4 20* 6.1 oil may be limited by the concentration of its high-molecular weight components, biological impacts were correlated with its Site 5 69** 13.4 initial release into the environment. Observed effects within the * Toxic first 6 years included extensive mortalities of Fucus spiralis on **Borderline toxic =) rocky shores and of My a arenaria and Spartina alterniflora in the lagoons (Thomas, 1978). The 1997 toxicity tests provide conclusive evidence of site re- Discussion covery, which was validated by the observation of diverse benthic invertebrate communities within the sampling area, including the Oiled sediments for the 1997 toxicity studies were recovered green crab {Carcinus maenus), soft-shelled clam (Mya arenaria), from a depth of 10-16 cm within an oxygen-limited zone. Dis- common periwinkle (Littorina littored), amphipods (e.g., Gam- turbed sediment released residual oil into the interstitial water as marus oceanicus), and worms (e.g., Nereis sp., Arenicola sp.). oily brown/black globules. It is possible that these hydrocarbons Mya, which has been systematically sampled in Chedabucto Bay

004 1078 1999 INTERNATIONAL OIL SPILL CONFERENCE since the spill (MacDonald and Thomas, 1982; Thomas, 1978) 10. MacDonald, B.A., and M.L.H. Thomas, 1982. Growth has shown declining levels of tissue hydrocarbons. Trace amounts Reduction in the Soft-Shell Clam Mya arenaria from a were still detected in 1990 (Vandermeulen and Singh, 1994), but Heavily Oiled Lagoon in Chedabucto Bay, Nova Scotia. today residents are consuming these clams harvested from the Marine Environmental Research, v6, pp. 145-156. study area, which now includes a public park. 11. Mearns, A, K. Doe, W. Fisher, R. Hoff, K. Lee, R. Siron, C. Mueller, and AVenosa, 1995. Toxicity Trends During an Oil Spill Bioremediation Experiment on a Sandy Acknowledgements Shoreline in Delaware, USA. Proceedings of the 18 Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, Environment Canada, Ottawa, Ont., pp. 1133— The Panel of Energy Research and Development, Canada, and 1145. Exxon Research and Engineering Company, USA, funded this 12. Microbics Corporation, 1992. Microtox Manual Detailed research program. The authors acknowledge the technical assis- Protocols, v2, Carlsbad, CA., 178 p. tance of S.E. Cobanli, J. Gauthier, and S. St-Pierre. 13. Prince, R.C, R.M. Garrett, C.E. Haith, J.H. Vandermeu- len, S. Cobanli, D. Mossman, and K. Lee, 1998. The Role of Biodegradation in the Weathering of Oil from the 1970 Biography ARROW Spill. Proceedings of the 21st Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, Environ- Dr. Kenneth Lee is a Research Scientist with Fisheries and ment Canada, Ottawa, Ont., pp. 717-728. Oceans Canada whose expertise in microbial ecology is focused 14. Sergy, G., and E. Owens, 1993. What Remains of the on the development and implementation of research programs to ARROW Oil? Spill Technology News, ed. M. Fingas, En- assess potential environmental impacts of oil spills, development vironmental Protection Publications, Ottawa, Ont., vl8, of offshore oil industry, and ocean disposal. çÀ,ññ. 1-5. 15. Teal, J.M., J.W. Farrington, K.A Burns, J.J. Stegeman, B.W. Tripp, B. Woodin, and C. Phinney, 1992. The West References Falmouth Oil Spill After 20 Years: Fate of Fuel Oil Com- pounds and Effects on Animals. Marine Pollution Bulle- 1. Anderson, J.W., and D.C. Malins, 1978. Physiological tin, v24, nl2, pp. 607-614. Stresses and Response in Chronically Oiled Organisms. 16. Thomas, M.L.H, 1973. Effects of Bunker C Oil on Inter- Journal of the Fisheries Research Board of Canada, v35, tidal and Lagoonal Biota in Chedabucto Bay, Nova Sco- pp. 679-680. tia. Journal of the Fisheries Research Board of Canada, 2. Collier, T.K, C.A. Krone, M.M. Krahn, J.E. Stein, S-L v30, pp. 83-90. Chan, and U. Varanasi 1996. Petroleum Exposure and As- 17. Thomas, M.L.H, 1978. Comparison of Oiled and Unoiled sociated Biochemical Effects in Subtidal Fish after the Intertidal Communities in Chedabucto Bay, Nova Scotia. Exxon Valdez Oil Spill, Proceedings of the Exxon Valdez Journal of the Fisheries Research Board of Canada^ v35, Oil Spill Symposium, American Fisheries Society Sympo- pp. 707-716. sium nl8, American Fisheries Society, Bethesda, MD., 18. Vandermeulen, J.H., and J. Singh, 1994. Arrow Oil Spill pp. 671-683. 1970-90: Persistence of 20-yr Weathered Bunker C Fuel 3. Douglas, G.S., AE. Bence, R.C. Prince, SJ. McMillen, Oil. Canadian Journal of Fisheries and Aquatic Sciences, and E.L. Butler, 1996. Environmental Stability of Se- v51, pp. 845-855. lected Petroleum Hydrocarbon Source and Weathering 19. Vandermeulen, J.H., and D.C. Gordon, Jr., 1976. Re-En- Ratios. Environmental Science and Technology, v30, pp. try of 5-Year Stranded Bunker C Fuel Oil from a Low- 2332-2339. Energy Beach into the Water, Sediment and Biota of Journal of the Fisheries 4. Environment Canada, 1992a. Biological Test Method: Chedabucto Bay, Nova Scotia. Research Board of Canada, Acute Test for Sediment Toxicity Using Marine or Estua- v33, pp. 2002-2010. rine Amphipods. Report EPS 1IRMI26, Environment 20. Vignier, V., J.H. Vandermeulen, J. Singh, and D. Moss- Canada, Ottawa, Ont., 83 p. man, 1994. Interannual Mixed Function Oxidase (MFO) 5. Environment Canada, 1992b. Biological Test Method: Activity in Winter Flounder (Pleuronectes americanus) Fertilization Assay Using Echinoids (Sea Urchins and from a Coal Tar Contaminated Estuary. Canadian Journal Sand Dollars), Report EPS 1IRMI27. Environment Can- of Fisheries and Aquatic Sciences, v51, n6, pp. 1368- ada, Ottawa, Ont., 45 p. 1375. 6. Environment Canada, 1996. 1996 National Compendium 21. Wang, Z., M. Fingas, and G. Sergy, 1994. Study of 22- Monitoring at Ocean Disposal Sites. Environment Can- Year Old Arrow Oil Samples Using Biomarker Com- ada, Ottawa, Ont., 34 p. pounds by GC/MS. Environmental Science and Technol- 7. Goks0yr, A., and L. Förlin, 1992. The Cytochrome P-450 ogy, v28, pp. 1733-1746. System in Fish, Aquatic Toxicology and Environment 22. Wang Z., and M. Fingas, 1995. Use of Methyldibenzo- Monitoring, Aquatic Toxicology, v22, pp. 287-312. thiophenes as Markers for Differentiation and Source 8. Lee, K., R. Siron, and G.H. Tremblay, 1995. Effectiveness Identification of Crude and Weathered Oils. Environ- of Bioremediation in Reducing Toxicity in Oiled Inter- mental Science and Technology, v29, pp. 2842-2849. tidal Sediments. Microbial Processes for Bioremediation, 23. Wilson, J.E.H. 1997. The grass shrimp embryo-larval v3, n8, eds. R.E. Hinchee, CM. Vogel, and FJ. Brock- toxicity test : a short-term predictive bioassay. eds. J.S. Goudey, S.M. Swanson, M.D. Treissman and A.J. Niimi. man, Battelle Press, Columbus, OH., pp. 117-128. rd 9. Lee, K., G.H. Tremblay, G.D. Wohlgeschaffen, J.H. Proceedings of the 23 Annual Aquatic Toxicity Work- Vandermeulen, D.C. Mossman, K. Doe, R.M. Garrett, shop: October 7-9, 1996, Calgary, Alberta. Canadian Te- C.E. Haith, and R.C. Prince, 1998. Residual hydrocarbon chical Report of Fisheries and Aquatic Sciences, n2144, toxicity in sediments impacted by the 1970 ARROW spill. pp. 53-65. Proceedings of the 21st Arctic and Marine Oilspill Pro- gram (AMOP) Technical Seminar. June 10-12, 1998, Ed- monton, Alberta, pp. 485-504.

005

C S A S S C C S

Canadian Science Advisory Secretariat Secrétariat canadien de consultation scientifique

Research Document 2007/061 Document de recherche 2007/061

Not to be cited without Ne pas citer sans permission of the authors * autorisation des auteurs *

Recovery Potential Assessment Évaluation du potentiel pour le for the northern abalone (Haliotis rétablissement de l’ormeau nordique kamtschatkana) in Canada (Haliotis kamtschatkana) au Canada

J. Lessard, A. Campbell, Z. Zhang, L. MacDougall, and S. Hankewich

Fisheries and Oceans Canada, Stock Assessment Division, Science Branch, Pacific Biological Station, Nanaimo, B.C., V9T 6N7

* This series documents the scientific basis for the * La présente série documente les bases evaluation of fisheries resources in Canada. As scientifiques des évaluations des ressources such, it addresses the issues of the day in the time halieutiques du Canada. Elle traite des frames required and the documents it contains are problèmes courants selon les échéanciers dictés. not intended as definitive statements on the Les documents qu’elle contient ne doivent pas subjects addressed but rather as progress reports être considérés comme des énoncés définitifs on ongoing investigations. sur les sujets traités, mais plutôt comme des rapports d’étape sur les études en cours.

Research documents are produced in the official Les documents de recherche sont publiés dans language in which they are provided to the la langue officielle utilisée dans le manuscrit Secretariat. envoyé au Secrétariat.

This document is available on the Internet at: Ce document est disponible sur l’Internet à: http://www.dfo-mpo.gc.ca/csas/

ISSN 1499-3848 (Printed / Imprimé) © Her Majesty the Queen in Right of Canada, 2007 © Sa Majesté la Reine du Chef du Canada, 2007

006

007

Table of Content

Abstract...... iii Résumé...... iii Introduction...... 1 Summary biology and ecology of northern abalone...... 2 Phase 1 ...... 4 Step 1: What is the present/recent species trend? ...... 4 Step 2: What is the present/recent status?...... 6 Steps 3 & 4: What is the expected target and time frame for recovery?...... 6 1. Recovery targets...... 7 ¾ Short term...... 7 ¾ Long term...... 7 2. Mortality rates...... 7 3. Mean densities of mature (>70 mm SL) abalone...... 8 4. Patch size...... 8 Step 4.5: Important sources of mortality...... 8 1. Illegal harvest or poaching...... 9 2. Sea otter predation ...... 9 Phase 2 ...... 11 Step 5: What is the maximum human-induced mortality which the species can sustain and not jeopardize survival or recovery? ...... 11 Steps 6 & 7: The major potential sources of mortality/harm and the amount of mortality/harm caused by each...... 11 1. Directed fishing...... 11 2. Habitat alterations...... 12 ¾ Finfish aquaculture...... 13 ¾ Log booms and log dumps...... 14 ¾ Dredging ...... 14 3. Abalone aquaculture...... 15 4. Fisheries on food supplies (i.e. kelp harvest) ...... 15 5. Scientific research...... 16 6. Rebuilding efforts...... 17 ¾ Larvae and juvenile outplanting...... 18 ¾ Adult aggregation...... 18 7. Dismissed activities...... 19 ¾ Military activities...... 19 ¾ Bycatch ...... 19 ¾ Detrimental impacts on habitats by fishing activities...... 19 ¾ Ecotourism and recreation...... 19 ¾ Shipping, transport, and noise...... 19 Step 8: Aggregate of the total mortality/harm from human activities and contrast with model.19

008

Phase 3: Options ...... 19 Summary and conclusions...... 19 Recommendations...... 21 References...... 21 Appendix 1 Abalone poulation model...... 34 Appendix 2 Impact assessment protocol for works and developments potentially affecting abalone and their habitat ...... 61 Appendix 3 Survey report of experiment to determine short-term impacts of finfish aquaculture on abalone...... 79 Appendix 4 Request for Working Paper...... 99

ii 009

Abstract This recovery potential assessment for the SARA-listed northern abalone includes a review of current status, population projections, and recommendations on permitting human-induced mortality and/or harm to abalone and their habitat. Recent surveys indicated northern abalone abundance was not recovering. Time series analyses of abalone survey data from areas free of sea otters in southeast Queen Charlotte Islands and Central Coast during 1978-2002 provided stock-recruitment relationships, recruitment trends and mortality estimates of > 0.20. Simulations indicated that abalone populations could continue to decline if mortality rates remain >0.20. Mortality rates of < 0.20 are required for abalone populations to recover. Several human activities were considered that could potentially harm and cause direct mortality to abalone populations. In order of importance, these activities were: 1) directed fishing; 2) habitat alterations, including finfish aquaculture, log booms and log dumps, and dredging; 3) abalone aquaculture; 4) fisheries on food supplies (i.e. kelp harvest); 5) scientific research; and 6) rebuilding activities, including larvae or juveniles outplanting and adult aggregations. Collectively, harmful activities that can be permitted under SARA cause little mortality relative to poaching or sea otter predation. No allowable direct mortality is recommended.

Résumé Ce document présente l’évaluation du potentiel pour le rétablissement de l’ormeau nordique, une espèce en voie de disparition, incluant la situation courante de l’espèce, des projections de populations futures et des recommandations pour permettre les activités qui pourraient tuer ou nuire aux ormeaux ou leurs habitat. Des échantillonnages récents ont démontré que l’abondance des ormeaux ne se rétablit pas. Les analyses de deux séries chronologiques d’échantillonnages, entre 1978-2002, dans des aires où les loutres de mer étaient absentes ont fourni des courbes de stock-recrutement, des projections de recrutement et des estimés de taux de mortalité de >0.20. Les simulations ont indiqué que les populations d'ormeaux pourraient continuer à diminuer si les taux de mortalité demeurent >0.20. Des taux de mortalité <0.20 sont requis pour rétablir les populations d’ormeaux. Plusieurs activités humaines qui pourraient potentiellement nuire et causer la mortalité directe d'ormeaux ont été considérées. Par ordre d'importance, ces activités étaient : 1) pêche dirigée; 2) changements de l’habitat, y compris l'aquaculture de poisson, l’entreposage et les dépotoirs de billots de bois, et le dragage ; 3) aquaculture d'ormeau ; 4) pêche sur les sources alimentaires (c.- à-d. moisson de varech) ; 5) recherche scientifique ; et 6) les activités de rétablissement de l’ormeau, y compris l’ensemencement de larves ou de juvéniles et les concentrations d'adulte. Collectivement, les activités humaines qui pourraient potentiellement nuire et/ou tuer les ormeaux, qui peuvent être autorisées par la loi pour les espèces en péril, causent peu de mortalité comparativement à la mortalité causée par la prédation des loutres de mer et/ou du braconnage. Aucune mortalité directe n'est recommandée.

iii 010

011 Introduction The northern abalone (Haliotis kamtschatkana), a patchily distributed marine mollusc, has declined in numbers and distribution in surveyed areas of British Columbia (BC) as documented by regular surveys since the late 1970s. In response to observations of population declines, all abalone1 fisheries (commercial, recreational and aboriginal) were closed in BC at the end of 1990. Despite the harvest closure, abalone numbers remained low and by April 1999, H. kamtschatkana was assigned a threatened status in Canada by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) (Jamieson 2001). This status was reconfirmed by COSEWIC in May 2000 (Campbell 2000). In June 2003, northern abalone was legally listed and protected as threatened under the Species at Risk Act (SARA). A recovery team was formed in November 2001 and prepared a ‘National Recovery Strategy for Northern Abalone in Canada’ (Abalone Recovery Team 2002). Following the completion of the recovery strategy, an action plan was drafted and is awaiting approval by the Minister. Both documents are available on the internet: www-comm.pac.dfo-mpo.gc.ca/pages/consultations/fisheriesmgmt/abalone/default_e.htm Recovery strategies identify what needs to be done to stop or reverse the decline of a species. The abalone recovery strategy’s short-term goal is to halt the decline of the wild northern abalone population in BC in order to reduce the risk of the species becoming endangered (Abalone Recovery Team 2002). Over the long-term, the goal is to increase numbers and densities to self-sustaining levels in each biogeographic zone of BC in order to remove the species from the threatened status. The goal of increasing northern abalone to sustainable levels can be expected to take several decades. The measurable short-term objectives over the next 5 years are to ensure that mean densities of large adult (≥100 mm shell length (SL)) abalone do not decline below 0.1/m2 at surveyed index sites and that the percentage of surveyed index sites with large adult abalone does not decline below 40%. From the most recent surveys the mean densities of large abalone (≥100mm SL) were only 0.04/m2 in Queen Charlotte Islands (Haida Gwaii, QCI) in 2002 and 0.02/m2 on the north and central mainland coast (CC) in 2006 (Fig. 2), well below the short-term recovery objective of maintaining densities at or above 0.1/m2. The percentage of sites with large abalone was also below the short-term recovery target of 40% (Fig. 3). The short-term objectives of the recovery strategy have not been achieved and the abalone populations continue to decline or oscillate at low levels. Northern abalone was the first species for which Fisheries and Oceans Canada (DFO), Pacific Region, prepared a recovery strategy, prior to the implementation of the SARA. Although recovery of abalone in BC was deemed feasible, recovery targets (sustainable levels; see above) were not defined and no assessment of potential harm to abalone population or their habitat were included in the recovery strategy. The objectives of this paper were to (1) assess the recovery potential and population status of abalone in BC and (2) review and assess the human activities that may harm abalone in BC and could be permitted under SARA section 73 or allowed under section 83. The request for a PSARC Working Paper is in Appendix 4.

1 “Abalone” refers to northern abalone in this paper unless otherwise stated.

012

This document follows the ‘Allowable Harm Assessment’ (AHA) framework defined at a national workshop in 2004 (DFO 2004c). The AHA framework requires that the recovery feasibility be determined before any harms from human activities can be assessed, as such, the framework also includes a recovery potential assessment (Phase 1, see below). The framework to assess harms to a species at risk that could be permitted under SARA includes a number of steps divided into three phases. The first phase assesses whether recovery of abalone is feasible if human activities which affect the species were to continue. The second phase reviews and assesses the important human activities and conditions within which they must operate if recovery is to be feasible. Finally, the third phase develops the specific options and recommendations considering all the activities assessed, consistent with the provision of section 73 of SARA. Two sections, or steps, have been added into the framework: (1) a biology section reviewing the elements of abalone biology and ecology necessary for understanding the impacts of the different human-induced activities reviewed and assessed, and (2) a review and assessment of sources of mortality of abalone populations that cannot be permitted under section 73 of SARA, but that are so important that they cannot be ignored when considering recovery (Step 4.5). For the sake of brevity and ease of reading, the steps of Phase 3 have been removed as alternative methods (Step 9), mitigation measures (Step 10), science advice discussion (Step 11), and recommended options (Step 12) are included for each potential harm when they are assessed under Steps 6&7. Phase 3 summarizes the discussions of Phases 1 and 2 and lays out the recommendations in the usual PSARC format.

Summary biology and ecology of northern abalone The biology of northern abalone was reviewed by Sloan and Breen (1988). In summary, northern abalone range from northern Alaska to Baja California in patchy distributions on exposed and semi-exposed areas from low intertidal to subtidal depths; most abalone off the coast of BC are found at depths shallower than 10m. Juvenile northern abalone (10-70 mm SL) are found under and on exposed areas of rocks, whereas the majority of adults (>70 mm SL) are found on exposed rock surfaces. As the juveniles develop to maturity, their diet changes from benthic diatoms and micro-algae to drift macro-algae. Abalone become sexually mature at about 50mm SL and all abalone are mature at 70mm SL (Campbell et al. 1992). Sea urchins and adult abalone have been shown to use the same habitats and may compete for algal food. Abalone predators include sea otters, some sea stars, octopus, crabs and fish. Abalone growth can vary considerably between areas depending on the extent of exposure to wave action and availability and quality of food. In BC, estimates of the age at which abalone reach 50 mm SL are 2 to 5 years old and 100 mm SL are between 6 to 9 years (or more). Growth of adults tends to be stunted in highly exposed outer coastal areas due to reduced opportunities for abalone to catch, and feed, on drift algae in strong wave action and water currents. Abalone growth is more rapid in moderately exposed areas with giant kelp, Macrocystis integrifolia, or bull kelp, Nereocystis luetkeana, kelp forests than at highly exposed areas with Pterygophora californica kelp forests (Sloan and Breen 1988). Breen (1980) speculated that ages of 50 years or more were not improbable based on shell appearances and extrapolation from other invertebrate species. However, longevity of 15-20 years for northern abalone seem more appropriate based on age determined by spire growth rings (Shepherd et al. 2000) and observed sizes taken with growth curves (Quayle 1971; Sloan and Breen 1988).

2 013

Northern abalone spawn synchronously, with groups of males and females in close vicinity to each other in shallow waters broadcasting their gametes into the water column (Breen and Adkins 1980b). These aggregations are believed to enhance reproductive success in many abalone species by increasing the chance of fertilization (Sloan and Breen 1988; Shepherd and Brown 1993; McShane 1995a, 1995b; Shepherd and Partington 1995; Babcock and Keesing 1999; Dowling et al. 2004). Recent studies on several abalone species (McShane 1995a, 1995b; Shepherd and Partington 1995; Babcock and Keesing 1999; Dowling et al. 2004) and sea urchins (Levitan et al. 1992, Levitan and Sewell 1998) have pointed to reduced fertilization success caused by dilution of gametes through reduced adult spawner densities. The planktonic phase of the northern abalone is short and temperature dependent (10-14 days at 14-10ºC) (Sloan and Breen 1988; Pearce et al. 2003). Since larvae of abalone species in general are non-feeding and are poor swimmers, many authors have suggested that larval dispersal is minimal and that recruitment is from local reproductive populations (Shepherd and Brown 1993; Tegner 1993; Tegner and Butler 1985; Prince et al. 1987; McShane et al. 1988; McShane 1992, 1995a, 1995b). Nevertheless, larval dispersal probably still occurs as genetic studies have found high genetic variation indicating gene flow over large areas with little population subdivision in several abalone species, including northern abalone (Brown 1991; Brown and Murray 1992; Shepherd and Brown 1993; Burton and Tegner 2000; Withler et al. 2003). Stock definition of northern abalone has been considered, in the fisheries management context, as an abalone population within arbitrarily chosen geographic or management areas. Consequently most of the stock assessment surveys of abalone in BC have been on a broad geographic scale. Currently, 5 biogeographic zones are recognized to recover abalone (Abalone Recovery Team 2002): Queen Charlotte Islands, north and central mainland coast, Queen Charlotte and Johnstone Straits, Georgia Strait, and west coast of Vancouver Island. Evidence from recent studies have suggested that some abalone species may be made up of many populations in which stock recruitment relations may occur in small geographic areas (on a scale of hundreds of metres to several kilometres) based on gene exchange (Brown 1991; Brown and Murray 1992) and larval exchange (Tegner and Butler 1985; Prince et al. 1987; McShane et al. 1988). Shepherd and Brown (1993) suggested that an abalone stock be defined as a metapopulation made of several local discrete populations that have limited larval interchange. This stock definition allows for managing local abalone populations that may have variable demographic processes. Few studies have shown strong stock recruitment relationships and the requirement of maintaining high adult abalone densities to ensure sufficient recruitment. Shepherd and Partington (1995) showed that there was a critical stock density threshold (0.15/m2) for the H. laevigata in Waterloo Bay, South Australia, below which the risk of recruitment failure was high. Shepherd and Brown (1993) found that a “minimum viable population” of more than 800 individuals of H. laevigata was required; anything less at West Island caused recruitment failure. Shepherd and Baker (1998) suggested that recruitment to an abalone fishery could be relatively poorer and more variable in small than in large abalone populations, in which case small populations would need to conserve relatively more egg production to prevent depletion. These studies supported the influence of the Allee effect or depensation (Allee et al. 1949) in which low abalone densities and aggregations reduced reproductive success due to low fertilisation of gametes.

3 014

Campbell et al. (1992) reported fecundity of 156,985 eggs for a 57 mm SL abalone and of 11.56 million eggs in a 139 mm SL female. The largest female, 144 mm SL, carried 11.31 million eggs. From cumulative size frequency data of abalone surveyed off eastern Moresby Island during June 1990 (Thomas et al., 1990), Campbell et al. (1992) estimated that 50% of the total potential eggs could be produced by the mature females < 100 mm SL which constituted 80% of the total population surveyed. The remaining 20% of mature females in the 100-152 mm SL size group could produce about 50% of the total potential egg production. They concluded that large females are potentially important in contributing eggs to the total potential egg production.

Phase 1

Step 1: What is the present/recent species trend? Since 1978, fishery independent surveys provided a time series through sampling abalone densities and size frequencies from QCI and the Central Coast of BC (CC), every 3–5 years (Fig. 1) (Adkins and Stefanson 1979; Atkins et al. 2004; Boutillier et al. 1984, 1985; Breen and Adkins 1979, 1980a, 1981; Breen et al. 1978b, 1982; Campbell et al. 1998, 2000a; Carolsfeld et al. 1988; DFO 2004a; Farlinger and Bates 1986; Farlinger et al. 1991; Lessard et al. 2007; Thomas and Campbell 1996; Thomas et al. 1990; Winther et al. 1995; and unpublished data for 2006 survey). In the early years (1978-1983), the surveyed sites were chosen because of harvestable commercial abalone abundances. The general survey method, by consistently using the standard 16 one m2 quadrat survey method developed by Breen and Adkins (1979) at indicator sites (commonly known as the ‘Breen’ method), provided a time series of abalone abundance indices in the QCI and CC. Although there were a few published surveys of southern BC (Quayle 1971; Breen et al. 1978a; Adkins 1996; Wallace 1999; Atkins and Lessard 2004; Davies et al. 2006) they did not provide the extended coverage and the time series of the surveys in the northern half of BC. Most surveys were conducted in northern BC where historically the bulk of BC commercial abalone harvest occurred and abalone were considered most abundant (Sloan and Breen 1988). Consequently, the results from surveys at index sites in northern BC have been used by DFO, and others, notably COSEWIC, to make management decisions. (N.B.: The 2006 CC survey densities are presented, but should be considered preliminary as the survey report is in preparation and analyses are pending; J. Lessard, DFO, Nanaimo; pers. comm.). The mean total abalone density2 at comparable index sites declined from 2.40 to 0.40 abalone/m2 for CC, during 1978-2006, and from 2.22 to 0.34 abalone/m2 for QCI during 1978-2002 (Fig. 1). During the same periods, the mean large (≥100 mm SL) density decreased from 1.10 to 0.02 abalone/m2 for CC and from 0.36 to 0.04 for QCI (Fig. 2). While there were significant declines of total densities with the previous surveys in both QCI 2002 and CC 2001, proportionally, the densities for large and mature abalone decreased more rapidly than that for small individuals (Atkins et al. 2004; Lessard et al. 2007). The mean size of abalone surveyed significantly dropped from 76.4 mm SL in 1998 to 67.0 mm SL in 2002 in QCI and from 80.7 mm SL in 1997 to 77.6 mm SL in 2001 in CC. The larger decreases in large and mature abalone densities as well as the decline in mean SL suggest size-selective fishing (poaching) mortality. Sea otters

2 unless otherwise stated, all densities are of emergent/exposed abalone

4 015

were only present in a small portion of the CC surveyed areas (Nichol et al. 2005) and therefore, sea otter predation could not explain these reductions in density and mean size estimates. Other surveys using different sampling designs also confirmed the low densities of abalone found by the index surveys in the same areas (Lucas et al. 1999; Cripps and Campbell 1998; Campbell and Cripps 1998; Lessard et al. 2002; Jones et al. 2003). The similarity in abalone density between new random sites and index sites indicated that the mean densities from all index sites were reasonably representative of adult abalone sampled in areas of CC in 1997 and QCI in 1998 (Lucas et al. 1999; Cripps and Campbell 1998; Campbell and Cripps 1998). New index site surveys were initiated on the west coast of Vancouver Island in 2003 (WCVI) (Atkins et al. 2004) and in Queen Charlotte and Johnstone Straits in 2004 (Davies et al. 2006). The mean total density estimates were 0.06 abalone/m2 in Queen Charlotte Strait and 0.02 abalone/m2 in Johnstone Strait. Previous surveys done in a particular location in Johnstone Strait indicated high abalone density, up to 10 abalone/m2 based on timed-swims in 1977 (Breen et al. 1978a), and 1.13 abalone/m2 from a 1986 using the Breen survey method (Adkins, 1996). This particular location was the only place in Johnstone Strait where divers surveyed any abalone at all in 2004, save one individual at another site. Harvest logs confirm there were commercially harvestable numbers in PFMA 12 (Harbo 1997), but a survey by Breen et al (1978a) suggested that the fishery was largely confined to the north of Port Hardy, where densities were higher. In Queen Charlotte Strait in 1977, Breen et al. (1978a) visually estimated abalone densities to be generally low, 1/m2 or less and noted a scarcity of juveniles. Although not directly comparable due to differences in survey designs, abalone densities in Queen Charlotte Strait have certainly decreased since 1977. Abalone densities in the Queen Charlotte and Johnstone Straits were at levels where the likelihood of recruitment failure is high (Shepherd and Brown 1993, Shepherd and Partington 1995; Babcock and Keesing, 1999; Campbell 2000). On north west side of WCVI (north of Brooks Peninsula), the mean total density was 0.09 abalone/m2 from all sites sampled, but 0.21 abalone/m2 from sites in Quatsino Sound where more sheltered abalone habitat was present (Atkins and Lessard 2004). Sea otters, Enhydra lutris, have inhabited the surveyed area of WCVI since 1989 and more specifically since 1991 in Quatsino Sound (Watson et al. 1997). Even sampling only exposed abalone during 2003 were abalone present in areas where otters were established at higher abalone densities than that estimated by Watson (1993) in areas with sea otters. This may be the result of the relatively low densities which make abalone a scarce food resource for sea otters and are therefore not selected as sea otters often exploit seasonally abundant food resources (Watson et al. 1997). No other abalone surveys exist for the area surveyed in 2003; therefore trends for this area cannot be assessed. A new index sites survey was planned for Georgia Strait in 2005, but because of budget constraints and other emerging priorities, only a small portion on the southern tip of Vancouver Island was surveyed in February 2005 (J. Lessard, unpublished data). Only 3 individual abalone were found at two (11%) of the 19 sites surveyed. The mean density for all sites surveyed was 0.0098 abalone/m2. This estimate was drastically lower than density estimates from two previous surveys on the south coast; one in 1982 in PFMA 19, 0.73 abalone/m2, and one in 1985 in PFMA 20, 1.15 abalone/m2 (Adkins 1996). Sampling methodology in the 1982/85 surveys used preliminary visual estimates of abalone density, and the Breen method was only applied to sites where density was estimated to be >0.5 abalone/m2. This would have resulted in higher mean density estimates for both Adkins (1996) surveys than if a random sampling method had

5 016

been employed. Nevertheless, even assuming overall densities of half of the estimates of 1982/85 (0.47 abalone/m2), the population of abalone in these areas has declined by more than 97% since the mid-80’s. Due to access restrictions enforced by a prison, the waters around William’s Head have been a marine reserve since 1958 (Wallace 1999). During a 1996/97 survey divers found 211 abalone in 275 minutes of diving (0.77 abalone/min) at this prison site. At 4 sites surveyed around Williams Head in 2005 (J. Lessard unpublished data), only one abalone was measured. Although a few more were observed around the quadrats, the abalone population around Williams Head was disappearing. This large decrease could be a result of poaching, but at this particular location because of the security, the most likely reason is simply that the large abalone found at the previous surveys have died and there has been no, or little, recruitment. The sizes of the three sampled abalone during the 2005 survey were 141mm, 135mm, and 135mm SL, well above the historical size limit of 100mm SL in place during the former fishery. The results from the 2005 survey were similar to surveys conducted in the San Juan Islands by Washington Department of Fish & Wildlife (D. Rothaus, WDFW; pers. comm.). Despite the fisheries closure, clearly the abalone population on the south coast has continued to decline and may have reached critical levels.

Step 2: What is the present/recent status? From the most recent surveys, the mean densities of large adult abalone (≥100mm in SL) were only 0.04/m2 in Queen Charlotte Islands (Haida Gwaii, QCI) in 2002 and 0.02/m2 on the north and central mainland coast (CC) in 2006 (Fig. 2), well below the short-term recovery objective of maintaining densities at or above 0.1/m2 (see Introduction). The percentage of sites with large abalone was also below the short-term recovery target of 40% (Fig. 3). Other surveys in other parts of BC show even lower densities (see above). The short-term objectives of the recovery strategy (Abalone Recovery Team 2002) have not been achieved and abalone populations continue to decline or oscillate at low levels. Indeed the prospect of further precipitous abalone population declines must be considered a possible reality with the continued illegal harvest (poaching) by humans and the continued sea otter population growth and spread in BC (see below).

Steps 3 & 4: What is the expected target and time frame for recovery? The short-term goals of the recovery strategy dealt with large abalone as these individuals have the most reproduction potential and if any future fisheries were to be considered, the portion of the abalone populations that would be exploited would probably be over 100 mm SL. At the time of writing, short-term objectives were set to observe if abalone population continued to decline, but no recovery targets were set. The short-term recovery objective of 0.1 large abalone/m2 was chosen as a reference point to measure declines or growth of the abalone populations comparatively from the time the fisheries were closed. Furthermore, the long-term recovery targets will have to be based on smaller size classes in areas where sea otters are present and may be partly based on these larger abalone where sea otter population expansion is not expected in the next 10-20 years. Restoration of abalone populations to the levels seen in the late 1970s will not be possible in areas affected by the continuing expansion of sea otter populations. Figure 5 shows the density estimates of large (≥100mm SL) and mature (≥70mm SL) abalone from all index sites surveys in QCI and CC. In general, the trends in both mature

6 017

and large abalone densities were similar. Around the time of the closure, the mature density estimates were 0.27 and 0.41 mature abalone/m2 in QCI and CC, respectively. It is instructive to use simple generic models to inform recovery goals for endangered species. Such a model was developed in order to determine recovery targets and time frame of abalone recovery. The simulation model and its various components are described in Appendix 1. The model is derived from survey time series data of >20 years in 2 geographic areas briefly described in Step 1 and more explicitly in Appendix 1. The model is based on the best available data in BC and comparison of simulated and observed densities of exposed abalone showed that the majority of observed densities fell within 90% confidence intervals of the simulated values (see Fig. 4-7 in Appendix 1). For these reasons, we consider the model reasonable and use it to derive recovery targets. However, as the surveys were completed in areas without sea otters, the recovery targets can not be set for areas currently or future occupied by sea otters.

1. Recovery targets We propose recovery targets with a two phase approach using short term and long term time horizons. Measurable targets, using standard surveys of abalone, should include: (1) annual mortality estimates; (2) density estimates of emergent or “exposed” reproductive broodstock (>70 mm SL); and (3) frequency of patch sizes of abalone. The chosen recovery targets are listed for both the short- and long-term. Each measurable target is then discussed. As discussed earlier and in Appendix 1, abundance of immature abalone (<70 mm SL) is difficult to measure and study. If better methods are developed to estimate immature densities, recovery targets should include a measure of juvenile abalone abundance.

¾ Short term The measurable short-term objectives over the next 10 years are to (1) reduce annual estimated mortality rates to <0.20, (2) ensure that mean densities of mature (>70 mm SL) abalone increase to >0.32/m2 at surveyed index sites (twice the current densities in 2001 CC and 2002 QCI) and (3) increase the percent of surveyed quadrats with abalone to >40% (index sites surveys).

¾ Long term The measurable long-term objective for the next 30 years are to (1) reduce and maintain annual estimated mortality rate to <0.15, (2) ensure that mean densities of mature (>70 mm SL) abalone increase to >0.5/m2 at surveyed index sites and (3) increase the percent of surveyed quadrats with >1 mature abalone to >20% (index sites surveys).

2. Mortality rates For adult abalone (different size fraction depending on method, see Appendix 1), mean mortality rates (Z) estimated in areas without sea otters were 0.23±0.04SE and 0.29±0.05SE for QCI and 0.21±0.06SE and 0.36±0.07SE for CC (Table 2 in Appendix 1). The higher Z estimates were probably more accurate for both QCI and CC as more of the survey-derived density estimates fell within the confidence limits of the model simulations (Fig. 4-7 in Appendix 1). Simulation model predictions indicate that should annual mortality rates remain higher than 0.20, abalone populations will continue to decline (Fig. 8, 9 and 11 in Appendix 1). Indeed, in areas with sea otter presence, mean annual mortality values would be expected to be considerably

7 018

higher than 0.20 (e.g., >0.35) and consequently abalone densities would decline at a more rapid rate with a possible earlier projection of critical densities (here set at 0.001) which would be considered un-recoverable. For example, with a Z=1.0, abalone can be expected to be below 0.001 abalone/m2 in 25 years, and with Z=0.26 abalone would reach critical densities in >300 years (Fig. 14 in Appendix 1). Decreasing mortality rates is essential to the recovery of abalone in BC and should be a short-term priority.

3. Mean densities of mature (>70 mm SL) abalone The slope of the stock-recruitment relationship decreases around 0.2 kg/m2 spawning biomass which correspond to about 1 spawner/m2 (1 mature abalone ≥70mm SL/m2) (Fig. 2-3 in Appendix 1). As the slope decreases, an increase in spawning biomass does not correspond to a comparable increase in the number of recruits and therefore the benefit of further increasing the number of spawners decreases. The majority of the survey data are below this level (Fig. 5). The recovery target for delisting could be set at a mean of 1 spawner (mature)/m2 at index sites surveys in the CC and QCI as this is the best available information we have. At a local scale (e.g., site level), a less conservative target could be set at half of this level to allow for potential conditional small harvest (see activity 1 under Steps 6-7). The time frame to recover abalone at the 1 mature abalone/m2 level is not realizable within 50 years with the current mortality rates (Fig. 8-13 in Appendix 1). The mortality rate would have to decrease to at least 0.15 to recover abalone to the 1 mature/m2 level within 70 years (Fig. 12-13 in Appendix 1 and assuming mature abalone densities of 0.15 in QCI in 2002 and 0.17 in CC in 2001). However, doubling the current mature density estimates in 10 years could be possible if mortality rates were lowered (i.e., without illegal harvest and/or sea otter predation) (Fig. 12-13 in Appendix 1).

4. Patch size The size and distribution of abalone populations required for effective reproduction and subsequent sufficient recruitment are unknown. Current knowledge of abalone, in general, suggests that there needs to be sufficient densities within patches of large mature abalone close enough together to successfully spawn and produce viable offspring (see biology section). Figure 4 shows the proportion of quadrats surveyed with different counts of abalone. The proportion of quadrats with no abalone (zero counts) was around 80% in the latest survey in QCI and CC. Although patch size is not directly measured in these surveys, the frequency of quadrats with small counts is an indication that patch size and frequency were decreasing. As the number of abalone close together is important for fertilization success (see biology section), the relationship between patch size and the proportion of quadrat with different abalone counts needs to be established. If no relationship can be determined, patch size should be measured directly in future surveys. In the mean time, abalone counts per quadrat (1 m2) could be used as a surrogate measure of patch size and frequency. An increase of 40% of total quadrats surveyed with abalone would effectively double the CC 2001 and QCI 2002 observations. Ideally, however, aggregations of reproducing abalone are needed to recover the populations and therefore an increase in the patch size and frequency of mature abalone should be the long-term goal.

Step 4.5: Important sources of mortality This step is an addition to the framework proposed in DFO (2004c). The causes of mortality for which we cannot issue a permit under Section 73 of SARA are so important that they cannot be

8 019

ignored when considering abalone recovery. In addition, without proper perspective in the different causes of mortality to abalone populations, the harm caused by the activities described in Steps 6-7 can seem considerable when in fact the aggregated mortality of these activities (those that can be permitted under Section 73 of SARA) is small compared to poaching and sea otter predation. Watson (2000) suggested that the effects of sea otters and human harvesters have on abalone populations differ in several aspects. Sea otters have no regard for size limits and other management measures, but as abalone become rare, sea otters are energetically constrained and switch to alternate prey (Wild and Ames 1974; Ostfeld 1982). In contrast, as the abundance of abalone declines, abalone becomes more valuable, making it worthwhile to pursue a rare species for human harvesters. Furthermore human harvesters using specially designed tools have the ability to remove abalone from areas where sea otters cannot. The combination of sea otter predation and human harvest may prove too much for many abalone populations (Watson 2000). These two sources of mortality to abalone populations are discussed below.

1. Illegal harvest or poaching Illegal harvest or poaching of abalone is considered an important source of abalone mortality (Jubinville 2000). Campbell (1997) estimated F (instantaneous fishing mortality) to be at least 0.20 for south east QCI and from 0.14 to 0.70 in some areas in CC during the post-fishery closure (1993-96) period. The mortality rates (Z) calculated in Appendix 1, 0.21-0.36, are higher than natural mortality rates (M), 0.15-0.20, estimated by Breen (1986) for adult abalone population in areas closed to the commercial fishery. However, Z estimates are within the range, 0.21-0.41, Breen (1986) estimated for areas exposed to the commercial fishery. This indicates that illegal harvest was probably still ongoing and was a major source of mortality. Using the M estimates from Breen (1986), fishing mortality (F) after the fisheries closure can be estimated to be 0.06-0.26. Recently, 1997 – 2006, approximately 30 abalone poaching convictions have been made. In some cases, multiple charges were laid and strict sentences meted out to repeat offenders. However, fishery officers suggest that poaching remains a major concern and estimate that only 10-20% of poaching activity is prosecuted (B. Hume, DFO Conservation and Protection, Campbell River; pers. comm.). There were also 14 reports of suspected poaching in 2004 and 23 in 2005. The majority of reports received were from northern BC (B. Hume; pers. comm.).

2. Sea otter predation Sea otters are considered a threat in the abalone recovery strategy (Abalone Recovery Team 2002). The sea otter is a natural major carnivore of many invertebrates, including abalone, and as a consequence can have a significant effect on the nearshore coastal ecosystems of BC (Watson 2000). However, sea otter abundance and distribution has been drastically manipulated and controlled by humans for over a century. Midden remains indicated that prior to the arrival of Europeans, First Nations may have extirpated local populations of sea otters (Simenstad et al. 1978). A massive fur trade occurred from the mid-1700s until 1911 when sea otters were protected under the International Fur Seal Treaty. By that time few populations remained and the last sea otter in BC was shot in 1929 resulting in the extirpation of sea otters in BC. Subsequently the sea otter was reintroduced into BC in three separate translocations from Alaska

9 020

between 1969 and 1972 and, with absence of exploitation, the populations have grown and continue to spread in BC (Watson et al. 1997; Nichol et al. 2005). Studies have shown that abalone, in areas where sea otters are present, are restricted to crevices and other cryptic habitats where they are inaccessible or hidden from sea otters (Lowry and Pearse 1973; Cooper et al. 1977; Pollard 1992; Watson 1993, 2000). It is unclear whether sea urchins and abalone inhabit crevices as a direct result of sea otter predation or because the abundant supply of food (perhaps an indirect result of sea otter predation) reduced their foraging activity (Lowry and Pearse 1973). Abalone populations have been present at low stable abundances in sea otter areas in California (Cooper et al. 1977, Hines and Pearse 1982; Wendell 1994). In BC, abalone do co-exist in area with sea otters, but it is not known at this time if this will persist when sea otters’ preferred food item, sea urchins, is depleted. Sea otters and abalone have co-existed in BC for millennia prior to the extermination of sea otters by humans in BC (Watson 2000). However, the exact mechanisms for this coexistence and the survival of abalone at low abundance in BC are not known. Ecosystems are clearly complex and are always changing in spatial and temporal scales and can develop counter intuitive population changes (Sinclair and Byrom 2006). Carter et al (2007), using sea urchins, have shown that the “sea otter-trophic cascade paradigm” is not applicable in all locations and habitat types. Many factors, such as environmental variability (e.g., storm frequency, climate change) and biological factors (e.g., disease, invertebrate predators and competitors, density dependent effects on growth, reproduction, and survival) may positively or negatively influence abalone density and abundance in an area. Although there are indications of the influence of local sea otter populations on nearshore ecosystems on a small scale (e.g. < 50 km shoreline) (Lowry and Pearse 1973, Cooper et al. 1977, Hines and Pearse 1982, Watson 1993; Wendell 1994), there is no clear understanding of the distribution profiles and influence of sea otter populations will have on the stock recruitment relationships of abalone and other predator species on a small (<50 km shoreline) or large scale (e.g., >50 km shoreline). In areas where abalone have been severely depleted by natural factors and or human poaching, subsequent sea otter predation may significantly accelerate the decline and contribute to the demise of abalone populations in many areas of BC. Mortality of red abalone, H. rufescens, populations in California in areas occupied by sea otters has been estimated at 0.3-1.0 (Hines and Pearse 1982 cited in Shepherd and Breen 1992) and 1.3 (Deacon 1989 cited in Shepherd and Breen 1992). In BC, there are no estimates of predation mortality by sea otters. However, rough estimates of mortality were calculated from Watson (1993, 2000) study on the effect of sea otters (Enhydra lutris) on nearshore ecosystems. At a permanent site in Kyuquot Bay, the mean number of abalone was 0.97/m2 in 1988, decreased to 0.36/m2 the following year and was 0.22/m2 in 1990 (Watson 1993). At that site, sea otters were first observed in November 1988 and foraged sporadically thereafter (Watson 1993). The abalone mortality rates were estimated to be 0.99 and 0.49. No legal fishery took place once sea otters had occupied the area (Harbo 1997). At the two control sites (without sea otter foraging), abalone densities fluctuated between 0.01-0.09/m2 and 017.-0.27/m2. These mortality rate estimates should be considered carefully as they do not include the number of sea otters in the area recently occupied nor do they distinguished between natural mortality and sea otter predation. Nevertheless, these estimates are similar to those of California. The Sea Otter Recovery Team (2003) recognized that conflicts between shellfisheries as well as abalone recovery will increase as the sea otter population(s) continues to spread to new areas.

10 021

Throughout the sea otter range in the Pacific, there is mounting evidence that many shellfish fisheries can not co-exist in the presence of an established sea otter population (Sea Otter Recovery Team 2003). During the consultations on the sea otter recovery strategy, the shellfish industry and some First Nations expressed the view that sea otter numbers have rebounded sufficiently in some areas, and that sea otters should be managed to control their numbers in those areas. Sea otters have been referred to as ‘keystone’ predators (Soulé et al. 2003), and contribute to the structure of nearshore ecosystems, with both direct and indirect effects on other species at risk and their associated habitats. Most of what we know about abalone in BC is based on studying systems without sea otters. There is no evidence that sea otter will enhance abalone spawning success by concentrating abalone in refugial habitats (crevices, under rocks, etc.). Ecological studies of northern abalone need to be conducted in experimental areas with sea otter populations present and absent to determine if future management of sea otters is required in some areas given that sea otter populations continue to grow and spread which threaten to accelerate the decline of abalone populations in BC.

Phase 2

Step 5: What is the maximum human-induced mortality which the species can sustain and not jeopardize survival or recovery? Based on the best information available, including index site surveys time series and the derived model, mortality rate estimates were >0.20 in sea otter free areas of northern BC after the closure to all fisheries in 1990, which likely contributed to continuing abalone population low abundances. Model simulations predicted that with mortality rates >0.20, abalone populations would decline further. As abalone are not recovering, the maximum human-induced mortality should be near zero as an increase in mortality rate will further reduce the abalone populations in BC.

Steps 6 & 7: The major potential sources of mortality/harm and the amount of mortality/harm caused by each At present, assigning estimated specific annual mortality rates due to each potential source or causative agent is difficult. However, we provide a qualitative relative estimate of direct and indirect mortality (with an estimated range of mortality rate (F) values) assigned to each potential source in Table 1 as follows: High = F >0.10; Medium = 0.10>F>0.03; Low = F <0.03; Minimal = F <0.01. Mortality causative agents can vary between small areas and large areas and between years. Clearly the cumulative effect of all the natural and human induced mortalities can be overwhelming to abalone populations. The potential mortality agents are listed below in order from the largest to the lowest estimated impact on abalone populations in BC. Unless stated, ‘mortality’ refers to mortality caused directly by the activity and harm refers to impacts of the activity that may lead to increase mortality (indirect mortality).

1. Directed fishing Directed commercial and recreational fisheries will not be considered here as they are unlikely to be opened in the near or distant future. In addition, criteria to re-open the commercial fishery were already set in Campbell (1997). While abalone fisheries were opened, there was little

11 022

known about the extent of First Nations and recreational harvesting, although the level was probably well below the commercial harvest (Sloan and Breen 1988). Conservation takes precedence over First Nation harvest for food, social and ceremonial purposes. Given current mortality rates calculated from index sites time series, directed fisheries are not possible. On the other hand, education and community projects have had little effect to decrease illegal harvest because the benefits of conserving and enhancing the abalone resource are not seen and are even thought to be not realizable. The recovery is simply taking too long for most people to see a benefit in becoming involved. Prince et al. (1998) advocated Territorial User Rights Fisheries (TURF) for protecting abalone populations. A form of TURF management was recently implemented to reduce poaching in the heavily exploited gastropod Concholepas fishery in Chile (Castilla et al. 1998). Several First Nation communities have been involved in abalone rebuilding projects in BC for several years. One possible action to increase stewardship would be to consider linking success in enhancement/rebuilding projects (e.g., juvenile outplanting) at specific sites by local First Nations with a small harvest of abalone for food, social, and ceremonial purposes under strict controls (e.g., within 1-2 days, with DFO enforcement personnel present on the grounds, etc.). The harvest would only occur if sufficient abalone densities are present at the site where rebuilding activities have been taking place The site would need to be monitored and have yearly surveys to determine success (e.g., above a certain abalone density threshold). Along with education, these small projects may increase the abalone population locally and give an incentive to protect this resource in their surrounding areas. Similar to requirements for the abalone aquaculture industry, a harvest could be considered with a requirement to contribute to abalone rehabilitation. As different methods of rebuilding may be used depending on the area (e.g., adult abalone aggregation in an area with sea otters may be counter productive to rehabilitation), decision rules will have to be project specific, but overarching decision rules on required abalone densities, minimum recovery work necessary, and monitoring requirements will have to be decided in consultation with First Nations and all stakeholders. The recovery targets given in Steps 3-4 can help set required densities at which a given site would be harvested after several years of rebuilding work. At the present time, the mortality from directed (legal) fishery is zero. Future harvest, if approved after consultations and under set protocol, cannot be estimated at this time.

2. Habitat alterations A protocol was developed in April 2004 to address the siting of proposed finfish aquaculture tenures and their impacts on abalone populations and their habitat. Portions of this protocol have been integrated into ‘DFO Marine Fish Habitat Information Requirements (HIR) for Finfish Aquaculture Projects’ (DFO 2005). The protocol, included in Appendix 2, has been revised as part of this AHA to include all works and developments in, on, or under the water proposed in areas of abalone habitat. The construction of underwater pipes or cable placement, installation of pilings, or other developments may have similar impacts as dredging and may be of interest if they occur in areas that contain abalone populations. Similar to dredging projects, however, if a project is small and the impacts on abalone populations are localized the transfer of individuals out of the area to another suitable habitat may be feasible, according to the protocol described in Appendix 2.

12 023

¾ Finfish aquaculture The impact of aquaculture on abalone has not been studied, except for a small experiment conducted in 2005-06 in the Broughton Group, Queen Charlotte Strait (Appendix 3). However, the effects of aquaculture on the benthic ecosystem in general have been reported, with an emphasis on soft sediment habitats. The nature of suspended solids, the rate at which they are released from the farm, water turnover time and sediment quality are key factors in determining the effects and impacts of eutrophication (Winsby et al. 1996; Black et al. 2001; Chamberlain et al. 2001; Nordvarg 2001). Environmental impacts from fish farms are similar to that found near other forms of organic pollution (sewage outflows, pulp mill effluent). Inputs which exceed the assimilative capacity of the benthic environment can result in increased nutrient levels; reduced oxygen concentration; lowered redox potentials; release of hydrogen sulphide, ammonium, and methane; and physical changes to the substrate (Winsby et al. 1996; Black et al. 2001). These changes can initiate successional changes in macrobenthic species diversity, abundance, and biomass, and continued inputs may result in reduced macrobenthic species richness and changes in community structure. Specifically, highly impacted areas directly under pens where waste accumulation is constant and substantial are characterized by few, if any, macroinvertebrate taxa (Winsby et al. 1996). This may be due to constant smothering or alterations in oxygen level. The intensity and duration of aquaculture impacts are related to current and water depth, extent of the disturbance, length of time the aquaculture facility is in operation, size of the facility, feeding rates, and sediment particle size. Facilities that are located in areas of low current and shallow water will have a greater accumulation of sediment directly under the pens; however an area of higher current may transport the sediments and chemicals further from the facility. Uneaten feed particles have a higher capacity than digested materials to impact the environment in terms of energy content and degradation rate, although some reports suggest that feed losses are being reduced in net pen culture (Black et al. 2001). Toxic effects from effluent sources, such as aquaculture facilities, may have varying degrees of impact depending on the invertebrate species and life stage. Haya et al. (2001) studied the effects of chemicals used in salmon farming on bivalve and lobster populations. They observed that late stage larvae and adult lobster were more susceptible to toxic effects from insecticides than earlier stage lobster larvae; however, there were no significant toxic effects on any of the bivalves included in the study. A study in California found that chemical effects on red abalone larvae in the laboratory may be analogous to processes occurring in the environment; zinc effects included abnormal larval shell development and reductions in successful metamorphosis (Hunt and Anderson 1989). The authors observed that toxin levels in southern California effluents have been found at concentrations higher than those used in the study, and declines in abalone abundance and growth rate have been observed near outfalls (Hunt and Anderson 1989). An experiment in the Broughton Group in Queen Charlotte Strait showed that abalone at control sites grew significantly more in two months than abalone placed within the finfish aquaculture tenures (Appendix 3). The animals outplanted were in poor health; this might have exaggerated the differences between tenures and control sites. In Queen Charlotte Strait, abalone densities found at index sites (0.080 abalone/m2) and random sites (0.024 abalone/m2) were lower than densities in the immediate vicinity of finfish aquaculture sites (0.125 abalone/m2), although this difference was not statistically significant (Davies et al. 2006). Surveyed sites at aquaculture tenures were not randomly selected, but were placed at the nearest good abalone habitat found by the pen-system which could have biased density estimates. Nonetheless, it appears that abalone

13 024

can live in the vicinity of pen-systems as four out of seven tenures surveyed in Queen Charlotte Strait had abalone at the site. The tenures where abalone were found have been in operation since 1991-1995 (M. Ayranto, Pan Fish Canada Ltd, Campbell River, BC; pers. comm.) indicating that abalone can survive several years within salmon aquaculture tenures. Abalone were also naturally present at other tenure sites in the Broughton Group in the vicinity of the experiment summarized above (C. Blackman, Marine Harvest Canada, Campbell River, BC; pers. comm.). Although there is no clear answer as to the impacts of finfish aquaculture on abalone populations, to be precautionary an impact on growth, at the minimum, should be assumed. A SARA permit is not required as it has been determined that the level of protection provided by the Fisheries Act is consistent with the level of protection required by section 58(1) of SARA. Therefore, when it has been determined that there is no harmful alteration, disruption or destruction of fish habitat (HADD) under section 35 of the Fisheries Act, there will not be any destruction of aquatic habitat under SARA (DFO 2004b). No direct mortality on abalone is expected from this activity as netpens are 10-15m deep and consequently have to be placed in deeper water than abalone preferred habitat. Harms on abalone populations will have to be assessed through the monitoring phase of ‘Impact Assessment Protocol’ (Appendix 2) and HADD will be authorized through the Fisheries Act, if need be. The mortality associated with this activity is expected to be low as long as the protocol outlined in Appendix 2 is followed.

¾ Log booms and log dumps While there are no studies specifically directed at these structures and their effects on abalone, log handling and storage is known to result in shading, water quality degradation, and modifications to habitat. Bark and wood debris may smother clams, mussels seaweed, kelp and sea grasses, and bark coverage may persist in the area for decades (Hanson et al. 2003). Accumulation of bark can result in locally decreased epifauna richness and abundance (Jackson 1986; Kirkpatrick et al. 1998). Storage of log materials and the loss of bark can result in the release of soluble organic compounds, increasing the oxygen demand within the area of accumulation. Increased oxygen demand can create anaerobic zones where toxic sulphide compounds are generated, particularly in brackish or marine waters. Shading can affect marine plant growth, including kelp and seagrass beds (Hanson et al. 2003), which can reduce the amount of food and cover available to adult abalone (Tegner et al. 2001). If abalone are present at a proposed log booms or log dump site considered for approval, they could be moved to a suitable location following the protocol outlined in Appendix 2. The expected mortality from this activity is near zero if the ‘Impact Assessment Protocol’ (Appendix 2) is followed. A SARA permit will be required if abalone are to be moved.

¾ Dredging Dredging can affect benthic and water column habitats by direct removal and burial of organisms, siltation effects, contaminant release and uptake, release of oxygen consuming substances, and alterations to hydrodynamic regimes and physical habitat. Physical factors including particle size, distribution, currents and compaction/stabilization processes can regulate recovery after dredging (Hanson et al. 2003). Recolonization can take up to 1-3 years in strong current areas, or as long as 10 years in lower current areas. Sensitive habitats may be damaged; dredging can physically destroy kelp and eelgrass beds, or modify current patterns and water

14 025

circulation affecting vegetation and larval settlement. Disposal of dredged material results in varying degrees of change in physical, chemical and biological characteristics of the substrate (Hanson et al. 2003). Discharges can smother benthic organisms and force mobile animals to leave. Erosion, slumping or lateral displacement of the surrounding bottom of such deposits can affect the substrate beyond the dump site by changing or destroying benthic habitat. Suspended solids reduce light penetration which may affect kelp beds as well (Hanson et al. 2003). In general, most dredging activities take place around rivers mouth or where sediments may accumulate due to wave action. As such abalone habitat is unlikely to be impacted, but the potential impacts should be kept in mind when habitat managers are considering this activity. If abalone are present in the proposed impacted zone, they could be moved to a suitable location following the protocol outlined in Appendix 2. The expected mortality from this activity is near zero. A SARA permit will be required if abalone are to be moved.

3. Abalone aquaculture As part of the strategy to rehabilitate northern abalone in BC, initial attempts were to include development of aquaculture methodology for use in stock rebuilding initiatives (Abalone Recovery Team 2002). This required the removal of mature abalone from the wild from a number of areas to provide broodstock for seed production at aquaculture facilities throughout BC. A protocol, reviewed by PSARC, is already in place which has been used in the latest broodstock collections and a similar protocol was used for the earlier collections in 1999 to 2001 (Lessard et al. 2002). Hatchery reared abalone that are not outplanted are sold to recoup costs and provide an economic incentive to the coastal community to support abalone recovery. Tracking protocols are in place to limit the avenues through which wild abalone may be “laundered” as cultured product. Provided current tracking protocols remain in place, no negative impact to the wild abalone population is expected. Hatchery reared abalone may be used for scientific research as an alternative to the use of wild abalone and can provide important information to fill knowledge gaps identified in the abalone recovery strategy. Most recently, a study of diseases in hatchery reared abalone raised in local seawater was conducted (S. Bower, DFO, Nanaimo, pers. comm.). At the time of writing, there is only one active abalone aquaculture facility located at the Bamfield Marine Science Centre and operated by the Bamfield Huu-ay-aht Community Abalone Project (BHCAP). There are abalone in two other aquaculture facilities, but they do not have an agreement with DFO to collect more broodstock. An agreement is in place between BHCAP and DFO outlining the required rebuilding work, amongst other conditions, necessary to justify broodstock collections. Although provisions are in place to return broodstock to the wild, the protocol assumes 100% mortality as a risk adverse approach. A survey is required in order to calculate the number of abalone that can be collected for broodstock that year. BHCAP requires about 100 or less abalone per year.

4. Fisheries on food supplies (i.e. kelp harvest) Adult abalone populations are affected by the availability of drift kelp as a food source (Tegner et al. 2001). In particular, the collapse of red abalone populations in California have been attributed to a combination of warmer water temperatures, fishing induced declines in adult

15 026

abalone density, and reduction of kelp densities due to El Nino related storm damage (Tegner et al. 2001). Ino (1968) reported that over-utilization of kelp contributed to the decline of abalone populations and the closure of abalone fisheries in some areas. The blacklip abalone (H. ruber) showed the strongest response to the removal of kelp canopy amongst common macroinvertebrate species in an Australian study; population numbers were reduced by half, likely due, in part, to increased predation of these individuals (Edgar et al. 2004). The Ministry of Agriculture and Lands (MAL) is responsible for the management of the commercial harvest of marine plants in BC. A Licence to Harvest Marine Plants is required to undertake a commercial harvest of any marine plant, including harvest for the purposes of the commercial Spawn-on-Kelp (or Roe-on-Kelp) fishery. The licence stipulates the species, quota, method of harvest and area of harvest. At present, all harvest of marine plants in BC is conducted by hand. There are conditions stipulated on the licence about where a plant may be cut, what portion of the plant may be harvested, and, on occasion, a condition that only one plant in four may be harvested in a given area to ensure that the integrity of the bed is not affected. MAL guidelines stipulate that no more than 20% of the total biomass of a marine plant bed may be harvested. This is to ensure long term sustainability of the resource and to minimize the impact on habitat. Most harvest levels are set substantially below the 20% maximum harvest level. In 2003, there were 69 licences issued to harvest marine plants. Over 250 tonnes of giant kelp (Macrocystis integrifolia), bull kelp (Nereocystis luetkeana), sea lettuce (Ulva spp.), bladderwrack (Fucus gardneri), sea asparagus (Salicornia pacifica, a vascular plant), and many other marine plants are commercially harvested and processed each year in BC (Ministry of Agriculture and Lands 2006). The expected mortality or harm from this activity is unknown. Although, over-utilization of kelp and canopy removal have been shown to affect abalone populations (Ino 1968; Edgar et al. 2004), the harvest levels set by MAL are probably low enough to ensure that abalone are not at risk of starvation. As such, given the harvest level guidelines, mortality caused by this activity is probably low to near zero.

5. Scientific research The majority of the scientific research currently ongoing involves non-destructive sampling. In recent years, because of the low abundance, destructive sampling has not been considered. Such sampling would give information on reproductive potential, metabolic condition, diets, and several diseases. Sex determination and an index of reproduction potential is possible without killing the individual, but this is not often done because of the increased diving time required. The recent research has involved mostly underwater survey where disturbance is minimal. An individual abalone may be wedged between rocks and may have to be picked up underwater for a more accurate measurement. The sunflower star Pycnopoda helianthoides is used to elicit an escape response to dislodge abalone without injury. This technique is in every protocol in use to survey abalone in BC. Several research projects, generally involved in rebuilding pilots, have brought the abalone out of the water in order to affix a numbered tag. Abalone are kept in circulating salt water or submerged in cages as long as possible to minimize exposure to air during tagging. Increased mortality associated with tagging appears to be largely due to increased predation on stressed abalone by P. helianthoides and not as a result of increased

16 027

injury from tagging, based on personal experience, video footage and recapture of tagged animals (J. Lessard, DFO Science, Nanaimo; B. Defrietas, Haida Fisheries Program, Queen Charlotte City; J. Harding, Kitasoo Fisheries Program, Klemtu). Information gathered during pilot projects has made valuable contributions to knowledge gaps identified in the recovery strategy. Therefore, non-destructive sampling, such as surveys, should be allowed to continue without a SARA permit. However, projects involving tagging or moving abalone to a different site, because of the potential mortality associated with exposure to air, should submit a SARA permit application to ensure proper procedures are followed. Mortality expected from scientific research is minimal.

6. Rebuilding efforts “The objective of ‘rehabilitation’, ‘rebuilding’ or ‘enhancement’ of abalone populations is to use a combination of methods to increase the depleted population size to a higher level of abundance and to increase population distribution by replacement of individuals in areas totally depleted of abalone patches” (Campbell et al. 2000b). Campbell et al. (2000b) reviewed and made recommendations of most, if not all, rebuilding methods available. They described each method and its relative success elsewhere in the world and gave pros and cons for each. However, several impacts were not considered. Rebuilding efforts that could harm abalone comprise (1) outplanting/seeding of larvae or juveniles from a hatchery and (2) adult aggregations usually collected from nearby wild populations. Both benefit reproduction output by increasing densities in a specific area thereby increasing fertilization rates. Adult aggregation has a potentially short-term as well as long-term benefit and the larvae or juveniles will potentially benefit the population in the long-term when the individuals reach reproduction age/size. There are two main risks associated with rebuilding efforts (1) the possible spread of diseases and (2) the loss of genetic diversity. Bower (2000) notes that comparatively little is known about northern abalone diseases and provides a literature review of available information. Bower (2000) classifies infectious diseases of abalone into three main categories. Category 1 identifies 6 pathogens, including Labyrinthuloides haliotidis, which decimated young cultured abalone in BC. Category 1 pathogens have been observed to cause severe disease in wild or cultured abalone; some pathogens have been reported to cause mass mortality in only one area but are known to be present worldwide. Category 2 includes pathogens such as nematode and trematode species that have been observed to cause infection but have not been linked to significant mortality or are infective only during a restrictive part of the life cycle. Category 3 includes organisms that can have serious impact under appropriate environmental conditions only. Transplantation of abalone from one area to another can extend the natural range of pathogens or introduce pathogens to different environmental conditions with unpredictable potential impact. Bower (2000) cautions that transplantation from infected areas is to be avoided, and transplant animals should be quarantined to ensure they are free of infection prior to introduction. To safeguard genetic variability, Withler et al. (2001) recommended that the number of abalone broodstock used to produce larvae or juveniles for outplanting to the wild should be at least 50 and preferably 100, with equal number of males and females. For the re-introduction or enhancement of endangered species, it has been recommended that a minimum of 20-30 animals

17 028

be used as a founding population (Ralls and Ballou 1992) and that animals be collected from several locations to provide adequate genetic diversity (Templeton 1990).

¾ Larvae and juvenile outplanting Releasing hatchery reared abalone larvae has met with limited success, with most studies concluding that larval release is not suitable for large scale restocking (Roberts et al. 1999). Successes have been reported for many areas of the world; to date Asia is the largest producer of hatchery reared abalone and South Africa as the second largest producer (Troell et al. 2006). In both Asia and South Africa, abalone hatcheries produce abalone for market sale as well as seeding of kelp beds for stock enhancement or rehabilitation of overfished stocks (Simizu and Uchino 2004, Troell et al. 2006). The disease risk is almost nil for larvae outplanting as the larvae do not come in contact with the adults potentially carrying the diseases and the duration of stay in the hatchery is short (2 weeks) (Pearce et al. 2003). A subsample of the juveniles to be outplanted can be tested prior to outplanting to determine if they are disease-free. All transfers of abalone to and from the wild are required to be permitted by the federal/provincial Introductions and Transfers committee (I&T) who will consider risk of disease resulting from the transfer. For SARA listed species, the I&T committee requires a SARA permit before an I&T permit is given. The harm associated with this activity is minimal if proper procedures are followed (i.e. I&T and SARA permits). Under the right conditions, outplanting can be successful and the benefits outweigh the risks.

¾ Adult aggregation Moving and aggregating mature abalone to increase the local density and consequently their reproductive potential could help in their rehabilitation. Campbell et al. (2003) discussed the implications of transplanting abalone from ‘poor’ to ‘good’ habitats to increase survival, growth and reproductive potential, as a rebuilding technique for northern abalone in BC. In the Broken Group Islands, a DFO-Parks Canada collaborative project is exploring methods and factors that improve abalone reproduction and recruitment by increasing adult densities and growth to increase reproductive success and capacity of normal and stunted ‘surf’ abalone. Preliminary results indicate that aggregation was successful in increasing densities of juvenile abalone at the experimental sites (J. Lessard, pers. comm.). Parks Canada, using night diving when small abalone come out of cryptic habitats, showed that <3 years old abalone were more abundant closer to one aggregation site (T. Tomascik, Parks Canada Agency, Vancouver; pers. comm.). In California, mature green abalone, H. fulgens, were collected from a site of abundance, tagged, and clustered in two sites where abalone populations had severely declined (Tegner and Butler 1985). Within the first year, the transplanted abalone showed signs of reproduction and were estimated to have a 10% natural mortality. In addition, large numbers of juveniles the following three years indicated an increase in recruitment in the transplanted areas. Abundance of the large tagged individuals declined significantly during the same time period, however, and lack of tagged shells suggested poaching was the major cause of mortality (Tegner 2000). As aggregation requires taking abalone out of the water, a SARA permit application will be necessary. Disease testing requires destructive sampling and will not be allowed (see Scientific Research section under Steps 6-7), therefore collection sites should be within the same geographic region to avoid the spread of disease as well as long exposure to air while travelling.

18 029

Adult aggregations can successfully enhance reproductive output of local population; as such the benefits outweigh the risks. So far, ongoing rebuilding projects have occurred in parallel with Coast Watch, or similar, programs where surveillance is increased. The expected mortality from this activity is minimal.

7. Dismissed activities

¾ Military activities There are no known military activities that impact abalone.

¾ Bycatch Abalone are not a legal bycatch of any fishery.

¾ Detrimental impacts on habitats by fishing activities Fisheries for which there are habitat impacts take place deeper than abalone distribution.

¾ Ecotourism and recreation Except for poaching, presumably due to ignorance or blatant disregard, there are no impacts from these activities.

¾ Shipping, transport, and noise Abalone are not known to be affected by noise and they are not affected by shipping and transport.

Step 8: Aggregate of the total mortality/harm from human activities and contrast with model. In general, little mortality is expected from legislated or permitted human activities. The direct and indirect causes of mortality are addressed through a published protocol for broodstock survey and collection and the protocol described in Appendix 2 for works and developments. A directed First Nations fishery is not considered at this time, at least until a protocol is in place which will require consultation and agreement between groups of minimum requirements for abalone densities, rebuilding work, monitoring requirements, and timing and enforcement of harvest. Harm may be expected from finfish aquaculture activity or other types of works and developments, but more information is needed and the protocol in Appendix 2 addresses this through a monitoring phase after approval.

Phase 3: Options

Summary and conclusions The northern abalone is vulnerable to over-exploitation due to its patchy distribution, short larval period, slow growth, long life, and low or sporadic recruitment. Mature individuals also tend to accumulate in shallow water where they may be easy to access. Survey results at index sites in QCI and CC showed that population densities declined by >80% during the period of 1978-2002

19 030

(Atkins et al. 2004; Lessard et al. 2007). The short-term objectives of the recovery strategy (Abalone Recovery Team 2002) have not been achieved and abalone populations continue to decline or oscillate at low levels. The projected population trajectory does indicate that survival and recovery are in jeopardy, but not from the human activities described in Steps 6-7. The most probable causes are illegal harvest and the resulting low mature densities leading to poor recruitment. The continued expansion of sea otters will further accelerate the decline of abalone abundances. Measurable targets, using standard surveys of abalone, should include: (1) annual mortality estimates; (2) density estimates of emergent or “exposed” reproductive broodstock (>70 mm SL); and (3) frequency of patch sizes of abalone. Both short-term (10 years) and long-term objectives are proposed for each of the measurable target. The mortality caused by poaching and sea otter predation, where they are present, is large when compared to aggregated mortality of the activities described in Steps 6-7 (those that can be permitted under Section 73 of SARA). The mortality rates (Z) calculated in Appendix 1, 0.21- 0.36, were higher than natural mortality rates. This indicates that illegal harvest was probably still ongoing and was a major source of mortality. Sea otter caused mortality rates have been estimated to be between 0.3 and 1.3. Literature indicates that in sea otter areas, initial abalone mortality rates are in excess of sustainability based on the current model. There are still knowledge gaps as to the long term implications on abalone due to ecosystem changes. Ecological and abalone population parameters need to be measured in areas without sea otters, in areas where sea otters have recently arrived and in areas where sea otters have been established for several years. As otters expand, it will be crucial to have this information if we are going to recover abalone. Halting the decline and increasing abalone abundance to sustainable levels as currently defined in the recovery strategy (Abalone Recovery Team 2002) will not be possible in areas occupied by sea otters. Model simulations indicate that with mortality rates >0.20, abalone populations will decline further. As abalone are not recovering, the maximum human-induced mortality should be near zero as an increase in mortality rate will further reduce the abalone populations in BC. We suggest that First Nations harvest be considered after a protocol is agreed upon in order to increase stewardship and help reduce poaching. The protocol development for First Nation harvest will require consultation and agreement between groups on minimum requirements for abalone densities, rebuilding work, monitoring protocol, timing of harvest and enforcement. SARA permits will be required for broodstock collections, tagging as part of research, except in situ tagging, and movement of abalone as part of rebuilding efforts or approved works and developments as described in the protocol in Appendix 2. Scientific research and rebuilding projects are essential to the recovery of northern abalone. To minimize handling stress, SARA permit conditions, similar to those described in Appendix 2, should continue to be set. When appropriate, SARA permit conditions should also include an I&T permit requirement to consider risk of disease. Reports are required as a condition of the SARA permits, including reporting of mortality, in which case alternate measures can be devised.

20 031

Recommendations 1. The proposed measurable short-term objectives over the next 10 years are to (1) reduce annual estimated mortality rates to <0.20, (2) ensure that mean densities of mature (>70 mm SL) abalone increase to >0.32/m2 at surveyed index sites (twice the current densities in 2001 CC and 2002 QCI) and (3) increase the proportion of quadrats with abalone at surveyed index sites to >40%. 2. The proposed measurable long-term objective for the next 30 years are to (1) reduce and maintain annual estimated mortality rate to <0.15, (2) ensure that mean densities of mature (>70 mm SL) abalone increase to >0.5/m2 at surveyed index sites and (3) increase the proportion of quadrat with >1 mature abalone to >20%. 3. No direct mortality should be allowed at this time. 4. Illegal harvest/poaching on abalone by humans should continue to be actively discouraged through enforcement and public education. 5. Additional research in the field and computer ecological simulations are required to further understand abalone population dynamics as well as sea otter and abalone interactions on small and large spatial and temporal scales. 6. Consultations could be considered to enhance First Nation stewardship through agreements under which small enhancement projects are carried out at specific sites by local First Nation communities followed by a small harvest of abalone with strict controls when agreed upon conditions are achieved (e.g., abalone density reaches a certain threshold). The harvest would only take place if enhancement activities have been carried out. 7. Research and rebuilding projects should continue. If moving abalone or collecting broodstock, a SARA permit is necessary. 8. Impact assessment protocol outlined in Appendix 2 should be followed for proposed works and developments on, in or under the water.

References Abalone Recovery Team. 2002. National Recovery Strategy for the Northern Abalone (Haliotis kamtschatkana) in Canada. 22pp. http://www.-comm.pac.dfo- mpo.gc.ca/pages/consultations/fisheriesmgmt/abalone/documents/AbaloneRecovStrategye. htm. Adkins, B.E. 1996. Abalone surveys in south coast areas during 1982, 1985 and 1986. Can. Tech. Rep. Fish. Aquat. Sci. 2089: 72-96. Adkins, B.E., and Stefanson, A.P. 1979. North coast abalone survey in harvested areas, November 1978. Fish. Mar. Serv. Manuscr. Rep. 1500: 15 p. Allee, W.C., A.E. Emerson, O. Park, T. Park, and K.P. Schmidt. 1949. Principles of Animal Ecology. Saunders, Philadelphia. 837 p. Atkins, M., and Lessard, J. 2004. Survey of northern abalone, Haliotis kamtschatkana, populations along north-west Vancouver Island, British Columbia, May 2003. Can. Manuscr. Rep. Fish. Aquat. Sci. 2690: 12 p.

21 032

Atkins, M., Lessard, J., and Campbell, A. 2004. Resurvey of northern abalone, Haliotis kamtschatkana, populations in Southeast Queen Charlotte Islands, British Columbia, April 2002. Can. Manuscr. Rep. Fish. Aquat. Sci. 2704: 37 p. Babcock, R., and Keesing, J. 1999. Fertilization biology of the abalone Haliotis laevigata: laboratory and field studies. Can. J. Fish. Aquat. Sci. 56: 1668-1678 Black, K., Angel, D., and Eden, N. 2001. A review of the environmental impacts of marine cage aquaculture process of biofiltration relevant to impact mitigation, the biological process of marine invertebrates relevant to biofiltration, and biofouling on artificial surfaces. BIOFAQS Q5RS-2000-30305 Tech. Annex Annual Report Dec. 2001. Boutillier, J.A., Carolsfeld, W., Breen, P.A., and Bates, K. 1984. Abalone survey in the Estevan Group and Aristazabal Island, May 1983. Can. Manuscr. Rep. Fish. Aquat. Sci. 1747: 60 p. Boutillier, J.A., Carolsfeld, W., Breen, P.A., Farlinger, S., and Bates, K. 1985. Abalone resurvey in the southeast Queen Charlotte Islands, July 1984. Can. Manuscr. Rep. Fish. Aquat. Sci. 1818: 87 p. Bower, S. 2000. Infectious diseases of abalone (Haliotis spp.) and risks associated with transplantation. Can. Spec. Public. Fish. Aquat. Sci. 130: 111-122. Breen, P.A. 1980. Muscled mollusc: the northern abalone. Diver 6(6): 26-28. Breen, P.A. 1986. Management of the British Columbia fishery for northern abalone (Haliotis kamtschatkana). Can. Spec. Public. Fish. Aquat. Sci. 92: 300-312. Breen, P.A., and Adkins, B.E. 1979. A survey of abalone populations on the east coast of the Queen Charlotte Islands, August 1978. Fish. Mar. Serv. Manuscr. Rep. 1490: 125 p. Breen, P.A., and Adkins, B.E. 1980a. Observations of abalone populations in Emily Carr Inlet and Lotbinière Bay, April 1980. Can. Manuscr. Rep. Fish. Aquat. Sci. 1576: 17 p. Breen, P.A., and Adkins, B.E. 1980b. Spawning in a British Columbia population of northern abalone, Haliotis kamtschatkana. Veliger 23: 177-179. Breen, P.A., and Adkins, B.E. 1981. Abalone surveys and tagging conducted during 1979. Can. Manuscr. Rep. Fish. Aquat. Sci. 1623: 88 p. Breen, P.A., Adkins, B.E., and Heritage, G.D. 1978a. Observations of abalone and subtidal communities made during a survey of Queen Charlotte Strait and upper Johnstone Strait areas. Fish. Mar. Serv. Manuscr. Rep. 789: 91 p. Breen, P.A., Adkins, B.E., and Sprout, P.E. 1982. Abalone populations on the west coast of Banks Island, June 1980. Can. Manuscr. Rep. Fish. Aquat. Sci. 1640: 42 p. Breen, P.A., Stefanson, A.P., and Adkins, B.E. 1978b. North coast abalone surveys in harvested areas, spring 1978. Fish. Mar. Serv. Manuscr. Rep. 1480: 61 p. Brown, L.D. 1991. Genetic variation and population structure in the blacklip abalone, Haliotis rubra. Aust. J. mar. Freshwat. Res. 42: 47-90. Brown, L.D., and Murray, N.D. 1992. Population genetics, gene flow, and stock structure in Haliotis rubra and Haliotis laevigata. pp. 24-33 in Shepherd, S.A., Tegner, M.J., Guzman del Proo, S.A. (eds). Abalone of the World - Biology, Fisheries and Culture. Fishing News Books, Oxford. Burton, R.S., and Tegner, M.J. 2000. Enhancement of red abalone Haliotis rufescens stocks at San Miguel Island: reassessing a success story. Mar. Ecol. Prog. Ser. 202: 303-308. Campbell, A. 1997. Possible criteria for reopening the northern abalone (Haliotis kamtschatkana) fishery in British Columbia. DFO Can. Stock Assess. Sec. Res. Doc. 1997/64: 47 p.

22 033

Campbell, A. 2000. Review of northern abalone, Haliotis kamtschatkana, stock status in British Columbia. Can. Spec. Publ. Fish. Aquat. Sci. 130. pp. 41-50. Campbell, A., Brouwer, D., Rogers, J., and Miller, D.C. 2000a. Abalone resurvey in South East Queen Charlotte Islands, 1998. Can. Manuscr. Rep. Fish. Aquat. Sci. 2528: 29 p. Campbell, A., and Cripps, K. 1998. Survey of abalone populations at Stryker Island, Tribal Group and Simonds Group, Central Coast of British Columbia, May 1997. Can. Manuscr. Rep. Fish. Aquat. Sci. 2451: 21 p. Campbell, A., Lessard, J., and Jamieson, G.S. 2003. Fecundity and seasonal reproduction of northern abalone, Haliotis kamtschatkana, in Barkley Sound, Canada. J. Shellfish Res. 22: 811-818. Campbell, A., Lucas, B., and Parker, G. 2000b. Discussion on an experimental approach for northern abalone stock rebuilding in British Columbia. DFO Can. Stock Assess. Sec. Res. Doc. 2000/047: 24 p. Campbell, A., Manley, I., and Carolsfeld, W. 1992. Size at maturity and fecundity of the abalone, Haliotis kamtschatkana (Jonas), in northern British Columbia. Can. Manuscr. Rep. Fish. Aquat. Sci. 2169: 47-65. Campbell, A., Winther, I., Adkins, B., Brouwer, D., and Miller, D. 1998. Survey of the northern abalone (Haliotis kamtschatkana) in the central coast of British Columbia, May 1997. DFO Can. Stock Assess. Sec. Res. Doc. 1998/89: 28 p. Carolsfeld, W., Farlinger, S., Kingzett, B.C., Sloan, N.A., and Thomas, G. 1988. Abalone resurvey in the southeast Queen Charlotte Islands, June 1987. Can. Manuscr. Rep. Fish. Aquat. Sci. 1966: 90 p. Carter, S.K., VanBlaricom, G.R., and Allen, B.L. 2007. Testing the generality of the trophic cascade paradigm for sea otters: a case study with kelp forests in northern Washington, USA. Hydrobiologia 579: 233-249. Castilla, J.C., Manríquez, P., Alvarado, J., Rosson, A., Pino, C., Espoz, C., Soto, R., Oliva, D., and Defeo, O. 1998. Artisanal “Caletas” as units of production and co-managers of benthic invertebrates in Chile. Can. Spec. Publ. Fish. Aquat. Sci. 125: 407-413. Chamberlain, J., Fernandes, T.F., Read, P., Nickell, T.D., and Davies, I.M. 2001. Impacts of biodeposits from suspended mussel (Mytilus edulis L.) culture on the surrounding surficial sediments. ICES J. Mar. Sci. 58: 411–416. Cooper, J., Wieland, M., and Hines, A. 1977. Subtidal abalone populations in an area inhabited by sea otters. The Veliger 20: 163-167. Cripps, K., and Campbell, A. 1998. Survey of abalone populations at Dalain Point and Higgings Pass, Central Coast of British Columbia, 1995-96. Can. Manuscr. Rep. Fish. Aquat. Sci. 2445: 31 p. Davies, K., Atkins, M., and Lessard, J. 2006. Survey of northern abalone, Haliotis kamtschatkana, populations in Queen Charlotte and Johnstone Straits, British Columbia, May 2004. Can. Manuscr. Rep. Fish. Aquat. Sci. 2743: 17 p. Deacon, R.T. 1989. An empirical model of fishery dynamics. J. Env. Econ. Manag. 16: 167- 183. DFO. 2004a. Northern Abalone. DFO Science Stock Status Report. DFO. 2004b. Practitioners guide to the Species at Risk Act (SARA) for habitat management staff. DFO. 2004c. Proceedings of the National Peer Review Meeting on the Level of Allowable Harm for Newfoundland and Labrador Atlantic Cod, Laurentian North Atlantic Cod, Cusk

23 034

and Bocaccio in Support of Species at Risk. DFO Can. Sci. Advis. Sec. Proceed. Ser. 2004/040: 65 p. DFO. 2005. DFO Marine Fish Habitat Information Requirements (HIR) for Finfish Aquaculture Projects. Dowling, M.A., Hall, S.J., and McGarvey, R. 2004. Assessing population sustainability and response to fishing in terms of aggregation structure for greenlip abalone (Haliotis laevigata) fishery management. Can. J. Fish. Aquat. Sci. 61: 247-259 Edgar, G.J., Barrett, N.S., Morton, A.J., and Samson, C.R. 2004. Effects of algal canopy clearance on plant, fish and macroinvertebrate communities on eastern Tasmanian reefs. J. Exp. Mar. Biol. Ecol. 312: 67-87. Farlinger, S., and Bates, K.T. 1986. Abalone survey in the Estevan Group and Aristazabal Island, June 1985. Can. Manuscr. Rep. Fish. Aquat. Sci. 1896: 45 p. Farlinger, S., Thomas, G.A., Winther, I., and Carolsfeld, W. 1991. Abalone resurvey in the Estevan Group and Aristazabal Island, June 1989. Can. Manuscr. Rep. Fish. Aquat. Sci. 2104: 39 p. Hanson, J., Helvey, M., and Strach, R. 2003. Non-fishing impacts to essential fish habitat and recommended conservation methods. Nat. Marine Fish. Service (NOAA Fisheries) Version 1. Southwest Region, Long Beach CA. Harbo, R. 1997. Abalone Dive Fishery (closed). Can. MS Rep. Fish. Aquat. Sci. 2369: 86-92. Haya, K., Burridge, L.E., and Chang, B.D. 2001. Environmental impact of chemical wastes produced by the salmon aquaculture industry. ICES J. Mar. Sci. 58: 492-496. Hines, A.H. and Pearse, J.S. 1982. Abalones, shells, and sea otters: dynamics of prey populations in central California. Ecology 63: 1547-1560. Hunt, J.W., and Anderson, B.S. 1989. Sublethal effects of zinc and municipal effluents on larvae of the red abalone Haliotis rufescens. Mar. Biol. 101: 545-552. Ino, T. 1968. The abalone science and its propagation in Japan. Fish. Res. Bd Transl. Ser. 1078: 209 p. Jackson, R.G. 1986. Effects of bark accumulation on benthic infaunal at a log transfer facility in Southeast Alaska. Mar. Pol. Bull. 17(6): 258-262. Jamieson, G.S. 2001. Review of status of northern abalone, Haliotis kamtschatkana, in Canada. Canadian Field-Naturalist 115: 555-563. Jones, R., DeFreitas, B., Sloan, N., Lee, L., von Boetticher, K., and Martin, G. 2003. Abalone stewardship in Haida Gwaii: forging a long term commitment. Can. Tech. Rep. Fish. Aquat. Sci. 2482: 5-19. Jubinville, B. 2000. Enforcing the fishery closure for northern (pinto) abalone (Haliotis kamtschatkana) in British Columbia (abstract). Can. Spec. Public. Fish. Aquat. Sci. 130: 52. Kirkpatrick, B., Shirley, T.C. and O’Clair, C.E. 1998. Deep-water bark accumulation and benthos richness at log transfer and storage facilities. Alaska Fishery Res. Bull. 5:103–115. Lessard, J., Atkins, M. and Campbell, A. 2007. Resurvey of northern abalone, Haliotis kamtschatkana, populations along the central coast of British Columbia, April 2001. Can. Manuscr. Rep. Fish. Aquat. Sci. 2791: 35 p. Lessard, J., Campbell, A., and Hajas, W. 2002. Survey protocol for the removal of allowable numbers of northern abalone, Haliotis kamtschatkana, for use as broodstock in aquaculture in British Columbia. DFO Can. Sci. Advis. Sec. Res. Doc. 2002/126: 40 p.

24 035

Levitan, D.R., and Sewell, M.A. 1998. Fertilization success in free-spawning marine invertebrates: review of the evidence and fisheries implications. Can. Spec. Public. Fish. Aquat. Sci. 125: 159-164 Levitan, D.R., Sewell, M.A., and Chia, F.-S. 1992. How distribution and abundance influence fertilization success in the sea urchin Strongylocentrotus franciscanus. Ecology 73: 248- 254 Lowry, L.F., and Pearse, J.S. 1973. Abalones and sea urchins in an area inhabited by sea otters. Mar. Biol. 23: 213-219 Lucas, B.G., Campbell, A., and Cripps, K. 1999. Resurvey of abalone populations at Tribal Group, Simonds Group and Stryker Island, Central Coast of British Columbia, 1998. Can. Manuscr. Rep. Fish. Aquat. Sci. 2487: 18 p. McShane, P.E. 1992. Early life history of abalone: a review. pp. 120-138 In Shepherd, S.A., Tegner, M.J., Guzman del Proo, S.A. (eds), Abalone of the World - Biology, Fisheries and Culture. Fishing News Books, Oxford. McShane, P.E. 1995a. Estimating the abundance of abalone: the importance of patch size. Marine and Freshwater Research 46: 657-662 McShane, P.E. 1995b. Recruitment variation in abalone: its importance to fisheries management. Aust. J. mar. Freshwat. Res. 46: 555-570 McShane, P.E., Black, K.P., and Smith, M.G. 1988. Recruitment processes in Haliotis rubra (Mollusca: Gastropoda) and regional hydrodynamics in southeast Australia imply localised dispersal of larvae. J. Exp. Mar. Biol. Ecol. 124: 175-203. Ministry of Agriculture and Lands. 2006. (http://www.agf.gov.bc.ca/fisheries/commercial/commercial_mp.htm; accessed Nov. 06, 2006) Nichol, L.M., Watson, J.C., Ellis, G.M., and Ford, J.K.B. 2005. An assessment of abundance and growth of the sea otter population (Enhydra lutris) in British Columbia. DFO Can. Sci. Advis. Sec. Res. Doc. 2005/094: 22 p. Nordvarg, L. 2001. Predictive models and eutrophication effects of fish farms. Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 602. 44 pp. Uppsala, Sweden. Otsfeld. 1982. Foraging strategies and prey switching in the California sea otter. Oecologia 53: 170-178. Pearce, C.M., Ågerup, P., Alabi, A., Renfrew, D., Rosser, J., Whyte, G., and Yuan, F. 2003. Recent progress in hatchery production of pinto abalone, Haliotis kamtschatkana, in British Columbia, Canada. Can. Tech. Rep. Fish. Aquat. Sci. 2482: 29-44. Pollard, S. 1992. Red abalone, Haliotis rufescens, relative impacts of recreational fisheries and sea otter predation on the abundance, size frequency, and microhabitat distribution of red abalone populations in central and northern California. M.Sc. Thesis, University of Santa Cruz, California. Prince, J.D., Sellers, T.L., Ford, W.B., and Talbot, S.R. 1987. Experimental evidence for limited dispersal of haliotid larvae (genus Haliotis; Molluscs: Gastropoda). J. Exp. Mar. Biol. Ecol. 106: 243-263. Prince, J., Walters, C., Ruiz-Avila, R., and Sluczanowski, P. 1998. Territorial user’s rights and the Australian abalone (Haliotis sp.) fishery. Can. Spec. Publ. Fish. Aquat. Sci. 125: 367- 375. Quayle, D.B. 1971. Growth, morphometry and breeding in the British Columbia abalone (Haliotis kamtschatkana Jonas). Fish. Res. Bd Can. Tech. Rep. Ser. 279: 84 p.

25 036

Ralls, K., and Ballou, J.D. 1992. Managing genetic diversity in captive breeding and reintroduction programs. Trans. 57th N. Am. Wildl. Nat. Res. Conf. : 263-282. Roberts, R.D., Kawamura, T., and Takami, H. 1999. Abalone recruitment - an overview of some recent research in New Zealand, Australia and Japan. Bull. Tohoku Natl. Fish. Res. Inst. 62: 95-107. Sea Otter Recovery Team. 2003. National Recovery Strategy for the Sea Otter (Enhydra lutris) in Canada. Fisheries and Oceans Canada. 60 pp. Shepherd, S.A., and Baker, J.L. 1998. Biological reference points in an abalone (Haliotis laevigata) fishery. Can. Spec. Public. Fish. Aquat. Sci. 125: 235-245 Shepherd, S.A., and Breen, P.A. 1992. Mortality in abalone: its estimation, variability and causes. in Shepherd, S.A., Tegner, M.J., Guzman del Proo, S.A. (eds). Abalone of the World - Biology, Fisheries and Culture. Fishing News Books, Oxford. Shepherd, S.A., and Brown, L.D. 1993. What is an abalone stock: implications for the role of refugia in conservation. Can. J. Fish. Aquat. Sci. 50: 2001-2009 Shepherd, S.A., and Partington, D. 1995. Studies on southern Australian abalone (genus Haliotis). XVI. Recruitment, habitat and stock relations. Marine and Freshwater Research 46: 669-680. Shepherd, S.A., Woodby, D., Rumble, J.M., and Avalos-Borja, M. 2000. Microstructure, chronology and growth of the pinto abalone, Haliotis kamtschatkana, in Alaska. J. Shellfish Res. 19, 219-228. Simenstad, C.A., Estes, J.A., and Kenyon, K.W. 1978. Aleuts, sea otters, and alternate stable- state communities. Science 200: 403-411 Simizu, T. and Uchino, K. 2004. Effects of extensive seeding on abalone Haliotis discus discus abundance on the Pacific coast of boso peninsula, Japan. J. Shellfish Res. 23: 1209-1211. Sinclair, A.R.E., and Byrom, A.E. 2006. Understanding ecosystem dynamics for conservation of biota. J. Animal Ecol. 75: 64-79. Sloan, N.A., and Breen, P.A. 1988. Northern abalone, Haliotis kamtschatkana in British Columbia: fisheries and synopsis of life history information. Can. Spec. Public. Fish. Aquat. Sci. 103: 46 pp Soulé, M.E., Estes, J.A., Berger, J., and Martinez del Rio, C. 2003. Ecological effectiveness: Conservation goals for interactive species. Conserv. Biol. 17: 1238-1250. Tegner, M.J. 1993. Southern California abalones: can stocks be rebuilt using marine harvest refugia? Can. J. Fish. Aquat. Sci. 50: 2010-2018 Tegner, M.J. 2000. Abalone (Haliotis spp.) enhancement in California: what we've learned and where we go from here. Can. Spec. Public. Fish. Aquat. Sci. 130: 61-71. Tegner, M.J., and Butler, R.A. 1985. Drift-tube study of the dispersal potential of green abalone (Haliotis fulgens) larvae in the southern California Bight: implications for the recovery of depleted populations. Mar. Ecol. Prog. Ser. 26: 73-84. Tegner, M.J., Haaker, P.L., Riser, K.L., and Vilchis, L.I. 2001. Climate variability, kelp forests, and the southern California red abalone fishery. J. Shellfish Res. 20: 755-763. Templeton, A.R. 1990. The role of genetics in captive breeding and reintroduction programs for species conservation. Endangered Species Update 8: 15-17. Thomas, G., and Campbell, A. 1996. Abalone resurvey in Aristazabal Island, the Estevan Group and Banks Island, June 1993. Can. Tech. Rep. Fish. Aquat. Sci. 2089: 97-109. Thomas, G., Farlinger, S., and Carolsfeld, W. 1990. Abalone resurvey in the southeast Queen Charlotte Islands in 1990. Can. Manuscr. Rep. Fish. Aquat. Sci. 2099: 66-82.

26 037

Troell, M., Robertson-Andersson, D., Anderson, R.J., Bolton, J.J., Maneveldt, G., Halling, C., and Probyn, T. 2006. Abalone farming in South Africa: An overview with perspectives on kelp resources, abalone feed, potential for on-farm seaweed production and socioeconomic importance. Aquaculture 257: 266-281. Wallace, S.S. 1999. Evaluating the effects of three forms of marine reserve on northern abalone populations in British Columbia, Canada. Conserv. Biol. 13: 882-887. Watson, J. 1993. The effects of sea otter (Enhydra lutris) foraging on shallow rocky communities off northwestern Vancouver Island, British Columbia. Ph.D., University of California, Santa Cruz. 169 p. Watson, J. 2000. The effects of sea otters (Enhydra lutris) on abalone (Haliotis spp.) populations. Can. Spec. Public. Fish. Aquat. Sci. 130: 123-132. Watson, J.C., Ellis, G.M., Smith, T.G., and Ford, J.K.B. 1997. Updated status of the sea otter, Enhydra lutris, in Canada. Canadian Field-Naturalist 111: 277-286. Wendell, F. 1994. Relationship between sea otter range expansion and red abalone abundance and size distribution in central California. Calif. Fish Game 80: 45-56. Wild, P.W. and Ames, J.A. 1974. A report on the sea otter, Enhydra lutris L., in California. Calif. Fish Game Mar. Res. Tech. Rep. 20. Winsby, M., Sander, B., Archibald, D., Daykin, M., Nix, P., Taylor, F.J.R., and Munday, D. 1996. The environmental effects of salmon netcage culture in British Columbia - A literature review. Report prepared for the Ministry of Environment Lands and Parks, Victoria, BC, by Hatfield Consultants Ltd and EVS Environmental Consultants. Winther, I., Campbell, A., Thomas, G.A., Adkins, B.E., and Clapp, B.G. 1995. Abalone resurvey in the southeast Queen Charlotte Islands in 1994. Can. Manuscr. Rep. Fish. Aquat. Sci. 2273: 43 p. Withler, R.E., A. Campbell, S. Li, K.M. Miller, D. Brouwer, and B.G. Lucas. 2001. High levels of genetic variation in northern abalone Haliotis kamtschatkana of British Columbia. DFO Can. Sci. Advis. Sec. Res. Doc. 2001/097: 27 p. Withler, R.E., Campbell, A., Li, S., Miller, K.M., Brouwer, D., Supernault, K.J., and K.M. Miller. 2003. Implications of high levels of genetic diversity and weak population structure for the rebuilding of northern abalone in British Columbia, Canada. J. Shellfish Res. 22: 839-847.

27 038

Table 1. Human-influenced relative mortality (direct and indirect) to abalone populations, in order of importance. Relative mortality scale: High = F >0.10; Medium = 0.10>F>0.03; Low = F < 0.03; Minimal = F < 0.01 Human-influenced activity Relative Comment Mortality Illegal harvest High Sea otter predation High Possible directed fishing Medium - Prior rebuilding work necessary and harvest localized allowed only above set target Habitat alterations - Finfish Low Assumes ‘Impact Assessment Protocol’ aquaculture (Appendix 2) is followed Habitat alterations - Log booms Low Follow ‘Impact Assessment Protocol’ and log dumps (Appendix 2) Habitat alterations – Dredging Low Follow ‘Impact Assessment Protocol’ (Appendix 2) Abalone aquaculture Low Combined with rebuilding work: beneficial Fisheries on food supplies (i.e. Low – kelp harvest) localized Scientific research Minimal Incidental mortality only (e.g., tagging), no direct mortality allowed Rebuilding efforts - Larvae and Minimal All outplanting approved by I&T juvenile outplanting Committee: beneficial Rebuilding efforts - Adult Minimal Requires a SARA permit: beneficial aggregation

28 039

5 ) 2 0.8 4 0.6 0.4 3 0.2 CC 0 1985 1989 1993 1997 2001 2005 2

1 QCI Mean Total Density (abalone/m 0 1977 1981 1985 1989 1993 1997 2001 2005 Year

Figure 1. Mean density of exposed abalone, all sizes, from all surveys in the Central Coast (CC) (solid line) and the Queen Charlotte Islands (QCI) (dashed line). Error bars represent two standard errors. Inset graph displays greater resolution of densities for survey years after 1985.

29 040

1.6

) 0.5 2 1.4 0.4 0.3 1.2 0.2 1.0 0.1 0.8 0 1984 1988 1992 1996 2000 2004 0.6 0.4 CC 0.2 QCI Mean Large Density (abalone/m Mean Large Density 0.0 1977 1981 1985 1989 1993 1997 2001 2005 Year

Figure 2. Mean density of large (≥100mm SL) abalone from all surveys in the Central Coast (CC) (solid line) and the Queen Charlotte Islands (QCI) (dashed line). Error bars represent two standard errors. Inset graph displays greater resolution of densities for survey years after 1985.

30 041

100 27 CC 80 33 62 26 33 60 QCI 70 70 69 70 47 40 Short-term recovery objective 55 66 68 20 68 % sites with large abalone %

0 1977 1981 1985 1989 1993 1997 2001 2005 Year

Figure 3. Percent of repeated index sites with large abalone (≥100 mm SL) from surveys in the Central Coast (CC) (solid line) and the Queen Charlotte Islands (QCI) (dashed line) (some years were excluded due to small number of ‘index sites’ in those years). Numbers are sample sizes.

31 042

90 1978 80 1990 70 2002 60

40 Percent 30

0 012345678910>10 Abalone/m2

90 1978-80 80 1989 70 2001 60

40 Percent 30

0 012345678910>10 Abalone/m2

Figure 4. Percentage of surveyed quadrats with different abalone counts (all sizes) from all sites. Top graph: Queen Charlotte Islands during 1978, 1990 (last survey year when fisheries were opened) and 2002 surveys; Bottom graph: Central Coast during 1978-80 (combined surveys), 1989 (last survey year when fisheries were opened), and 2001 surveys.

32 043

2.5 QCI Mature CC Mature QCI Large 2.0 CC Large ) 2

1.5

Proposed target for delisting: 1 spawner/m2 1.0 Abalone Density (#/m Density Abalone Proposed target for conditional FN harvest: 0.5 spawner/m2 0.5

RS: 0.1 large/m2 0.0 1977 1980 1983 1986 1989 1992 1995 1998 2001 Year

Figure 5. Mature (≥70mm SL) (dashed lines) and large (≥100mm SL) (solid lines) abalone densities from index sites surveys in QCI (triangles) and CC (squares). The lines across indicate the proposed recovery targets (solid lines) and the existing short-term recovery objective from the abalone recovery strategy (dashed line) (Abalone Recovery Team 2002).

33 044

Appendix 1 Abalone population model

Abalone population model

Zane Zhang

34 045

1. Introduction. All fisheries on northern abalone (Haliotis kamtschatkana) stocks were closed in 1990 in British Columbia due to greatly reduced abundances of abalone populations. Sizes of abalone populations have remained low or even decreased since the closure. The objectives of this paper are to analyze time series survey data on northern abalone populations in British Columbia using computer simulation models to estimate mortality and stock recruitment relationships as well as predict population trends under different mortality values. The results will provide guidance for the recovery targets of abalone populations.

2. Material and Methods.

2.1. Geographic Locations This modelling study was conducted for abalone populations in two geographic areas: Queen Charlotte Island (QCI) and Central Coast (CC) of British Columbia. QCI is divided into seven sub areas (Atkins et al. 2004), and CC is divided into nine sub areas (Lessard et al. 2007) (Table 1). The model is derived from surveys conducted during 1978-2002, usually with intervals of 4 years between consecutive surveys (Table 1).

2.2. Size-specific Proportion of Cryptic Abalone Juvenile northern abalone (10-70 mm shell length (L)) are typically cryptic and are often found under rocks or in crevices, whereas the majority of adults (>70 mm L) are found on exposed rock surfaces (Sloan and Breen 1988, Cripps and Campbell 1998). To model abalone population dynamics, we need to know the proportions of cryptic abalone at different shell lengths. In the 1984 (Boutillier et al. 1985), 1987 (Carolsfeld et al. 1988) and 1990 (Thomas et al. 1990) surveys in QCI, extra efforts were made to search for and record (other surveys had searched for but had not recorded) cryptic abalone. We used the survey data from these three surveys to study the relationship between the proportion of cryptic abalone and shell length. We used a generalized linear model to describe the relationship, assuming a binomial distribution error structure with the logit link function (McCullagh and Nelder 1989). Abalone shell lengths (L) in mm were log-transformed before fitting the model:

y α β L)log( ++= ε (1) where α and β are the model parameters, ε is the error variability from a binomial

⎛⎞PL distribution, and y = log ⎜⎟ where PL is the probability of being cryptic at shell length L . ⎝⎠1− PL

The parameters α and β were estimated to be 13.85 (se = 2.18) and -3.69 (se = 0.53) respectively. The probability of being cryptic at shell length L can, therefore, be estimated as:

35 046

exp() 13.85− 3.69log(L ) P = (2) L 1+− exp() 13.85 3.69log(L )

2.3. Abalone Growth Model A number of von Bertalanffy growth equations were established for the abalone populations in QCI (Breen 1986). We chose to use the mean values for the parameters k (= 0.265) and

L∞ (112.6) in this study. The growth model was used to determine the size of age 3 and 4 juvenile as well as the size of adult abalone in a previous or subsequent survey to estimate mortality. As the sample size from each individual sub area is too small for the analyses, we pooled together the data from the surveys conducted in QCI or CC in years shown in bold in Table 1.

2.4. Abalone Density Estimation We estimated mean abalone density and standard error through bootstrapping for each of the 7 sub areas in QCI and each of the 9 sub areas in CC. One thousand bootstraps were conducted. In each bootstrapping, the survey sites were randomly selected with replacement, and quadrats in each of the chosen sites were then randomly selected with replacement. Abalone density was estimated simply by dividing the total number of abalone in the selected quadrats by the total number of selected quadrats. The mean density and standard error were calculated from the 1000 estimated densities. Density of abalone for any particular size group was calculated as the product of the density of all sized abalone and the proportion of abalone in this size group in the sample.

2.5. Estimation of Mortality Rate Using the survey data and the growth model, we estimated the mortality rates, Z , in two ways. In the first method, forward-calculation (section 2.5.1), we used the density estimate of cryptic and exposed abalone of all sizes in one survey to calculate the density of cryptic and exposed abalone larger than a selected size (70 or 80 mm L) at the time the subsequent survey was conducted. In the second method, back-calculation (section 2.5.2), we used the density estimate of cryptic and exposed abalone larger than a selected size range (50-70 mm L or 60-80 mm L) in one survey to back-calculate the density of cryptic and exposed abalone in the selected size range at the time the previous survey was conducted. We estimated Z by comparing the calculated density with the survey-derived density. Surveys conducted before 1990 were not used, because the fishery was not closed until 1990.

2.5.1. Forward-calculation Method Abalone length at the subsequent survey time was estimated by rearranging the von-Bertalanffy equation:

LL21=−∞∞exp( −× ktLL )( − ) (3)

36 047

where t is the number of years between this survey and subsequent one, and the subscripts 1 and 2 indicate this survey and subsequent survey respectively. Density of both cryptic and exposed abalone >70 or 80mm L was estimated as:

∑ 1L − PN L )1( ˆ L 12 ×−×= tZDN )exp( (4) TN 1 where D is the estimated density of all sized exposed abalone in this survey except for year

1990 for QCI, Z is the mortality rate to be estimated (see below), NL is the number of abalone at shell length L , which is expected to be >70 or >80 mm L at the subsequent survey time, PL is the probability of being cryptic at shell length L , TN is the total number of measured abalone in this survey, and t is the number of years between this survey and subsequent one. For the 1990 survey in QCI, both cryptic and exposed abalone were observed, Nˆ was estimated without using the term − PL )1( . The survey-derived density of both cryptic and exposed abalone was estimated to be:

∑ 2L − PN L )1( L DN 22 ×= (5) TN 2

where D is the estimated density of all sized exposed abalone in the subsequent survey, and N L is the number of measured abalone at shell length L larger than the preset size at the subsequent survey. A large proportion of abalone which are larger than 70 or 80 mm L result from the growth of smaller abalone observable in the previous survey. A small proportion results from recruitment, i.e. abalone that settled in the previous survey year and have grown to >70 or >80 mm L. The number of recruits among the abalone >70 or >80mm L was estimated to be:

R ×= PRN (6) where R is the number of recruits randomly generated using the stock-recruitment model described in section 2.6, P is the proportion of recruits >70 or >80mm L. The value for P was calculated by assuming that shell lengths of the recruits are normally distributed with the mean estimated from the growth model and a coefficient of variation of 0.08. Various Z values were trialed, and Z was estimated by minimizing the summed squared ˆ difference between NN− R and N . To assess the variation in the Z estimation, 999 Monte Carlo simulations were conducted. In each simulation, survey-derived densities were randomly regenerated using the estimated mean density and standard error.

37 048

2.5.2. Back-calculation Method. To back-calculate shell length, the von-Bertalanffy equation was re-arranged:

LL12=−∞∞exp( ktLL × )( − ) (7) where t is the number of years between this survey and previous one, and L is the shell length. The subscripts 1 and 2 indicate the previous survey and this survey respectively. The size range was selected to be either 50-70 mm L or 60-80 mm L. Density of both cryptic and exposed abalone within the selected size range at the previous survey time was estimated as:

∑ 2L − PN L )1( ˆ L 21 ××= tZDN )exp( (8) TN 2 where D is the estimated density of all sized exposed abalone in this survey, Z is the mortality rate to be estimated (see below), N L is the number of measured abalone at shell length L at this survey, which is estimated to have been within the selected size range at the previous survey, PL is the probability of being cryptic at shell length L , TN is the total number of measured abalone in this survey, and t is the number of years between the previous and this survey. The survey- derived density of both cryptic and exposed abalone was estimated to be:

∑ 1L − PN L )1( L DN 11 ×= (9) TN 1

where N L is the number of measured abalone at shell length L within the preset size range at the previous survey, and D is the estimated density of all sized exposed abalone at the previous survey except for year 1990 in QCI. For the 1990 survey in QCI, both cryptic and exposed abalone were observed, N was calculated without using the term − PL )1( . Various Z values were trialed, and Z was estimated by minimizing the summed squared difference between N and Nˆ . To assess the variation in the Z estimation, 999 Monte Carlo simulations were conducted in the same way as described above.

2.6. Stock and Recruitment Abalone has a short planktonic larval stage which lasts for 5-11 days (Olsen 1984, Pearce et al. 2003), and dispersal of abalone larvae is likely to be limited (Sloan and Breen 1988). We, therefore, estimate the correlation between spawning stock biomass and recruitment for each of the seven sub areas in QCI and for each of the nine sub areas in CC. Size at 100% maturity for northern abalone is approximately 70 mm L (Campbell et al. 1992). Spawning stock biomass was estimated to be the biomass of abalone ≥ 70 mm L/ m2. We used the equation W = 0.0000578L3.2 to convert shell length (L in mm) into body weight (W in g). The two parameter values are the means estimated by Breen and Adkins (1982).

38 049

To correlate spawning stock biomass with the corresponding recruitment, we estimated the spawning biomass of both cryptic and exposed abalone at one survey and density of cryptic and exposed abalone at age 3 or 4 at the subsequent survey depending on the interval between the two consecutive surveys. For instance, to correlate the spawning biomass in the 1987 survey with the recruitment in the 1990 survey, we estimated density of age 3 abalone in 1990; to correlate the spawning biomass in the 1990 survey with the recruitment in the 1994 survey, we estimated density of age 4 abalone in 1994. Abalone cannot be aged easily, therefore the proportion of abalone at the interested age was estimated using the growth model (section 2.3). We combined the survey data from all the surveys in the 7 sub areas of QCI or in the 9 sub areas of CC, because the measured sample size was small for each sub area. We assume that shell length (L) for each age is normally distributed and overlap of the length frequency distribution for one age group over the means for the two neighbouring age groups is negligibly small. For instance, nearly all age 4 abalone are larger than the mean L of age 3 abalone, but smaller than the mean L of age 5 abalone. We also assume that coefficient of variation (cov) is the same for the each L distribution. We set cov to be 0.08, so that the overlap is not too substantial.

We used the growth model to calculate the mean shell length, La−1 , La and La+1 , corresponding to ages a −1, a and a +1. Assuming the numbers of abalone for age a −1, a and a +1 to be

Na−1 , Na and Na+1 respectively, we calculated the expected number of abalone at each shell length between La−1 and La+1 as:

PN ˆ PN ˆ PN ˆ Nˆ = −− ,11 Laa ,Laa ++ ++ ,11 Laa L L L L ` a+1 a+1 a+1 (10) ˆ ˆ ˆ P − ,1 La P ,La ∑∑∑ P + ,1 La =LL a−1 =LL a−1 =LL a−1

ˆ ˆ ˆ where P − ,1 La , P ,La , and P + ,1 La are the expected proportions of abalone at shell length L among ˆ ˆ ˆ abalone at ages a −1, a and a +1 respectively. P − ,1 La , P ,La , and P + ,1 La were calculated based on ˆ the assumed normal distributions. The expected proportion of abalone at shell length L ( PL ) ˆ was estimated by dividing NL by N , the observed number of abalone between La−1 and La+1 . We assumed a multinomial distribution for the number of abalone at each shell length, and the log likelihood is:

La+1 Likelihood = log(PN ˆ ) ∑ L L (11) La−1

39 050

where NL is the number of abalone at shell length L . Various combinations of values for Na−1 ,

Na and Na+1 were trialed until the maximum likelihood was reached. Among the abalone at shell length L the proportion of age a was calculated to be: PN ˆˆ ˆ ,Laa Q ,aL = a+1 ˆˆ (12) ∑ PN αα ,L α a−= 1

ˆ ˆ where N a is the maximum likelihood estimate of Na . We assume that the estimate of P ,La is applicable to the sample for each sub area. Thus, the number of abalone at age a in each sample is:

La+1 ×= QNR ˆ a ∑ ,LaL (13) =LL a−1

where N L is the number of abalone at shell length L ( La−1 ≤ L ≤ La+1 ) in a sample. For CC, only density of cryptic and exposed abalone at age 4 was estimated. For QCI, density of cryptic and exposed abalone at age 4 was estimated for years of 1994, 1998 and 2002, whereas density of cryptic and exposed abalone at age 3 was estimated for years of 1987 and 1990. For consistency, recruitment in this study is represented by the density of cryptic and exposed abalone at age 4. Densities of age 4 abalone in 1988 and 2001 were thus estimated from the densities of age 3 abalone in 1997 and 2000 in QCI using a mortality rate of 0.3. We fit a separate Beverton-Holt stock-recruitment model to the data for QCI and CC with a multiplicative error structure: × Sa R = ε )exp( (14) 1 ×+ Sb where R is the recruitment represented by number of both cryptic and exposed age-4 abalone/m2, S is the spawning biomass represented by matured biomass in kg/m2, a and b are the model parameters, and ε is a random error variability from a normal distribution, N ,0( σ 2 ) . The ratio of a and b determines the maximum recruitment. For CC, a and b were estimated to be 1.73 and 2.8 respectively, suggesting that the maximum recruitment would be 0.62/m2 (Fig. 2). The coefficient of determination, R 2 , was equal to 0.56, and σ was estimated to be 0.61. For QCI, a and b were estimated to be 22.86 and 409.47 respectively, suggesting that the maximum recruitment would only be 0.056/m2 and mean recruitment does not practically change when the spawning biomass is above 0.02 kg/m2 (Fig. 3). This stock-recruitment curve does not appear to be biologically sensible. The carrying capacity for the recruitment is likely to be much higher, and increases in biomass from the current low level are likely to result in higher recruitment. We, therefore, abandoned the mathematically fitted model, and re-fit the model in a

40 051

biological manner. We assumed that the maximum recruitment in QCI is the same as the one estimated for CC. Thus, the ratio of ab/ is fixed to be 0.62, and only one of the two parameters (either a or b ) needs to be estimated. The parameters, a and b , were estimated to be 1.59 and 2.56 respectively, and the resultant stock and recruitment function appears to be much more meaningful, and was used for simulation studies (Fig. 3). The Beverton-Holt stock-recruitment model implicitly assumes that the survival rate for young animals before the recruitment age is made up of both density-independent and density- dependent effects. Existing of the assumed density-dependent effect would result in a higher survival rate for young animals with decreasing density of spawning biomass. It is possible that the abalone population might not have the density-dependent effect at low density levels. We also fit a density-independent stock-recruitment model to the estimated spawning stock biomass and recruitment data:

κ ×= SR ε )exp( (15) where κ is the density-independent parameter. As this stock-recruitment relationship is not realistic for large population size, we only fit the model to the data with the estimated spawning biomass less than 0.2 kg/m2. One data set was excluded for QCI, and two data sets were excluded for CC.

2.7. Comparison of Simulated and Survey-derived Abalone Densities The stock and recruitment functions were established by correlating the estimated spawning biomass with density of recruits in each individual sub area. Simulations, however, were conducted for the entire geographic region. We tested the reliability of our model predictions by comparing the model simulation results with observed survey results. The commercial fishery ended in 1990 due to low abalone population levels. The tests were carried out during the years after the commercial fishery ended. The simulation started in 1990 and ended in 2002 for QCI, and started in 1993 and ended in 2006 for CC. The simulation was conducted using the established Beverton-Holt stock-recruitment model, the growth model, and estimated mortality rates. Maximum age was assumed to be 20 years, and abalone reaching age 21 would be removed from the simulated population. The density estimates of exposed abalone and size frequencies of the 1990 survey (for QCI) and 1993 survey for (CC) were used to set up the initial population structure. The size interval for each age was set according to the growth model. For age a , the size interval was defined to be between the mid-point of La−1 and La and the mid-point of La and La+1 . Abalone within this interval was assigned to be age a . Abalone larger than L∞ was assigned to be age of 20 years.

Spawning biomass was estimated to be the matured biomass (in kg) of abalone ≥ 70 mm L/ m2 , and recruitment (number of age 4 abalone/ m2 ) were randomly generated from the Beverton-Holt stock-recruitment model. One thousand simulations were conducted, and in each simulation variations in the stock-recruitment functions and the mortality rates were incorporated.

41 052

2.8. Trajectory of Population Growth The abalone population dynamics were simulated for the next 50 years starting from the latest survey available (2002 for QCI and 2001 for CC). The simulations were conducted in the similar way as described above, using the same Beverton-Holt stock-recruitment models and growth model. The density estimates of exposed abalone and size frequencies of the 2002 survey (for QCI) and 2001 survey for (CC) were used to set up the initial population structure. Impact of various mortality rates on the population growth was examined.

3. Results When abalone are <20 mm L, the cryptic proportion is very high (Fig. 1). As observed in surveys (Cripps and Campbell 1998), the cryptic proportion declines with increasing shell length (Fig. 1). There are considerable variations in the stock-recruitment relationships, and variation in recruitment increase with increasing spawning biomass (Figs. 2, 3). The maximum mean recruitment in CC was estimated to be 0.62/m2. When this recruitment carrying capacity was used for abalone populations in QCI, the resultant Beverton-Holt stock-recruitment model appears to be reasonable as compared with the one fitted in the mathematical manner (Fig. 3). The fitted density-independent stock-recruitment curves appear to be comparable to the fitted Beverton-Holt curves for both QCI and CC when spawning biomass is low (<0.1 kg/m2) (Figs. 2, 3). Thus, only Beverton-Holt models were used in the simulations. Estimated mortality rates using the backward method are lower than those using the forward method in Table 2. Comparison of observed and simulated exposed densities using the two different morality rates seem to indicate, at least for CC, that the rate estimated by the forward method is closer to reality. With a mortality rate of 0.23 in the QCI simulation, the survey-derived mean densities of exposed abalone in 1998 and 2002 are within the 90% confidence bounds of simulated densities, and the observed mean density of exposed abalone in 1994 is outside the bound (Fig. 4). This simulation suggests that the abalone population is sustainable at the current low level, as the density trend does not appear to be increasing or decreasing considerably. With a mortality rate of 0.29 in the QCI simulation, the survey-derived mean densities of exposed abalone in 1994 and 2002 are within the 90% confidence bounds, and the survey-derived mean density of exposed abalone in 1998 is outside the bound (Fig. 5). This simulation suggests that the abalone population is generally still decreasing which is more likely to be closer to the reality, since large abalone densities have decreased in QCI (Atkins et al. 2004). With a mortality rate of 0.21 in the CC simulation, the survey-derived mean densities of exposed abalone in 1997 and in 2006 are within the 90% confidence bounds, and the survey-derived mean density of exposed abalone in 2001 is outside the bound (Fig. 6). This simulation suggests that the abalone population is in general slightly increasing. With a mortality rate of 0.36 in the CC simulation, the observed mean densities of exposed abalone in 1997, 2001, and 2006 are all within the 90% confidence bounds (Fig. 7). This simulation, therefore, is more reasonable than the simulation with a mortality rate of 0.21, indicating that the abalone population is in general still decreasing.

42 053

With a mortality rate of 0.23, the 50-year simulation suggests that abalone population in QCI would sustain at the current level (Fig. 8). With a mortality rate of 0.29, the 50-year simulation suggests that abalone population in QCI would continue to decrease. The mean total density of exposed abalone would be below 0.1/m2 in 50 years (Fig. 9). With a mortality rate of 0.21, the 50-year simulation suggests that abalone population in CC would be increasing. The mean total density of exposed abalone would be around 0.6/m2 in 50 years (Fig. 10). With a mortality rate of 0.36, the 50-year simulation suggests that abalone population in CC would be decreasing. The mean total density of exposed abalone would be below 0.05/m2 in 50 years (Fig. 11). Mortality rates will have a substantial impact on the growth of the abalone population. Results from simulations varying mortality rates suggest that the abalone populations in QCI and CC would increase with mortality rate below 0.2, would be sustainable with mortality rate between 0.2 and 0.25, and would decrease with mortality rate above 0.25 (Figs. 12, 13). When mortality rates remain high, the exposed abalone population (all sizes) would decline to a density level (0.001/m2) which probably would not be sustainable and lead to eventual extinction. For instance, when annual mean mortality rates are at 0.3 or 0.75, the exposed population would decline to the potential extinction level in about 270 or 30 years, respectively (Fig. 14).

4. Discussion In this study, we used the survey information and published abalone growth models to estimate some crucial abalone population parameters. Based on the estimated size-specific proportion of cryptic abalone, mortality rates, and Beverton-Holt stock-recruitment curves, we examine the trajectories of abalone population growth and investigate impacts of mortality rates on the population growth. Current mortality rates abalone populations endure in QCI and CC are rather high, around 0.3 in QCI and around 0.35 in CC. The mortality rate estimated for abalone populations includes natural mortality rate and poaching-induced mortality rate. In review of abalone biology, Sloan and Breen (1988) reported that natural mortality rate for abalone populations is around 0.15-0.2. If the abalone populations in QCI and CC indeed have a natural mortality rate of 0.15-0.2, an approximate estimate of the poaching rate would be around 0.1. To restore the abalone populations, measures need to be taken to eliminate poaching and to reduce mortality rates down to 0.15-0.2. The proportion of cryptic abalone is high for small abalone, declines quickly for medium sized abalone, and decreases slowly for large abalone. This sigmoid curve is well modelled using the generalized linear model with a binomial distribution error structure. This model enables us to estimate entire spawning stock biomass and recruits. The Beverton-Holt model appears to be a good stock-recruitment model for the abalone populations. The recruitment carrying capacity could be readily derived from the model, and was calculated to be 0.62/m2 for abalone populations in CC. Due to large variations in the stock and recruitment relationship and small range of available spawning stock biomass, the mathematically fitted Beverton-Holt curve for the stock-recruitment relationship in QCI does not appear to be biologically meaningful, as the resultant recruitment carrying capacity is too low. We assumed that the recruitment carrying capacity in QCI is the same as the estimated one in CC. The Beverton-Holt stock-recruitment curve fitted in this manner appears to be much more meaningful in the biological sense, and resembles the Beverton-Holt stock-recruitment curve in CC. The stock-recruitment models form

43 054

the main engines for running population simulations. Comparisons of simulated densities with the observed ones seem to indicate that the established Beverton-Holt stock-recruitment models are reasonable. However, the recruitment carrying capacity in QCI possibly differs appreciably from the estimated one in CC. More stock and recruitment data, especially those containing larger spawning stock biomass, are needed to determine with more certainty the recruitment carrying capacity for abalone populations in QCI. No age information is available for individual abalone. We used the established growth model with mean parameter values published to estimate the abundance of abalone at recruitment age (age 4) and to estimate mortality rates. Abalone in different habitats grow at different rates. Abalone in protected locations with high-quality food, such as Macrocystis and Nereocystis forests, grow faster and to a larger size than those in exposed places with low quality food, such as Pterygophora forests (Sloan and Breen 1988). Ideally, specific growth models should be applied for abalone populations in different habitats. However, the sample size for each sub-area was too small, and data had to be pooled together to apply the mean growth model. The amount of recruitment in each sub-area was then estimated based on the estimated proportion of abalone at recruitment age among the pooled data. Due to likely different growth rates, some biases or errors would inevitably be introduced in the estimations. Therefore, precaution needs to be taken in interpreting the modeling results.

5. References Atkins, M, Lessard, J., and Campbell, A. 2004. Resurvey of northern abalone, Haliotis kamtschatkana, populations in southeast Queen Charlotte Islands, British Columbia, April, 2002. Can. Man. Rep. Fish. Aquat. Sci. 2704. Boutillier, J.A., Carolsfeld, W., Breen, P.A., Farlinger, S. and Bates, K. 1985. Abalone resurvey in the southeast Queen Charlotte Islands, July 1984. Can. Manuscr. Rep. Fish. Aquat. Sci. 1818: 87 p. Breen, P.A. 1986. Management of the British Columbia fishery for northern abalone (Haliotis kamtschatkana). Can. Spec. Publ. Fish. Aquat. Sci. 92: 300-312. Breen, P.A. and Adkins, B.E. 1982. Observation of abalone on the north coast of British Columbia, July 1980. Can. MS Rep. Fish. Aquat. Sci. 1633: 55 p. Campbell, A., Manley, I., and Carolsfeld, W. 1992. Size at maturity and fecundity of the abalone, Haliotis kamtschatkana, in northern British Columbia. Can Manuscr. Rep. Fish. Aquat. Sci. 2169: 47-65. Carolsfeld, W., Farlinger, S., Kingzett, B.C., Sloan, N.A. and Thomas, G. 1988. Abalone resurvey in the southeast Queen Charlotte Islands, June 1987. Can. Manuscr. Rep. Fish. Aquat. Sci. 1966: 90 p. Cripps, K., and Campbell, A. 1998. Survey of abalone populations at Dalain Point and Higgings Pass, Central Coast of British Columbia, 1995-96. Can. MS Rep. Fish. Aquat. Sci. 2445: 31 p. Lessard, J., Atkins, M. J. and Campbell, A. 2007. Resurvey of northern abalone, Haliotis kamtschatkana, populations along the central coast of British Columbia, April 2001. Can. Manuscr. Rep. Fish. Aquat. Sci. 2791: 35 p. McCullagh, P. and Nelder, J.A. 1989. Generalized linear models, 2nd edn. Chapman and Hall. Olsen, S. 1984. Shellfish enhancement report. NOAA, NMFS, Final report for project 1-144-R. Wash. State Dep. Fish., Seattle, Wash. 85 p.

44 055

Snedecor, G.W. and W.G. Cochran. 1989. Statistical Methods, Eighth Edition. Iowa State University Press. Pearce, C.M., Ågerup, P., Alabi, A., Renfrew, D., Rosser, J., Whyte, G., and Yuan, F. 2003. Recent progress in hatchery production of pinto abalone, Haliotis kamtschatkana, in British Columbia, Canada. Can. Tech. Rep. Fish. Aquat. Sci. 2482: 29-44. Sloan, N.A., and Breen, P.A. 1988. Northern abalone, Haliotis kamtschatkana, in British Columbia: fisheries and synopsis of life history information. Can. Spec. Publ. Fish. Aquat. Sci. 103. 46 p. Thomas, G., Farlinger, S., and Carolsfeld, W. 1990. Abalone resurvey in the southeast Queen Charlotte Islands in 1990. Can. Manuscr. Rep. Fish. Aquat. Sci. 2099: 66-82.

45 056

Table 1. Number of sub areas and survey years in Queen Charlotte Island and Central Coast.

Queen Charlotte Island Seven Sub Areas: Carpenter Bay Cumshewa Inlet Kunghit Island Juan Perez Sound Selwyn Inlet Skincuttle Inlet Tanu Island

Survey Years: 1978 1979 1984 1987 1990 1994 1998 2002

Central Coast Nine Sub Areas: Lotbiniere Bay North Aristazabal North Banks Island Oswald Bay Pemberton Bay South Aristazabal Simonds Spider Island Striker Island

Survey Years: 1978 1979 1980 1983 1985 1989 1993 1997 2001

** Survey years in bold were used in the stock-recruitment studies.

Table 2. Estimated mortality rates and standard errors using two calculation methods for abalone populations in Queen Charlotte Island and Central Coast.

Method Queen Charlotte Island Central Coast

Forward 0.23 (0.042) 0.21 (0.064)

Backward 0.29 (0.046) 0.36 (0.065)

46 057

1.2

0.8

0.6

0.4 Cryptic Proportion Cryptic

0.2

0 0 20406080100120140 Shell Length (mm)

Figure 1. Proportion of cryptic abalone in Queen Charlotte Island at shell length.

47 058

0.3

)

2 0.25

0.2

0.15

0.1

Density of Recruits (per m (per Recruits Density of 0.05

0 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22

Spawning Biomass of Cryptic and Exposed Abalone (kg/m2)

0.7

) 0.6

0.5

0.4

0.3

0.2

m (per Recruits Density of 0.1

0 0.2 0.4 0.6 0.8

Spawning Biomass of Cryptic and Exposed Abalone (kg/m2)

Figure 2. Stock recruitment models for abalone in Central Coast. The solid line is the Beverton- Holt curve, and the broken line is the density-independent curve. (Recruitment is represented by the density of both cryptic and exposed abalone at age 4).

48 059

0.3

) 0.25 2

0.2

Biological Fitting 0.15

0.1 Mathematical Fitting

Density of Recruits (per m (per Recruits of Density 0.05

0 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 Spawning Biomass of Cryptic and Exposed Abalone (kg/m2)

Figure 3. Stock recruitment model for abalone in Queen Charlotte Island. The solid lines are the Beverton-Holt curves, and the broken line is the density-independent curve. (Recruitment is represented by the density of both cryptic and exposed abalone at age 4).

49 060

0.9 ) 2 0.8 A 0.7

0.6

0.5 0.4 0.3 0.2 0.1 Exposed Abalone Density (per m AbaloneExposed (per Density 0 1990 1992 1994 1996 1998 2000 2002 Year

0.7 )

2 0.6 B

0.5

0.4

0.3

0.2

0.1

m AbaloneExposed (per Density 0 1990 1992 1994 1996 1998 2000 2002 Year

Figure 4. Comparison of simulated and observed densities of exposed abalone in Queen Charlotte Island. The simulation starts from 1990 with a mortality rate of 0.23. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals. ■ – Observed mean density of exposed abalone in the surveys.

50 061

0.6 )

0.5 A

0.4

0.3

0.2

0.1

Exposed Abalone Density (per m AbaloneExposed (per Density 0 1990 1992 1994 1996 1998 2000 2002 Year

0.6

) 2 B 0.5

0.4

0.3

0.2

0.1

m AbaloneExposed (per Density 0 1990 1992 1994 1996 1998 2000 2002 Year

Figure 5. Comparison of simulated and observed densities of exposed abalone in Queen Charlotte Island. The simulation starts from 1990 with a mortality rate of 0.29. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals. ■ – Observed mean density of exposed abalone in the surveys.

51 062

0.9 )

2 0.8 A 0.7 0.6 0.5 0.4 0.3

0.2

0.1 Exposed Abalone Density (per m AbaloneExposed (per Density 0 1993 1995 1997 1999 2001 2003 2005 2007 Year

1.2

) 2

1 B

0.8

0.6

0.4

0.2

m AbaloneExposed (per Density 0 1993 1995 1997 1999 2001 2003 2005 2007 Year

Figure 6. Comparison of simulated and observed densities of exposed abalone in Central Coast. The simulation starts from 1993 with a mortality rate of 0.21. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals. ■ – Observed mean density of exposed abalone in the surveys.

52 063

0.7 )

2 0.6 A

0.5

0.4

0.3

0.2

0.1 Exposed Abalone m Density (per 0 1993 1995 1997 1999 2001 2003 2005 2007 Year

0.7

) 2 0.6 B

0.5

0.4

0.3

0.2

0.1

Exposed Abalone m Density (per 0 1993 1995 1997 1999 2001 2003 2005 2007 Year

Figure 7. Comparison of simulated and observed densities of exposed abalone in Central Coast. The simulation starts from 1993 with a mortality rate of 0.36. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals. ■ – Observed mean density of exposed abalone in the surveys.

53 064

0.9 )

2 0.8 A 0.7

0.6 0.5 0.4 0.3

0.2

0.1 Exposed Abalone Density (per m AbaloneExposed (per Density 0 0 1020304050 Number of Years since 2002

0.8

) 2 0.7 B

0.6

0.5

0.4

0.3

0.2

0.1

m AbaloneExposed (per Density 0 0 1020304050 Number of Years since 2002

Figure 8. Simulated densities of exposed abalone in Queen Charlotte Island. The simulation starts from 2002 with a mortality rate of 0.23. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals.

54 065

0.5 )

2 0.45 A 0.4

0.35

0.3

0.25

0.2

0.15

0.1 0.05 Exposed Abalone Density (per m AbaloneExposed (per Density 0 0 1020304050 Number of Years since 2002

0.5

) 2

B 0.4

0.3

0.2

0.1

m AbaloneExposed (per Density 0

0 1020304050 Number of Years since 2002

Figure 9. Simulated densities of exposed abalone in Queen Charlotte Island. The simulation starts from 2002 with a mortality rate of 0.29. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals.

55 066

1.2 )

1 A

0.8

0.6

0.4

0.2

Exposed Abalone Density (per m AbaloneExposed (per Density 0 0 1020304050 Number of Years since 2001

1.2

) 2

1 B

0.8

0.6

0.4

0.2

m AbaloneExposed (per Density 0

0 1020304050 Number of Years since 2001

Figure 10. Simulated densities of exposed abalone in Central Coast. The simulation starts from 2001 with a mortality rate of 0.21. A – Two randomly chosen simulated trajectories. B – Mean and 95% confidence intervals.

56 067

0.5 )

2 A 0.4

0.3

0.2

0.1

Exposed Abalone Density (per m AbaloneExposed (per Density 0 0 1020304050 Number of Years since 2001

0.5

) 2

B 0.4

0.3

0.2

0.1

m AbaloneExposed (per Density 0

0 1020304050 Number of Years since 2001

Figure 11. Simulated densities of exposed abalone in Central Coast. The simulation starts from 2001 with a mortality rate of 0.36. A – Two randomly chosen simulated trajectories. B – Mean and 90% confidence intervals.

57 068

9 Year 10 8 Year 20 7 Year 30 Year 40 6 Year 50 5 2002 4

1 Times of Exposed AbaloneTimes of Density in 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Mortality Rate

Figure 12. Expected changes in the density of exposed abalone relative to the observed mean density in 2002 in Queen Charlotte Island with different annual mortality rates and number of years after 2002. The dotted line indicates the level at which there is no change in population density relative to the current density.

58 069

12 Year 10 Year 20 10 Year 30 Year 40 8 Year 50

2001 6

2 Times of Exposed AbaloneTimes of Density in 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Mortality Rate

Figure 13. Expected changes in the density of exposed abalone relative to the observed mean density in 2001 in Central Coast with different annual mortality rates and number of years after 2001. The dotted line indicates the level at which there is no change in population density relative to the current density.

59 070

500

450

400 Central Coast

350 Queen Charlotte Island 300

250

200

Number of Years of Number 150

100

0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 Mortality Rate

Figure 14. Expected number of years for simulated exposed abalone populations (all sizes combined) to reach 0.001/m2 (or to approach potentially extinction) under different annual mortality rates in southeast Queen Charlotte Islands and Central Coast.

60 071

Appendix 2 Impact assessment protocol for works and developments potentially affecting abalone and their habitat

Impact assessment protocol for works and developments potentially affecting abalone and their habitat

Joanne Lessard Alan Campbell

61 072

Table of Content

1. Definitions...... 63 2. Background...... 63 3. How to determine impact on abalone ...... 65 4. Phase 1: Initial Survey ...... 66 Site definition...... 66 Nearshore-swims...... 66 Desired results...... 67 Data Management ...... 67 Decision rule for next step ...... 67 5. Abalone Habitat ...... 67 Physical Factors ...... 67 Biological Factors ...... 67 6. Phase 2: Transect survey...... 68 Transect placement ...... 68 Transect layout...... 68 Underwater survey...... 68 Analytical methods ...... 69 Data Management ...... 70 Decision rule for next phase ...... 70 7. Control site...... 70 8. Phase 3: Monitoring program – Plot survey ...... 70 Reference line placement...... 71 Underwater survey...... 71 Analytical methods ...... 72 Data Management ...... 72 Decision rule for next phase ...... 72 9. Phase 4: Feed back...... 72 10. References...... 73 Appendix A. Third party biologist requirements...... 74 Appendix B. Field data sheet for the transect survey ...... 75 Appendix C. Field data sheet for the plot survey...... 76 Appendix D. Dive Codes ...... 77 Appendix E. Database Field Descriptions ...... 78

62 073

1. Definitions 3rd PARTY BIOLOGIST: an established independent third party biological consultant company or an independent third party biologist accredited with a university or college degree in a related biological science that has preferably formed an independent company under his/her own name with experience working with DFO in accomplishing biological research including surveys. Other requirements are outlined in Appendix A. ABALONE HABITAT: description of physical and biological features of habitats where abalone are found; includes all abalone habitats as well as critical (not defined for abalone). See Section 5. CONTROL SITE: location outside of the area of influence and within 1000m of the potentially impacted site to minimize differences in current and temperature regimes CRITICAL HABITAT: the habitat that is necessary for the survival or recovery of a listed wildlife species that is identified as the species’ critical habitat in the recovery strategy or in an action plan for the species (as defined under SARA) IMPACT: unless other wise stated (e.g. impact on habitat) in this document, for the sake of brevity, impact refers to the direct or indirect impacts of works and developments on abalone abundance and distribution only. INITIAL SURVEY: See Section 4. MONITORING PROGRAM: the plot survey repeated at least once a year. PLOT SURVEY: See Section 8. PRECAUTIONARY APPROACH: Set of measures taken to implement the Precautionary principle. A set of agreed cost-effective measures and actions, including future courses of action, which ensures prudent foresight, reduces or avoids risk to the resource, the environment, and the people, to the extent possible, taking explicitly into account existing uncertainties and the potential consequences of being wrong. (Garcia S.M. (1996) The precautionary approach to fisheries and its implications for fishery research, technology and management: An updated review. FAO Fish. Tech. Paper, 350.2: 1-76) RECRUITMENT: for this document, juvenile abalone with a shell length <70 mm. SARA: Species at Risk Act SITE: proposed site, unless otherwise stated (e.g. control site). SL: Shell length, the maximum measurement of an abalone shell. TRANSECT SURVEY: See Section 6.

2. Background The provisions of SARA that were implemented on June 1, 2004 include: • prohibitions on killing, harming, harassing, possessing, buying or selling an individual of a species listed as an extirpated species, an endangered species or a threatened species etc. (section 32).

63 074

• prohibitions on damaging or destroying residences of individuals (section 33). • prohibitions on destroying the critical habitat of a listed endangered or threatened species or listed extirpated species (section 58). • provisions for effective enforcement measures and significant penalties where needed to serve as a deterrent. These prohibitions will apply to aquatic species that are listed under SARA as extirpated, endangered or threatened. There is a provision in SARA (section 73) that allows the competent minister (DFO for listed aquatic species) to authorise a person to engage in an activity that affects a listed wildlife species, its critical habitat or residence. However, this provision also includes a series of strict criteria that must be met prior to doing so. Currently there is ample habitat available in BC for the northern abalone population. In general, abalone populations have declined; however, there has been no known significant reduction in available habitat. Therefore, habitat loss is not a major concern in the recovery of northern abalone at this time in comparison with the identified threats. Although good abalone habitat is not believed to be limiting, there may be certain habitat where juvenile survival is better, or where the reproducing adults contribute to a larger portion of the total recruitment. Identification of this key habitat is being included as part of the abalone research and rebuilding plans. The abalone recovery strategy identified several knowledge gaps (Abalone Recovery Team 2002). The recovery strategy identified the need to clarify the extent of threat of works and developments on, in and under the water to northern abalone populations and habitat. The recovery strategy also identified the need for monitoring and regulation of projects to prevent losses to important spawning aggregations and maintain ecosystems in which abalone can recover. Once ‘critical habitat’ for northern abalone is defined (e.g., abalone beds or important spawning aggregations), specific criteria to protect it under the Fisheries Act and Regulations (1993) and SARA (2003) may be better developed and applied. Until then, it is recommended that the best science available be used, and where science is lacking that a precautionary approach be adopted in considering and approving location(s) for works and developments on, in, and under the water. Determining impacts on a site by site basis is impractical and will not provide meaningful data as other factors may affect abalone populations. However, when all sites with abalone are combined it may be possible to determine the impact(s) if a proper scientific method is followed. It is the intention of the monitoring program described in Section 8 to evaluate what these impacts are. Given DFO abalone stock assessment limited budget, no large scale studies are planned to determine the impacts of works and developments on, in or under the water on abalone populations. Therefore, the proponents will have to either pay for a DFO certified 3rd party biologist and/or give money to DFO to carry out the work, which will include field survey, analysis and reporting. This impact assessment protocol applies to any proposed works and developments where abalone habitat is present and the area affected will be larger than 20m2. The amount of abalone habitat necessary to trigger this protocol is arbitrary and is purposely small as the shape of the area impacted is important. For example, 20m2 distributed as a 1m vertical swath to a depth of 10m with a moderate slope is not equal to 20m wide 1m horizontal swath parallel to shore at 3m depth because abalone prefer shallow depths and more abalone preferred habitat is impacted in the second scenario. If abalone are present in a small area (i.e. <20m2) expected to be impacted, the

64 075

individual abalone shall be relocated, under a SARA permit, to suitable abalone habitat nearby. The SARA permit will include the following conditions (some conditions may vary depending on the available abalone habitat nearby): 1. The dive surveys of the area to be impacted must take place at night and search successive depth contours in a systematic and thorough manner. 2. All abalone observed within the survey area must be removed from the substrate by hand only, with the assistance of a Pycnopodia helianthoides (sunflower sea star) if necessary; prying abalone from the substrate is not permitted; 3. The shell length (to the nearest mm), depth located, substrate type and dominant algae species must be recorded for each abalone observed; 4. The abalone must be relocated underwater to a location of cover in rocky subtidal habitat no deeper than 6m depth (chart datum) and a minimum of 50 metres away from the construction footprint; taking abalone from the water is not permitted; 5. Plastic totes may be used to move the abalone underwater and abalone may be relocated in close proximity of another abalone to improve chance of spawning success; 6. As a requirement to report information under this project, the authorized persons must submit a written record containing the following summary information: i. Dates in which relocation surveys took place; ii. Number of abalone observed and relocated; iii. Shell length, depth located, substrate type and dominant algae cover for each abalone observed; iv. Overall perspective on the success or difficulties in conducting the work.

3. How to determine impact on abalone Except for surveyed sites, there is a general lack of data on abalone distribution and abundance throughout the BC coastline. Site specific information for proposed works and developments must be acquired before any decision can be made. In order to determine impacts of works and developments on abalone populations and make inference to their habitat, abalone will have to be present at some sites. Only abundance, and possibly distribution, data will be used to determine impacts in the short term (2-5 years) as other parameters of abalone population health are more difficult to measure (e.g. change in reproductive output, growth, disease incidence, etc.). Impacts may be determined by changes in density before and after the project is completed in conjunction with the continuous monitoring of control site(s) outside of the area of influence. For example, there may be a statistically significant changes (increases or decreases) in total abalone density within the site, but no change at the control site(s) or the total density does not change, but one of size category (juvenile, mature, etc.) becomes more dominant when compared with the control site(s). Observing changes in abalone spatial distribution will be more difficult unless some animals are uniquely identified (tagged), particularly if density decreases and few or no shells are recovered. Nevertheless, changes in depth distribution and aggregation will be possible under the proposed monitoring approach described in Section 8. To obtain information necessary to make a decision on the site and evaluate impacts if approved, we recommend a four phased approach: • Phase 1: Initial Survey

65 076

The site is assessed to determine the extent of abalone habitat present using nearshore swims. The abalone habitat is then mapped. • Phase 2: Transect Survey A quantitative survey is conducted to estimate abalone densities within the abalone habitat identified in Phase 1 as well as in an area outside the area of influence. • Phase 3: Monitoring program – Plot survey If the site is accepted and abalone are present, an intensive survey is conducted at 1-3 plots within the impacted site as well as within a control site, outside the area of influence. • Phase 4: Feed Back After 5 years, an analysis of abalone abundance and distribution data combining several sites of a given type of work or development should be completed to evaluate the impacts and determine if mitigation actions are required. Each phase is described in detail in the sections below. We recommend that Phase 1 to 3, if not done by DFO staff, be completed by a certified 3rd party biologist (see Appendix A for requirements).

4. Phase 1: Initial Survey The objectives of this phase are to (1) establish the area of Abalone Habitat present at the site, and (2) delineate these habitats on a chart. Although all habitats are important, for the purpose of this document, only abalone habitat is described in Section 5. Site definition The site is defined by using landmarks and geographic coordinates. The ‘site width’ is the linear distance between the two furthest points. Nearshore-swims Two divers swim (a few metres apart from each other) in a zigzag pattern (generally parallel to shore) between depths of 0-10 m chart datum. Very good notes need to be taken throughout the swim so that the GPS coordinates can be related to what was observed underwater. Habitat changes including changes in primary substrate (e.g., bedrock to boulders or sand), and algal community (e.g., from a Macrocystis to a Nereocystis kelp forest or understorey algae only), should be marked using one of two methods described below. Method 1: Floats can be deployed at the edges of each change in habitat. The boat can then use a GPS to obtain the coordinates. Because the edges of habitats do not usually form a straight line, several floats need to be release to accurately map the habitats. Method 2: One person is put on shore at a location where most of the surface water of the site would be visible and records his/her position using a portable GPS. Two divers swim throughout the site during several dives carrying a metal float. At a change in habitat, one of the divers pulls on the float several times while the other diver records the time, depth and other habitat information. Upon seeing the float bob at the surface, the shore person measures the distance to

66 077

the float using a laser range finder and the magnetic bearing of the float using a compass and records the time which will be matched with the time recorded underwater. Desired results The end product of this phase should be a digital map with depth contours and the important habitats delineated. Although all habitats should be outlined, for the purpose of this document, only abalone habitat is described here in detail (see next section). Data Management The GPS shore positions are imported into ArcView 3.2 or another GIS software. Method 1: The GPS positions from the boat are matched with the divers notes to digitize (create a polygon) abalone habitat. Method 2: From the shore positions, the measured distances and bearings are plotted using an extension from Jenness Enterprises called “Distance & Azimuth tool” (http://www.jennessent.com/arcview/arcview_extensions.htm). Polygons delineating abalone habitat are created using the plotted positions. The digital map, electronic file containing the GPS points and copies of the field notes must be sent to the Shellfish Data Unit, PBS, Nanaimo. Decision rule for next step If abalone habitat, as described in the next section, is present and the area of the abalone habitat is > 20m2, then the next phase is necessary to assess the abalone density at the site as well as in surrounding areas.

5. Abalone Habitat Physical factors include: i. Primary Substrate: bedrock and/or boulders ii. normal salinity (not low salinity as found close to river run off) iii. Depth: <10m depth (datum) iv. Good water exchange (tidal current or wave action present) v. Secondary Substrate: some cobble may be present and little or no gravel, sediment, sand, mud, or shell present. Biological factors include: i. Presence of encrusting coralline algae (e.g. Lithothamnium) ii. Presence of sea urchins Strongylocentrotus franciscanus and/or S. droebachiensis, Lithopoma (Astraea) gibberosa, sea stars. iii. Presence of kelp in surrounding area (e.g., Nereocystis, Macrocystis, Pterygophora). iv. Presence/absence of abalone

Physical and biological factors are listed in order of importance.

67 078

6. Phase 2: Transect survey The objective of the transect survey is to get quantitative estimates of abalone density and distribution within the abalone habitat delineated in the initial survey (Phase 1). This is necessary to evaluate if the work or development proposal will be accepted based on the 0.1 abalone/m2 criteria (see “Decision Rules” this section). The method described in this section is identical to Lessard et al. (2002) with two exceptions: (1) the higher confidence interval is used for the density calculation, and (2) the population size is not calculated as it is unnecessary to evaluate the site. The 0.1 abalone/m2 threshold was originally based on the measurable short- term goal of the National Recovery Strategy (see “Background” section). Although, this threshold in the recovery strategy is for the size category ≥100 mm SL, the higher confidence interval of the mean is used here. Transect survey(s) outside the area of influence is also necessary to assess possible control site(s). The transect survey at the control site(s) may be done after the transect survey at the proposed site is completed and the site is given approval to go ahead. However, to minimize seasonality effects, transect survey(s) at possible control sites should be conducted within a month, two at the most. For information on where the control site should be, see Section 7. Control Site. Transect placement Transect positions are marked on nautical charts before the survey begins. The positions are selected randomly using the ‘abalone habitat width’ defined as the linear distance between the two furthest points of the abalone habitat. Transects are perpendicular to the shoreline at these positions. If the abalone habitat is discontinuous, separated by large areas of unsuitable abalone habitat (e.g., area of sand), the process to select the transect positions is repeated for each area of abalone habitat. At least ten transects should be surveyed in each abalone habitat area. If the width of the abalone habitat is shorter than 300m, a lesser amount of transects may be considered. Transect layout The primary sampling unit is a transect, made up of a variable number of secondary units: quadrats. Each transect is one meter wide and variable in length, depending on the slope of the substrate. Prior to entering the water, a lead line, the transect, is laid perpendicular to the shore, from the boat. If this is not possible, because of thick kelp beds or other environmental factors, then the divers should sample along a compass bearing perpendicular to the shore. The compass bearing must be strictly followed to avoid possible bias in the density estimate(s). Transects begin at 10 m chart datum and extend all the way into the shore, or to the point where the surge makes it impossible for the divers to work effectively. Underwater survey (Filling out the “Abalone Field Sheet - Transect” Appendix B) The secondary sampling unit consists of a 1 m x 1 m square quadrat that is placed beside the transect, 1 m away to avoid the area potentially disturbed by the lead line placement. Divers flip the quadrat parallel to the transect line, from deep to shallow. One diver records the data while the other measure the abalone and flips the quadrat. In each quadrat, the recording diver writes down 1) the shell length (SL in mm) of each abalone, 2) the depth, 3) the time, 4) the substrate type, 5) the number of urchins, 6) the number and relative size of abalone predators (sunflower starfish, Dungeness and red rock crabs, octopus, etc.) and 7) the % cover and dominant species

68 079

of algae. The % cover of all algae combined is recorded by category: 1) canopy (kelp taller than 2m), 2) understorey (algae between 15cm and 2m in height), 3) turf (erect algae less that 15cm in height) and 4) encrusting (carpet-like algae). The dominant algal species (1-2) are recorded for the first 3 categories only. Appendix D lists the substrate and algae species codes to be used. The measuring diver must exercise caution when measuring abalone to ensure that the longest shell length is measured and the abalone is returned right side up on the rocks outside and behind of the quadrat. In order to minimize habitat damage, algae are not to be removed. Boulders are not to be moved to search for cryptic abalone. Caution must be exercised to ensure that abalone in upcoming quadrats are not disturbed. Where the transect length is greater than 20 m, only every second quadrat needs to be sampled completely. If transects are longer than 60 m, abalone and depth can be sampled every second quadrat, and substrate and algae cover can be sampled every fourth quadrat. The frequency of sampling must be written on the underwater sheet. Analytical methods Calculations are included here for information only. The analysis will be performed by DFO Stock Assessment. 2 For each site, the estimated mean density, ds (number/m ), of abalone is calculated as:

∑((cqtt /t )* L ) t ds = (1) ∑ Lt t

The standard error of the mean density, ses, is calculated as: 2 ∑ cqts L− d L ((t / )*tt * ) n t ses =−1* (2) T nn*(− 1)* L2 where n is the number of transects, ct is the number of abalone counted in transect t, qt is the number of quadrats sampled in transect t, Lt is the length of transect t, L is the mean transect length, T is the total possible number of transects that can be sampled in the surveyed area and is equal to the ‘abalone habitat width’. This method accounts for the variable length of transects and for the variable proportion of quadrats surveyed along each transect. To estimate the mean density (Equation 1) and standard error (Equation 2) for a specific size group (i) (i.e. ≥100 mm SL), the value ct is substituted with cti, the counts of size group i in transect t. At each site, the higher 90% confidence intervals of the mean density (H90CI), for all sizes or for a particular size group (≥100 mm SL) of abalone, are calculated using bootstrapping (Davidson and Hinkley 1997).

69 080

Data Management All the data must be entered using the “Transect Data Entry” form in the Access database provided by DFO Stock Assessment. The fields that need to be filled on the field sheets and in the database are described in Appendix E. The original field data sheets as well as the electronic version in Access must be sent the Shellfish Data Unit, PBS, Nanaimo. At PBS, a S-Plus script exists to analyze the data using the data directly from the database. Decision rule for next phase If the H90CI for all sizes is ≥0.1 abalone/m2, the site is automatically rejected. If the H90CI for all sizes is <0.1 abalone/m2, the responsibility of the decision to go ahead with permitting rests with the Habitat Management Program. If the site is accepted and an authorization is issued in accordance with section 35 of the Fisheries Act, the next phase is initiated.

7. Control site Proposed control site(s) should be outside of the area of influence and within 1000m of the impacted sites to minimize differences in current and temperature regimes. For aquaculture proposals, the area of influence is determined by DEPOMOD. The control site must be within abalone habitat as described in Section 5. In general, the control site should have the same relative exposure, current regime and habitat characteristics. For example, it would be unsuitable to have smooth bedrock substrate within the abalone portion of the impacted site and boulders at the control site. It may also be unsuitable to have the control ‘around the corner’ where exposure to wave action would be different.

8. Phase 3: Monitoring program – Plot survey The objective of this phase is to survey abalone within a small geographic area in order to calculate reliable density estimates with minimal variation. A density estimate with high precision is essential to detect impacts on abalone abundance as abalone density estimates have inherently high variance due to their aggregating behaviour. It is not rare to have standard deviations equal to or larger than the mean density estimates. For example, if the mean density estimate from the transect survey is 0.05 abalone/m2 with a pooled standard deviation of 0.025 abalone/m2, 34 samples would be necessary to detect a change in abundance of at least 50% with 95% confidence 80% of the time (17 impacted sites and 17 control sites). To increase precision, more plots can be placed in both the impacted and control sites; this would add a strata (high/low density areas) to the sampling design. In addition, more random transects can be added within each depth strata. The number of samples (transects) and strata can be determined using the transect survey results. The plot survey is based on a stratified random sampling design. The current plot survey design is based on past survey results and builds on the Parks Canada and Haida Fisheries Program survey designs. Figure 1 gives a schematic diagram of the plot survey design. A better design would involve using the quadrats as the primary sampling unit and have each quadrat randomly placed within the plot. Strata (e.g. deep/shallow and/or high/low density areas) could also be used. However, the underwater logistics of such a design are impractical. The sampling design described below is for the minimum number of strata and samples required: one plot at each of

70 081

the impacted and control sites with 2 depth strata in each plot and 10 or 8 transects for the shallow or deep reference lines, respectively.

Figure 1. Schematic view of the plot survey design. Numbers on the left side are depths in metres (datum). Reference line placement (in consultation with DFO Stock Assessment) Two reference lines, 40m long each, are placed at 2.5m and at 7.5m below chart datum. The location of the reference lines are the middle of the 2 depth zones (0-5m and 6-10m) that are going to be sampled (the 2 strata in a stratified random sampling design). On each of the reference lines, several short perpendicular 1 m wide transects are surveyed, alternating on either side of the reference lines to minimize disturbance. The start location of each transect is chosen randomly prior to the start of the survey. Choose 10 starting positions (out of 40 m) along the shallow reference line (2.5m) and 8 along the deep reference line (7.5m). Additionally, randomly choose the side of the transect where the first transect is placed and alternate thereafter (marked as “Start Down” or Start Up” on the field sheet). Underwater survey (Filling out the “Abalone Field Sheet - Plot” Appendix C) Each transect starts off the reference line at the randomly chosen location and the quadrat is flipped perpendicular to the reference line until the top, or bottom, of the depth zone is reached. No lead line is laid out for the random transects (a compass bearing can be taken, but this is not necessary as the transects are usually short, 4-8 quadrats long). One diver records the data while the other measures the abalone and flips the quadrat. In each quadrat, the recording diver writes down 1) the shell length (SL in mm) of each abalone, 2) the depth, 3) the time, 4) the substrate type, 5) the number of urchins, 6) the number and relative size of abalone predators (sunflower starfish, Dungeness and red rock crabs, octopus, etc.) and 7) the % cover and dominant species

71 082

of algae. The % cover of all algae combined is recorded by category: 1) canopy (kelp taller than 2m), 2) understorey (algae between 5cm and 2m in height), 3) turf (erect algae less that 5cm in height) and 4) encrusting (carpet-like algae). The dominant algal species (1-2) are recorded for the first 3 categories only. The measuring diver must exercise caution when measuring abalone to ensure that the longest shell length is measured and the abalone is returned right side up on the rocks outside and behind of the quadrat. In order to minimize habitat damage, algae are not to be removed. Boulders are not to be moved to search for cryptic abalone. Caution must be exercised to ensure that abalone in upcoming quadrats are not disturbed. All quadrats are sampled completely. Once the transect is completed, the divers move to the random location and repeat the procedure until all locations have been completed within the depth strata. Analytical methods To calculate the mean and standard error within each strata a, the analysis is identical to the Transect Survey in Section 6. 2 For each site, the estimated mean density, ds (number/ ), of abalone is calculated as:

ds = (1/N)∑nada (3)

The standard error of the site mean density, ses, is calculated as:

ses = (1/N)∑nasea (4) where N is the total number of transects in all strata

na is the number of transects in strata a

da is the estimated mean density in strata a

sea is the estimated standard error of the mean in strata a The data will probably not be normally distributed and a nonparametric test such as the Wilcoxon paired-sample test should be used to look at differences between control and impacted sites. Data Management All the data must be entered using the “Plot Data Entry” form in the Access database provided by DFO Stock Assessment. The fields that need to be filled on the field sheets and in the database are described in Appendix E. The original field data sheets as well as the electronic version in Access must be sent to the Shellfish Data Unit, PBS, Nanaimo. Decision rule for next phase Once the monitoring is initiated at more than one site, the next phase should be instigated after 2- 5 years depending on the extent of the changes. For example, if densities decrease at all impacted sites, but not at the control sites, by >50% within 2 years, then Phase 4 should be initiated.

9. Phase 4: Feed back In phase 4 all monitoring data for a given type of work or development are pooled to determine overall impacts of this given type of work or development on abalone populations. Due to natural variation in abalone density and the low initial densities at approved sites

72 083

(<0.1abalone/m2), a small change in abundance or distribution will be difficult to detect. Detecting changes less than 50% is therefore impractical because of the high variance and a much larger number of samples would be required. Because of the possible implications of such an analysis, the results should be presented at PSARC.

10. References Davidson, A.C., and Hinkley, D.V.. 1997. Bootstrap Methods and their Application. Cambridge University Press, Cambridge. 578 p. Lessard, J, Campbell, A, and Hajas, W. 2002. Survey protocol for the removal of allowable numbers of northern abalone, Haliotis kamtschatkana, for use as broodstock in aquaculture in British Columbia. DFO Can. Sci. Advis. Sci. 2002/126: 41 p. Sloan, N.A., and Breen, P.A.. 1988. Northern abalone, Haliotis kamtschatkana, in British Columbia: fisheries and synopsis of life history information. Can. Spec. Public. Fish. Aquat. Sci. 103: 46 p.

73 084

Appendix A. Third party biologist requirements The minimum requirements for biological expertise for an independent third party biologist to conduct abalone surveys are: • an established independent third party biological consultant company with experience working with DFO in accomplishing biological research including surveys; or • an independent third party biologist accredited with a university or college degree in a related biological science that has preferably formed an independent company under their own name and has experience working with DFO in accomplishing biological research including surveys. And • meets a reference check for experience, competency, and demonstrated independent ‘arms length’ work experience; And • has passed a training session with DFO-Stock Assessment Division on conducting abalone surveys, including data collection and reporting. Training will be given by DFO-Stock Assessment Division and may be expected to include dive surveying. And • SCUBA dive certification, meeting WCB requirements And • bonded (to ensure confidentiality) And • knowledge of common algae, invertebrates and fish species. And • has access to Microsoft Access database software

74 085

Appendix B. Field data sheet for the transect survey See Appendix E for field descriptions. Abalone Field Sheet - Transect Page ____ of _____

Site Name: File number: Date: Measurer: Recorder: Time in: Out: LAT: LONG: Direction (bearing in °): Transect number: Quadrat Frequency:

Quad Depth Time Substrate Abalone Shell Length Urchin Predators Canopy Understory Turf En # ft. (mm) Count %

Substrate codes: 1 bedrock smooth 3 boulders 5 gravel 7 sand 9 mud 2 bedrock crevices 4 cobble 6 pea gravel 8 shell

General Algae codes: Specific algae codes: EN encrusting (flat) F foliose (leaf-like) AG Agarum IR Iridea PT Pterygophora AC articulated coralline B branched (tree-like) AL Alaria LA Laminaria SA Sargassum KK kelp H filamentous (hair-like) CO Costaria MA Macrosystis UL Ulva B other brown CY Cymathere NT Nereocystis R red algae Grasses (GR) DE Desmarestia PL Pleurophycus G green algae PH Phyllospadix EG Egregia PO Porphyra

75 086

Appendix C. Field data sheet for the plot survey note: Numbers/text in bold in the heading section of the field sheet are chosen randomly for each reference line for each period surveyed. See Appendix E for field descriptions. Abalone Field Sheet - Plot Page ____ of _____

Site Name: File number: Date: Measurer: Recorder: Time in: Out: LAT: LONG: Direction (bearing in °): Reference Line: (shallow or deep) Plot number: Transect start locations: 2, 6, 13, 19, 21, 25, 31, 32, 34, 38 Start Down Tide height (height@time):

Quad Depth Time Substrate Abalone Shell Length Urchin Predators Canopy Understory Turf En # ft. (mm) Count %

Substrate codes: 1 bedrock smooth 3 boulders 5 gravel 7 sand 9 mud 2 bedrock crevices 4 cobble 6 pea gravel 8 shell

76 087

Appendix D. Dive Codes Table D1. Substrate codes Code Substrate 1 Bedrock - smooth 2 Bedrock - crevices 3 Boulders (rock bigger than a basketball) 4 Cobble (basketball down to 3 inches) 5 Gravel (3 inches down to 3/4 inch) 6 Pea gravel (3/4 inch down to 1/8 inch) 7 Sand 8 Shell 9 Mud

Table D2. Algae Codes

Code Species Code Species Code Species AA Alaria nana EI Eisenia arborea MI Microcladia sp AB Agarum cribosum EN encrusting algae NO No Algae Present AC Articulated corallines ET Enteromorpha sp NT Nereocystis luetkeana AF Agarum fimbriatum FU Fucus gardneri OD Odonthalia sp AG Agarum sp GA Green Algae PH Phyllospadix sp AL Alaria sp GB green branched PL Pleurophycus gardneri AM Alaria marginata GE Gelidium sp PO Porphyra sp BB brown branched GF green foliose PR Prionitis sp BF brown foliose GG eelgrass & surfgrass PT Pterygophora californica BH brown filamentous GH green filamentous PV Pelvetiopsis sp. CA Callophyllis sp GI Gigartina sp RB red branched CF Codium fragile GR Gracilaria pacifica RF red foliose CN Constantinea sp. GS Gastroclonium RH red filamentous CO Costaria costata subarticulatum SA Sargassum muticum CR Cryptopleura sp HA Halosaccion glandiforme UL Ulva sp, Monostroma sp CS Codium setchellii HE Hedophyllum sessile or Ulvaria sp CY Cymathere triplicata IR Iridea sp UN Unknown DB Dictyota binghamiae KK Kelp ZO Zostera sp DE Desmarestia sp LA Laminaria sp DF Desmarestia foliacea LB Laminaria bongardiana DL Desmarestia ligulata LE Leathesia difformis DU Desmarestia munda LO Lessoniopsis littoralis DR drift algae LR Laurentia spectabilis DS Delesseria sp. LS Laminaria saccharina DV Desmarestia viridis LT Laminaria setchellii EG Egregia menziesii MA Macrocystis integrifolia

77 088

Appendix E. Database Field Descriptions Field Name Description Site Name The name of the proposed tenure as stated on the application File Number File number of the application *if available Date YYMMDD Measurer The name of the diver measuring and counting Recorder The name of the diver recording Time In The time (hh:mm) the diver leaves the surface *note: do not round to 5 mins. Time Out The time (hh:mm) the diver reaches the surface LAT Latitude of site in degrees and decimal minutes LONG Longitude of site in degrees and decimal minutes Direction (bearing in º) The bearing in which the transect is laid, in degrees Reference Line (shallow or deep) Plot Number Number assigned to the plot Transect start locations Randomly selected points along the transect to lay reference lines Start The direction from the main transect line to start the first quadrat, either shallow or deep Tide height (height @time) Several tide height to add to maximum depth to reach for strata (e.g., 4.5ft@10:30, 5ft@11:00, etc) Quad# The number of the quadrat being sampled Depth (ft) The gauge depth, in feet, for the quadrat being sampled Time The time (hh:mm) at which the diver was in that quadrat Substrate up to three codes for the most prominent substrate types in that quadrat (see sheet for codes) Abalone Shell length (mm) The measured shell length in mm of each abalone measured Urchin Count The number of urchins counted in that quadrat Predators (count/size/species) eg. 2MPy = 2 medium Pycnopodia Canopy % and species of the most dominant canopy species (kelp taller than 2m) (e.g., 50 MA = 50% Macrocystis) Understory % and species of the most dominant understorey species (algae between 5cm and 2m in height) Turf % and species of the most dominant turf species (erect algae less that 5cm in height) En% % (only) of cover of encrusting (carpet-like algae)

78 089

Appendix 3 Survey report of experiment to determine short-term impacts of finfish aquaculture on abalone

Survey report of an outplanting experiment using hatchery-reared abalone to determine short-term impacts of finfish aquaculture on abalone in the Broughton Group, July 2005 - March 2006

Sandie Hankewich

79 090

INTRODUCTION Aquaculture, while an ancient practice in other parts of the world (Iwama 1991), is an emerging field in Canada. Canadian aquaculture production increased, on average, 19.8% per year between 1986 and 2001 (Lanteigne 2002), and is expected to continue to rise. The majority of production, 43% of the Canadian total in 2004, occurs in BC (Alain 2005). Finfish, especially salmon, account for 79% of the total production, and 89% of the total value of aquaculture in Canada (Alain 2005). Recent studies have identified several possible detrimental effects of aquaculture to coastal ecosystems including, but not limited to: introduction of chemicals found in the feed and construction materials of the net pens (Winsby et al. 1996), eutrophication and/or possible oxygen deprivation spurred by excess food and feces in the water (Hargrave et al. 1993), release of pesticides and antibiotics used at fish farms (Haya et al. 2001), and burial by excess sedimentation (Ritz et al. 1989). However, the effect of fish farms on surrounding ecosystems is not yet fully understood (Milewski 2001). In particular, there is a serious lack of literature concerning aquaculture effects on hard- bottom substrates and moderate to high current speeds. Consequently, potential impacts on northern abalone (Haliotis kamtschatkana) populations and their habitat are unknown. As a result, the Department of Fisheries & Oceans Canada (DFO) initiated in a study to determine if finfish aquaculture has acute (i.e. immediate) impacts on abalone survival and growth in their natural environment. This nine month study assessed the growth and survival of small hatchery-produced abalone, which were placed within artificial habitats at existing aquaculture tenures as well as control sites in Queen Charlotte Strait (QCS).

METHODS

Experimental Sites In order to test conditions as they exist in nature, this experiment was conducted in situ, rather than in a laboratory situation. The broodstock of the hatchery-produced abalone were originally from Cormorant Island, QCS. Due to possible disease transfer, outplanting occurred within the same biogeographic zone (Abalone Recovery Team 2002). Therefore only finfish aquaculture sites in the QCS were considered for this study. To simplify logistics, tenure sites which were operated by a single company were selected. Furthermore, the tenure sites had to be in operation throughout most of the experiment (July 2005 to March 2006). Given the above considerations, and exploratory SCUBA surveys examining habitat characteristics (substrate, algae cover and invertebrate community), tenure sites located at Swanson and Bonwick Islands (Figure 1) were selected. Control sites selected had similar habitat characteristics to the chosen tenure sites, and were located close to the tenure sites, but outside their area of influence, to ensure similar oceanographic conditions. Two substations were selected at each control and tenure site based on appropriate substrate availability: 6-8 m depth (chart datum), hard substrate with minimum slope surrounded by appropriate abalone habitat. Despite

80 091

the fact that the sites constituted suitable habitat for abalone, no abalone were seen at the sites or the immediate surrounding areas during the preliminary dives.

Condo Description and Deployment Juvenile abalone tend to hide inside crevasses or under large boulders, making these cryptic abalone very difficult for divers to find (Sloan and Breen 1988; J. Lessard, DFO, Nanaimo, pers. comm.). Artificial habitats (further referred to as condos), provide hiding spaces for juvenile abalone as well as standardized sample areas to monitor juvenile recruitment and growth (Davis 1995). Condos also reduce the intensity and effort (in dive time and sample sorting) associated with other methods such as underwater magnifying glass (Shepherd and Turner 1985, McShane and Smith 1988), rock removal combined with anesthesia (Prince and Ford 1985, Sasaki and Shepherd, 2001), and venturi suction sampler (McShane and Smith 1988). Condos are non-destructive to habitat (DeFreitas 2003; Jones et al. 2003). In this study condos also increased the probability of retaining juveniles at the experimental sites. Each condo provided about 3.5m2 of surface area and consisted of 24 concrete mini- blocks stacked within a modified commercial crab trap. Standard concrete blocks were cut into quarters longitudinally to produce 4 individual mini-blocks. Discarded commercial crab traps were altered by removing the central ‘fishing’ component, leaving a structurally effective frame of corrosion-resistant metal enclosed with stainless steel mesh. Diamond shaped openings in the wire mesh frames as well as the entry/exit hole allowed access to most abalone sizes and a hinged lid allowed divers access to load, remove and examine the concrete mini-blocks during deployment and sampling. Condo deployment occurred on July 5-7, 2005. At each substation, three condos were placed 2-10 m apart, for a total of 24 condos in this study (Figure 1). Condos were deployed by lowering the traps and the cut-up blocks from a dive support vessel to the ocean floor. Divers then repositioned each structure with an industrial airlift bag. No anchoring mechanisms were needed to secure the condos in place as each condo weighed 120kg and possessed a stable base. The condos were located at an average depth of 4.7 m chart datum (Appendix A).

Abalone outplanting Due to the threatened status of abalone, this study could not use wild abalone. Instead, 1200 hatchery-raised juvenile abalone, from the broodstock years of 1999 and 2000, were purchased from Malcolm Island Shellfish Cooperative (MISC). To prevent the spread of disease to the wild a juvenile subsample from MISC were examined for parasites and pathogens prior to onset of the experiment. Although the average size of the abalone was determined to be small for their age, no abnormalities, parasites, or pathology of infectious disease of concern were found (G. Meyer, DFO, Nanaimo, pers. comm.). The juvenile abalone were outplanted directly into the condos on July 26-27, 2005, which allowed nearly a month for bacteria and diatoms to develop on the condos before abalone were transplanted into them. At the hatchery, abalone were measured and counted and placed in cages designed to hold abalone for transport and outplanting. Fifty abalone were placed in each cage. Every effort was made to ensure these abalone were in the best

81 092

possible condition during transport, such as providing cold and well oxygenated water. Divers then transferred the abalone from each cage into a submerged condo, ensuring that all abalone attached to the concrete blocks before closing the lid of the crab trap.

Survey Methods Subsequent to abalone outplanting in July, 2005, the condos were resurveyed on September 27-28, 2005, and again on March 21-22, 2006. A pair of divers sampled each condo by removing each concrete mini-block and examining it for abalone. All live abalone found were measured for maximum SL to the nearest millimeter using calipers. Abalone frequently develop drastically different shell colors when their environment or diet changes (Olsen 1968). This color change may be especially dramatic when abalone are moved from a hatchery (single food type) to the wild (varied food sources) (Gallardo et al. 2003). In our study, this distinct color band allowed the divers to visually determine and measure (for most of the abalone found) the amount each abalone had grown since outplanting. Empty abalone shells were measured and removed. After all blocks were examined, they were repositioned within the metal frame, and abalone were returned to the condo. Divers also recorded observations of urchins and common abalone predators such as Pycnopodia helianthoides, Cancer productus, and Octopus dofleini. Habitat data, such as percentage cover of the dominant algae, encrusting algae cover, and diatom presence was also recorded. Additionally, in the March survey only, divers recorded the relative abundance of other invertebrate or fish species (chitons, shrimp, hermit crabs, worms, barnacles, snails, and fish) using relative categories. At the outplanting in July and the surveys in September and March, an underwater video camera was used to record short videos of each of the condos. The purpose of the video was to qualitatively determine the amount of silt at each site. Other quantitative methods exist, but they are time-consuming and often require on-site monitoring which was not possible for this study. The videos were later examined to confirm substrate types at each condo, as well as provide estimates of the silt accumulated on the condos, and information about other species present on the condos in the months of July and September. The video also provided general site information such as visibility and current speed at survey times, and the composition of algal communities surrounding the condos. Since data gathered from the video is primarily qualitative, results are presented as generalizations and, therefore, no statistical analysis was performed. The water conditions at each site including: salinity (ppt), temperature (ºC), oxygen concentration (ppm), and oxygen percent saturation, were tested on September 27-28, 2005. An Oxyguard Handy Gamma was used to collect dissolved oxygen data, and a YSI 30 Salinity model 30/10ft collected data on salinity and temperature. Measurements at each substation were taken at the surface (1ft/0.30m depth), and at depth (30ft/9.14m).

Analytical methods A Mann-Whitney U test was used to determine overall differences in density between the control and tenure sites. Since the results appeared to differ at the two islands, separate Mann-Whitney U tests were used to test for differences in density between control and tenure sites at Bonwick and Swanson Islands individually. Only September data were

82 093

used in the density analyses, as the number of abalone recovered in March was too small for statistical comparison. There appeared to be differences observed in the number of abalone found in condos that rested on bedrock vs. those on boulder. To test whether these differences were significant a 2 x 2 contingency table was constructed using bedrock and boulder as columns, and abalone groups of few and many as rows. A Fisher’s exact test was then performed. In order to remove variation caused by site, the same procedure was repeated using only data from the control sites. ANOVA was also used to compare the mean size of the abalone outplanted as well as abalone found in the September survey. Again, the March data was not tested alone due to small sample size. Differences in abalone growth between the control and tenure sites in September were compared using ANCOVA, with original size (calculated as the measured shell length – growth recorded) as the covariate. In all statistical analyses results were considered significant at the α= 0.05 level.

RESULTS

Density Of the 1200 abalone initially released in July, 2005, a total of 96 live abalone were found in September, 2005, and 12 in March, 2006 (Table 1). In September divers found 69 abalone at control sites: 55 at Bonwick Island, and 14 at Swanson Island. Twenty seven abalone were found at tenure sites: 14 and 13 at Bonwick and Swanson Islands, respectively. Fifty seven empty shells were recovered in September: 28 from control, and 29 from tenure sites. In March, 10 live abalone were found at control sites, all of which came from Bonwick Island. Two abalone were found at tenure sites, both at Swanson Island. Three shells were recovered at control sites, and 10 were found at tenure sites in March (Table 1). Although there were more abalone found at the control than tenure sites in September, the difference was not significant (p=0.129). When considering sites separately, Bonwick Island had significantly more abalone at control sites than at tenure sites (p=0.035), while there was no difference in abundance at Swanson Island (p=1.00). The number of abalone found did not appear to be distributed evenly throughout substrate types forming the bases of the condos. Condos were split into groups based on the predominant substrate type of their base: predominantly boulder (n=12), boulder and bedrock mixed (n=2), and predominantly bedrock base (n=10), (Table 2, Appendix A). Bonwick Island consisted of mainly bedrock based condos (8/12), while Swanson Island had mainly boulder based condos (9/12). The number of abalone/condo for each of the substrate groups was determined to be: 2.7 abalone/condo for boulder based condos, 2.5 abalone/condo for mixed, and bedrock based condos had a mean of 7.1 abalone/condo (Table 2). The overall differences between substrate types were nonsignificant (p=0.501), however there was a significant difference between substrate types at the control sites (p=0.045).

83 094

Size Frequency At the time of outplanting (July, 2005), the mean shell length (SL) of all abalone was 23.92mm±0.09SE with a range of 11-35mmSL (Figure 2a). In September the mean SL was 26.68mm±0.40SE with a range of 17-37mm (Figure 2b). In March, the mean SL was 34.67mm±1.87SE with a range of 24-46mm (Figure 2c). When results for Bonwick and Swanson Islands are combined, the mean SL for the control sites in July is 24.07mm±0.12SE with a range of 14-34mm, and the mean SL for tenure sites is 23.76mm±0.13SE with a range of 10-31mm (Figure 2a). In September, the mean SL for control sites was 27.35mm±0.49SE with a range of 18-38mm, (Figure 2b). The abalone within the tenure sites in September had a mean SL of 24.96mm±0.60SE and a range of 21-33mm (Figure 2b). In March, the abalone at the control sites had a mean SL of 35.60mm±1.93SE and a range of 30-47mm (Figure 2c). Only two abalone were found at the tenure sites with a mean SL of 30.00mm±6.00SE and values of 25mm and 37mm (Figure 2c). There was no significant difference between abalone released in July at the control and tenure sites (p=0.079). In September, the abalone in tenure sites were significantly smaller than those in control sites (p=0.007). The March data were not tested alone as the data set was too small. The mean growth of abalone at Bonwick Island control sites one and two was 5.45mm and 2.67mm, respectively, in September, and 16.80mm by March (Table 3). At Swanson Island in September the mean growth at control sites one and two was 4.67mm and 5.78mm, respectively. The abalone at tenure sites at Bonwick Island in September grew an average of 4.86mm at T1, and 1.50mm at T2. Swanson Island tenure sites had mean abalone growth rates of 1.50mm at T1 and 2.00mm at T2 in September. By March, only two abalone were found, both at Swanson Island; the abalone at T1 had grown 16.00mm and the abalone at T2 had not grown at all. Highly significant results were obtained on the growth data, with abalone located at tenure sites showing far less growth than abalone located at control sites (p<0.001).

Other species and Habitat data Colonization of the condos between the time of deployment, and the first survey in July was low, consisting primarily of snails and hermit crabs which were more common at tenure sites than control sites (Appendix B). Urchins, especially green sea urchins (Strongylocentrotus droebachiensis), were common in the condos during both the September and the March surveys, and had similar abundances between the control and tenure sites. In September, nearly all of the condos had other species present. In general, Bonwick Island condos had more tenants than their Swanson Island counterparts, especially when considering urchins and barnacles. No clear pattern is present between control and tenure sites in September, except that chitons were seen only in the control sites. By March, all of the condos were being utilized by at least one other species. Predators consisted entirely of sunflower stars, Pycnopodia helianthoides, with the exception of a single red rock crab, Cancer productus, and showed the same general pattern of increasing numbers from July to March. Predators were also more prevalent at the control than tenure sites in March. Snails, mostly of the Genus Nucella, were

84 095

common in the condos throughout the study, and were always more abundant at control sites than tenure. While Nucella are predatory, and it is possible that some abalone were preyed upon, none of the empty shells recovered show evidence (bore-holes) of this occurring. Algae and encrusting algae were absent in July, and prevalent in September, especially at Bonwick Island (Appendix B). However, by March, few condos had algae left, and encrusting algae was absent again. Diatoms did not develop at any of the Bonwick Island control condos, though a few of the Swanson Island control condos showed diatom growth in July. At the tenure sites most of the condos had thick diatom growth in July and September, particularly at Swanson Island, though most condos were diatom-free in March. The silt on the condos showed a general pattern of increasing from July to September, then decreasing in March at all sites (Appendix B). There was more silt at the tenure sites than control sites at all survey times, with the exception of one site at Bonwick Island in March. However there was considerably more silt on the condos at Swanson Island than those at Bonwick Island.

Water Conditions Table 4 provides a summary of the water conditions on September 26-27, 2005 at all sites surveyed. All parameters measured were similar between locations, Swanson and Bonwick, as well as between tenure and control sites. Since only one measurement was taken at each site, statistical comparison was not possible. All sites had full strength sea water (no freshwater input) and saturation was relatively high.

DISCUSSION The objective of this study was to determine if any of the altered environmental conditions which may be present at aquaculture sites have acute effects on juvenile abalone growth and survival. It was not possible, given the short duration of the experiment and budget constraints, to test for chronic impacts or in-depth physical/ chemical examination of the water and sediments. Unfortunately, very few abalone were found in the September and March surveys, (only 8% and 1%, respectively, of the juveniles outplanted in July). These recovery rates are considerably lower than might be expected: in a California based study, Davis (1995) reports one year survival rates of 32% for juvenile Haliotis rufescens when transplanted from a hatchery to artificial habitats similar to the condos used in this study. Previous condo studies conducted by DFO, the Kitasoo, and the Haida, as well as the presence of other species in the condos in this study indicate that condos are indeed suitable habitat. Thus, there are two main reasons for the low recovery rates: the abalone have moved out of the condos or they have died. A combination of both these events is likely responsible for our observations. Through tag recovery studies, adult abalone have been found to migrate as much as 48m over a period of a year (J. Lessard, DFO, Nanaimo, pers. comm.). Juvenile abalone in this study only needed to travel a few centimeters to exit the condos. In fact, the quality

85 096

of the habitat surrounding the condos at control sites may have determined the amount of migration to a large extent, as considerably more abalone were found in condos with a bedrock base than in condos with a base of boulders, which had more available hiding spaces. It is also likely that there was high mortality of the outplanted juveniles. The empty shells recovered account for only 6% of the original population; however, mortality cannot be estimated based on the shells recovered alone (Shepherd 1998). Many of the empty shells could have been swept away by current or wave action. Furthermore, the diver’s ability to recover shell was dependant on the type of substrate present. The winter of 2005/2006 in coastal BC was characterized by severe storms (Environment Canada). These storms destroyed much of the local algae (seen quite clearly in the video data), including Macrocystis integrifolia and Agarum sp., which usually persist through the winter (Druehl and wheeler 1986; J. Lessard, DFO, Nanaimo, pers. comm.). The juvenile abalone had likely switched from feeding on diatoms to feeding on kelp (Sloan and Breen 1988; Breen and Adkins 1980), especially in control sites where diatoms were rare. Thus, winter die-off of the abalone was possibly exacerbated by loss of perennial kelp in the storms, resulting in a significant decrease in abalone found between September and March. Preexisting health conditions of the study animals also probably contributed to the high mortality rates in this experiment. Following the conclusion of the experiment, several juvenile abalone from MISC were sent for examination due to unexplained mortality in the hatchery. The test results showed that the abalone were free of serious disease and parasites, however, they appeared to be seriously malnourished (S. Bower, DFO, Nanaimo; pers. comm.). Given this information, the abalone used in this study may have been unhealthy and/or starved at the time of outplanting. High mortality rates in the control group, as well as in the tenure group, support this conclusion. This should not, however, have significantly altered our findings in regards to the comparison between the control and tenure sites, as the original populations would have been equally afflicted. In retrospect, the compromised health of the abalone may have actually contributed to our comparison; nutrition is known to play a vital role in the stress tolerance of aquatic organisms (Martins 2006). In our study, malnourishment may have exaggerated the abalone’s reaction to adverse conditions at the tenure sites, allowing us to observe effects more quickly than we would expect under normal conditions. Sutherland et al. (2005), found increased macrofaunal abundance of some species under a farm at Kent Island, in the QCS. Our results did not show any indication that aquaculture is beneficial to abalone abundance. However, the Sutherland study was focused on less sensitive species than abalone and the fish farm at Kent Island had only been in operation for a short time (six months). Winsby et al. (1996) report that: …although initial inputs of organic matter often stimulate benthic production, continued inputs result in a reduced macrobenthic species richness and changes in community structure, as sensitive species die or migrate. A recent study by Hall-Spencer et al. (2006) found that aquaculture can have effects on environments with hard-bottom substrates, even in areas with moderate currents. Visible

86 097

waste was noted up to 100m from the cage edges, and abundances of scavenging fauna increased significantly, while sensitive species and overall biodiversity decreased with proximity to farms. Yet, it is clear that some abalone are able to survive extended periods of time in the vicinity of aquaculture sites (Davies et al. 2006). Overall effects on mortality in this study were not significant, though this may largely be due to small sample sizes and high standard error. Furthermore, differences in density between control and tenure groups were far more pronounced at Bonwick Island, than at Swanson Island. The reason for this is unclear. Effects of organic waste from aquaculture sites are dependant on, water depth, farm size, bottom topography, and current velocity (Winsby et al. 1996). The mean depth of condos was slightly shallower at Swanson Island than Bonwick, and the net pen is larger at Swanson. All of which, theoretically, should lead to greater effects observed at Swanson Island. However, the current is much stronger at Swanson Island than Bonwick Island, which may be an important factor in decreasing the effects observed at Swanson Island. Substrate type may also be critical to understanding the differences between the two locations, based on the results at the control sites; Swanson Island has considerably more boulders surrounding the condos than Bonwick Island, which may have led to disproportional migration from the control condos at Swanson Island, minimizing the difference in numbers of abalone found at control and tenure sites. Nearly all of the difference in result at Bonwick and Swanson emerged from a single substation (BC1), where considerably more abalone were found than at all other sites. There were no observable physical differences between BC1 and BC2 (J. Lessard, DFO, Nanaimo, pers. comm.), or the sites at Swanson Island (aside from substrate). However, the circumstances in this study and the analysis do not allow us to determine whether the observed differences were due to an effect of substrate, or a product of some other anomaly present at BC1. Thus, we can only speculate. Though we cannot state the cause, the results at this single substation appear to be driving the observed differences in density of abalone. In fact, the results of this study suggest that the effects of aquaculture on abalone growth may be more critical than effects on mortality: the growth shown by abalone at the tenure sites was much lower compared to those at the control sites. Over time, reduced growth may significantly influence abalone populations and recovery, as fecundity is strongly linked to body size (Campbell et al. 1992). The water quality data gathered did not indicate any obvious differences between the control and tenure sites which would explain the suppressed growth rates at tenure sites. Furthermore, all values obtained were well within H. kamtschatkana tolerance limits for temperature (Dahlhoff and Somero 1993; Sloan and Breen 1988), and the salinity measured in this study is similar to average surface salinities for the QCS (Foreman et al. 2006). Likewise, the oxygen concentration measured in our study was much greater than possible late summer – early fall low concentrations (3-4ppm) in the QCS reported by Williams et al. (2003). However, only a single sample was taken at each depth strata per site, therefore, this data is not sufficient for comparison. Divers did observe some reduction (seen as black coloration on the bricks at the bottom of the condos) at some of the tenure sites, which suggests that differences in growth may be due to a parameter of water quality that was not measured.

87 098

CONCLUSION Despite the short duration of the experiment, high mortality rates, and the resultant low sample sizes in the September and March surveys, the data strongly suggests slower growth rates in abalone located near aquaculture tenure sites than those in nearby control sites, though effects on mortality are inconclusive. Nevertheless, the topic of this study is one that merits further research, and the methods and results of this experiment may provide direction for future study.

REFERENCES Abalone Recovery Team. 2002. National Recovery Strategy for the Northern Abalone (Haliotis kamtschatkana) in Canada. 22 pp. http://www-comm.pac.dfo- mpo.gc.ca/pages/consultations/fisheriesmgmt/abalone/documents/AbaloneRecovStr ategy_e.htm Alain, B. 2005. Taking stock: a profile of Canada’s aquaculture industry. VISTA on the agri-food industry and the farm community. 6 pp. http://www.statcan.ca/english/freepub/21-004-XIE/21-004-XIE2005005.pdf (Accessed August 29, 2006). Breen, P.A., and Adkins, B.E. 1980. Spawning in a British Columbia population of northern abalone, Haliotis kamtschatkana. Veliger 23: 177-179. Campbell, A., Manely, I., and Carolsfeld, W. 1992. Size at maturity and fecundity of the abalone, Haliotis kamtschatkana (Jonas), in northern British Columbia. Can. Manuscr. Rep. Fish. Aquat. Sci. 2169: 47-65. Dahlhoff, E., and Somero, G.N. 1993. Effects of temperature on mitochondria from abalone (Genus Haliotis): adaptive plasticity and its limits. J. exp. Biol. 185:151- 168. Davies, K., Atkins, M., and Lessard, J. 2006. Survey of northern abalone, Haliotis kamtschatkana, populations in Queen Charlotte and Johnstone Straits, British Columbia, May 2004. Can. Manuscr. Rep. Fish. Aquat. Sci. 2743: iii + 17 p. Davis, G.E. 1995. Recruitment of juvenile abalone (Haliotis spp.) measured in artificial habitats. Marine and Freshwater Research 46: 549-554. DeFreitas, B. 2003. Estimating juvenile northern abalone (Haliotis kamtschatkana) abundance using artificial habitats. J Shellfish Res. 22: 819-823. Druehl, L.D., and Wheeler, W.N. 1986. Population biology of Macrocystis integrifolia from British Columbia, Canada. Marine Biology 90: 173-179. Environment Canada. Climate Data and Information Archive. http://www.climate.weatheroffice.ec.gc.ca/Welcome_e.html. (Accessed October 30, 2006). Foreman, M.G.G., Stucchi, D.J., Zhang, Y., and Baptista, A.M. 2006. Estuarine and Tidal Currents in the Broughton Archipelago. 17pp. http://www.ccalmr.ogi.edu/CORIE/modeling/elcirc/broughton.pdf. (Accessed September 12, 2006). Gallardo, W.G., Bautista-Teruel, M.N., Fermin, A.C., and Marte, C.L. 2003. Shell marking by artificial feeding of the tropical abalone Haliotis asinina Linne juveniles for sea ranching and stock enhancement. Aquaculture Research 34: 839.

88 099

Hall-Spencer, J., White, N., Gillespie, E., Gillham, K., and Foggo, A. 2006. Impact of fish farms on maerl beds in strongly tidal areas. Mar. Ecol. Prog. Ser. 326: 1-9. Hargrave, B.T., Duplisea, D.E., Pfeiffer, E., and Wildish, D.J. 1993. Seasonal changes in benthic fluxes of dissolved oxygen and ammonium associated with marine cultured Atlantic salmon. Mar. Ecol. Prog. Ser. 96: 249-257. Haya, K., Burridge, L.E., and Chang, B.D. 2001. Environmental Impact of chemical wastes produced by the salmon aquaculture industry. J. Mar. Sci. 58: 492-496. Iwama, G.K. 1991. Interactions between aquaculture and the environment. Critical Reviews in Environmental Control 21: 177-216. Jones, R., DeFreitas, B., Sloan, N., Lee, L., von Boetticher, K., and Martin, K. 2003. Abalone stewardship in Haida Gwaii: Forging a long-term commitment. Proceedings of the workshop on rebuilding techniques for abalone in British Columbia. Can. Tech. Rep. Fish. Aquat. Sci. 2482: 5-19. Lanteigne, S. 2002. Current status and potential of the Canadian aquaculture industry. 70 pp. http://www.dfo-mpo.gc.ca/aquaculture/ref/Study1_e.pdf. (Accessed October 31, 2006) Martins, T.G., Cavalli, R.O., Martino, R.C., Rezende, C.E.M., and Wasielesky, W. 2006. Larviculture output and stress tolerance of Farfantepenaeus paulensis postlarvae fed Artemia containing different fatty acids. Aquaculture 252: 525-533. McShane, P.E., and Smith, M.G. 1988. Measuring abundance of juvenile abalone, Haliotis rubra Leach (Gastropoda: Haliotidae); comparison of a novel method with two other methods. Aust. J. Mar. Freswater Res. 39: 331-336. Milewski, I. 2001. Impacts of salmon aquaculture on the coastal environment: A review. Marine Aquaculture: 166-197. Olsen, D. 1968. Banding patterns of Haliotis rufescens as indicators of botanical and animal succession. Biol. Bull. 134: 139-147. Prince, J.D., and Ford, W.B. 1985. Use of anaesthetic to standardize efficiency in sampling abalone populations (genus Haliotis: Mollusca: Gastropoda). Aust. J. Mar. Freshwater Res. 36: 701-706 Ritz, D.A., Lewis, M.E, .and Shen, M.A. 1989. Response to organic enrichment of infaunal macrobenthic communities under salmonid seacages. Mar. Biol. 103:211- 214. Sasaki, R. and Shepherd, S.A. 2001. Ecology and post-settlement survival of the ezo abalone Haliotis discus hannai, on Miyagi coasts, Japan. J. Shellfish Res. 20: 619- 626. Shepherd, S.A. 1998. Studies on southern Australian abalone (genus Haliotis) XIX: long-term juvenile mortality dynamics. J. Shellfish Res. 17: 813-825. Shepherd, S.A., and Turner, J.A. 1985. Studies on southern Australian abalone (genus Haliotis). VI. Habitat preferences, abundance, predators of juveniles. J. Exp. Mar. Biol. Ecol. 93: 285-298. Sloan, N.A., and Breen, P.A. 1988. Northern abalone, Haliotis kamtschatkana, in British Columbia: fisheries and synopsis of life history information. Can. Spec. Public. Fish. Aquat. Sci. 103: 11p. Sutherland, T.F., Levings, S.A., Peterson, S.A., Sinclair, D., Jepps, S., Gillard, B., Knight, J., McPhie, R., Walton, P., Lessard, J., McGreer, E., and Taekema, B.

89 100

2005. Data report on a benthic survey conducted near Kent Island, British Columbia. Can. Data Rep. Fish. Aquat. Sci. 1166: 1-27. Williams, I.V., Groot, C., and Walthers, L. 2003. Possible factors contributing to the low productivity of the 2000 brood year pink salmon, Oncorhynchus gorbuscha, that migrate through the Broughton Archipelago. 46 pp. http://www.davidsuzuki.org/files/Oceans/PinkSalmon_full_report.pdf (Accessed October 19, 2006). Winsby, M., Sander, B., Archibald, D., Daykin, M., Nix, P., Taylor, F.J.R., Munday, D. 1996. The environmental effects of salmon netcage aquaculture in British Columbia- A literature review. Report prepared for the Ministry of Environment, Lands, and Parks, Victoria, BC, by Hatfield Consultants Ltd. And EVS Environmental Consultants.

90 101

Table 1. Summary of density results of live abalone and empty shells at control and tenure sites on Bonwick and Swanson Islands, Queen Charlotte Strait during dive surveys on July, 2005, September, 2005, and March, 2006.

Control Tenure Total No. No. No. Survey No. Live Empty No. Live Empty No. Live Empty Date Abalone Shells Abalone Shells Abalone Shells Bonwick July 300 0 300 0 600 0 September 55 24 14 22 69 46 March 10 2 0 1 10 3

Swanson July 300 0 300 0 600 0 September 14 4 13 7 27 11 March 0 1 2 9 2 10

Combined July 600 0 600 0 1200 0 September 69 28 27 29 96 57 March 10 3 2 10 12 13

Table 2. Number of condos and total number of abalone/condo (September + March) grouped by the predominant substrate underneath the condos (boulder, bedrock, and a mixture of boulder and bedrock) by site at Bonwick and Swanson Islands, Queen Charlotte Strait. Bonwick Island Swanson Island Total Substrate No. No. Abs/ No. No. Abs/ No. No. Abs/ type condos abs condo condos abs condo condos abs condo Control Boulder 1 4 4.0 6 14 2.3 7 18.0 2.6 Mixed 1 4 4.0 0 1 4.0 4.0 Bedrock 4 57 14.3 0 4 14.3 3.6 Tenure Boulder 2 6 3.0 3 8 2.7 5 14 2.8 Mixed 0 1 1 1.0 1 1 1.0 Bedrock 4 8 2.0 2 6 3.0 6 14 2.3 Total Boulder 3 10 3.3 9 22 2.4 12 32 2.7 Mixed 1 4 4.0 1 1 1.0 2 5 2.5 Bedrock 8 65 8.1 2 6 3.0 10 71 7.1

91 102

Table 3. Number of abalone, mean shell length (SL), and mean growth in mm of abalone measured in control and tenure sites at Bonwick and Swanson Islands, Queen Charlotte Strait, in September, 2005, and March, 2006. Note: growth was not measured for all abalone, thus the numbers of abalone in this table may not match those in Table 1. Bonwick Island Swanson Island Mean Mean Mean Mean No. of SL Growth No. of SL Growth Site abalone (mm) (mm) abalone (mm) (mm) September Control 1 42 28.55 5.45 3 25.67 4.67 Control 2 9 25.56 2.67 9 26.44 5.78 Tenure 1 7 28.29 4.86 6 24.67 1.50 Tenure 2 6 23.17 1.50 2 22.00 2.00

March Control 1 10 35.60 16.80 0 Control 2 0 0 Tenure 1 0 1 36.00 16.00 Tenure 2 0 1 24.00 0.00

Table 4. Water quality data taken at each of the control and tenure sites at Bonwick and Swanson Islands, Queen Charlotte Strait, in September, 2005. Measurements were taken at the surface (1ft depth, =S) and at depth (30ft, =D). In the site names B=Bonwick Island, S=Swanson Island, C=control, F=tenure, followed by the site number. X denotes equipment malfunction when measurements could not be taken.

Date Site Salinity ppt Temp (ºC) O2 ppm O2 % Sat S D S D S D S D Control Sep-27 BC 1 31.7 31.7 10.3 10.7 6 6.3 72 79 Sep-27 BC 2 30.9 32 10.4 11.4 6.2 X 80 Sep-28 SC 1 31.8 X 9.4 X 6.7 6.5 79 76 Sep-28 SC 2 X X X X 6.7 6.4 79 75 Mean 31.5 31.9 10.0 11.1 6.5 6.4 76.7 77.5 Tenure Sep-27 BT 1 31.4 31.8 10.3 11.3 6.1 6.6 80 81 Sep-27 BT 2 31.3 32.4 10.7 10.2 7 6 87 72 Sep-28 ST 1 32.1 32.1 9.4 9.6 6.5 7 75 82 Sep-28 ST 2 31.6 32 9.4 9.8 6.3 6.5 73 76 Mean 31.6 32.1 10.0 10.2 6.5 6.5 78.8 77.8

92 103

Figure 1. Maps of the Queen Charlotte Straight (right), and the sites at Bonwick (top) and Swanson (bottom) Islands showing condo placement. Site labels begin with the first letter of the location name, followed by T for tenure and C for control sites, and the substation number. Condos are marked with X.

93 104

120

Control: 100 n= 600 Control Mean= 24.07 Tenure SE= 0.12 80 Tenure: n= 600 60 Mean= 23.76 Count SE= 0.13 40

0 1 6 11 16 21 26 31 36 41 46 Shell Length (mm) b

Control: 10 n= 69 Mean= 27.35 Control SE= 0.49 Tenure 8 Tenure: n= 27 6 Mean= 24.96

Count SE= 0.60

0 1 6 11 16 21 26 31 36 41 46 Shell Length (mm) c

3.5 Control: 3 n= 10 Mean= 35.6 Control SE= 1.93 2.5 Tenure Tenure: 2 n= 2 Mean= 30.0

Count 1.5 SE= 6.00

0.5

0 1 6 11 16 21 26 31 36 41 46 Shell Length (mm) Figure 2. Size frequency distributions of abalone measured in the Queen Charlotte Straight in (a) July, 2005, (b), September, 2005, and (c), March, 2006. Results from both locations (Swanson and Bonwick Islands) are combined and split into control (light) and tenure (dark) sites. Vertical lines across represent mean shell lengths for tenure and control sites, respectively.

94 105

Appendix A. Depths and predominant substrate types for each condo at control and tenure sites in Bonwick and Swanson Islands in the Queen Charlotte Strait The most common substrate types are listed (1 = dominant, 2 = second, and 3 = third) where substrate code 1 = smooth bedrock, 2 = bedrock with crevices, 3 = boulders, 4 = cobble, 7 = sand, 8 = shell. Bonwick Island Swanson Island Substrate Substrate Condo Depth Condo Depth Site Number (m) 1 2 3 Number (ft) 1 2 3 Control C1-13 4.0 1 2 C1-19 6.4 4 3 8 Control C1-14 4.6 1 2 8 C1-20 5.2 3 2 8 Control C1-15 4.6 1 2 8 C1-21 5.2 3 Control C2-16 6.4 3 8 C2-22 4.6 3 4 8 Control C2-17 4.9 1 2 3 C2-23 3.7 3 4 Control C2-18 6.4 1 2 8 C2-24 4.0 3 4 8

Tenure T1-7 5.8 2 71 T1-1 5.2 1 2 Tenure T1-8 5.2 3 78 T1-2 4.3 1 2 3 Tenure T1-9 4.3 7 81 T1-3 2.4 1 2 8 Tenure T2-10 3.4 3 T2-4 2.4 3 Tenure T2-11 4.6 1 2 8 T2-5 4.9 3 8 Tenure T2-12 5.2 2 1 8 T2-6 4.3 3 8

95 106

Appendix B. Summary of habitat data for each condo at Bonwick and Swanson islands, Queen Charlotte Strait, in July, 2005, September, 2005, and March, 2006 Condo numbers beginning with “C” indicate control sites, while “T” indicates tenure sites. The first digit of the condo number is the site number, and the number following the dash is the actual number of the condo. The other species observed on or in the condos where P= predators (PY= Pycnopodia helianthodoides ro RR= red rock crab, Cancer productus followed by a code for size as L= large, M= medium, and S= small;), U= urchins (all green sea urchin, Strongylocentrotus droebachiensis, except for a single purple sea urchin, Strongylocentrotus purpuratus, denoted with *), C= chitons, Sh= shrimp, W= tube worms, B= barnacles, Sn= snails, and F= fish. The relative abundance of other species are placed into categories where S= single, F= few, M= many, A= abundant, and VA= very abundant. The algae are listed as species code followed by the percentage cover on the condo, where AL= Alaria sp, AG= Agarum sp, BB= brown branched, NT= Nereocystis luetkeana, RB= red branched, and RF= red foliose. Encrusting algae (Enc) is listed as percentage cover. Diatoms (D) are listed as present (Y)/ absent (N). Silt accumulated on the condos has been given a number code from 0-4, where a higher number indicates more silt cover. Cells where and X occurs are times where the video camera failed, and no other data was available.

Other Species Location Condo # P U C Sh H W B Sn F Algae Enc (%) D Silt July Bonwick C1-13 F F 0 0 N 0 Bonwick C1-14 F 0 0 N 0 Bonwick C1-15 F 0 0 N 0 Bonwick C2-16 M 0 0 N 1 Bonwick C2-17 M 0 0 N 0 Bonwick C2-18 M 0 0 N 1

Bonwick T1-7 F 0 0 N 2 Bonwick T1-8 0 0 Y 1 Bonwick T1-9 F F 0 0 N 1 Bonwick T2-10 F F 0 0 Y 1 Bonwick T2-11 M 0 0 N 0 Bonwick T2-12 M F 0 0 Y 0

Swanson C1-19 F 0 0 Y 2 Swanson C1-20 0 0 N 3 Swanson C1-21 F 0 0 N 2 Swanson C2-22 0 0 N 3 Swanson C2-23 1 0 0 Y 3 Swanson C2-24 0 0 Y 3

Swanson T1-1 0 0 Y 4 Swanson T1-2 X X X X X X X X X 0 0 Y 4 Swanson T1-3 0 0 Y 4 Swanson T2-4 F 0 0 N 3 Swanson T2-5 F F 0 0 Y 4 Swanson T2-6 F 0 0 N 3

96 107

Appendix B. Con’t.

Other Species Location Condo # P U C Sh H W B Sn F Algae Enc (%) D Silt September Bonwick C1-13 0 3 M F A F RB 5 10 N 1 Bonwick C1-14 1 PYS 3 F F F VA F RB 5 30 N 1 Bonwick C1-15 0 0 VA M RB 10 20 N 1 Bonwick C2-16 1 PYS 1* X X X X X X X RB 10 40 N X Bonwick C2-17 0 1 F F M RB 10 10 N 1 Bonwick C2-18 0 3 F F F M S RB 10 20 N 1

Bonwick T1-7 0 4 M RB 5 10 Y 2 Bonwick T1-8 0 0 M A F RB 10 5 Y 2 Bonwick T1-9 I PYS 0 M M F RB 10 5 Y 2 Bonwick T2-10 0 0 M VA M RB 10, BB 5 0 Y 2 Bonwick T2-11 0 0 A VA M RB 5, BB 5 0 Y 2 Bonwick T2-12 0 3 F M VA F RB 5 0 Y 2

Swanson C1-19 0 2 F A F A RB 5 20 N 1 Swanson C1-20 0 1 M F F M RB 2 5 N 2 Swanson C1-21 0 0 F M RB 10 20 N 1 Swanson C2-22 0 1 F F 0 0 N 2 Swanson C2-23 0 0 F F AL, NT 0 N 3 Swanson C2-24 0 0 F F 0 0 N 3

Swanson T1-1 0 2 M 0 20 Y 3 Swanson T1-2 0 0 M F M F 0 20 Y 3 Swanson T1-3 0 0 M 0 0 Y 3 Swanson T2-4 0 0 F AG 5, RF 5 0 N 3 Swanson T2-5 1 PY 0 F 0 0 N 2 Swanson T2-6 0 0 0 0 N 3

March Bonwick C1-13 0 2 F A A A A 0 0 N 2 Bonwick C1-14 0 1 F F A F F A 0 0 N 1 Bonwick C1-15 2 PY 1 F F A VA F A 0 0 N 0 Bonwick C2-16 1 PY 0 M VA M F M 0 0 N 2 Bonwick C2-17 0 0 F A M M 0 0 N 2 Bonwick C2-18 1 RR 0 F F F A RB 2 0 N 1

Bonwick T1-7 1 PYS 0 F VA F F M F S RB 10 0 Y 0 Bonwick T1-8 3 PY 1 VA A F S 0 0 N 1 Bonwick T1-9 0 1 A F F RB 10 0 Y 2 Bonwick T2-10 0 0 M A VA M S 0 0 N 0 Bonwick T2-11 0 0 F F F VA F RF 5 0 N 0 Bonwick T2-12 0 0 M F M M RB 2 0 N 1

97 108

Appendix B. Con’t.

Other Species Location Condo # P U C Sh H W B Sn F Algae Enc (%) D Silt

Swanson C1-19 0 0 M A M F F M 0 0 N 1 Swanson C1-20 4 PY 2 F F F M 0 0 N 1 Swanson C1-21 1 PY 0 F A M M 0 0 N 1 Swanson C2-22 1 PY 0 F M 0 0 N 2 Swanson C2-23 1 PY 8 F F M A 0 0 N 2 Swanson C2-24 2 PY 0 F F F 0 0 N 2

Swanson T1-1 0 9 M M 0 0 N 3 Swanson T1-2 0 0 F A F 0 0 N 2 Swanson T1-3 0 1 M F 0 0 N X Swanson T2-4 0 0 F F F RB 10 0 N 1 Swanson T2-5 0 0 F F F RB 2 0 N 4 Swanson T2-6 0 0 F F 0 0 N 4

98 109

Appendix 4 Request for Working Paper

PSARC INVERTEBRATE SUBCOMMITTEE

Date Submitted: November, 2006

Individual or group requesting advice: • L. Convey (Resource Management Biologist – Abalone and Species at Risk), K. West (Species at Risk Recovery Planning Coordinator), and DFO Habitat Managers.

Proposed PSARC Presentation Date: November 2006

Subject of Paper (title if developed): Northern abalone ‘allowable harm assessment’.

Science Lead Author: Joanne Lessard

Resource Management Lead Author: (Laurie Convey lead but not author)

Rationale for request: • Northern abalone (Haliotis kamtschatkana) is listed and protected as threatened under Schedule 1 of the Species at Risk Act (SARA). • SARA prohibits killing, harming, harassing, capturing, and taking northern abalone, and damaging or destroying abalone residences or its critical habitat (once critical habitat is identified in a recovery strategy or action plan). • Activities affecting a listed species or its critical habitat may be permitted under SARA Section 73 or Section 83(4). • A framework for an ‘allowable harm assessment’ is being adopted nationally to provide science advice for permitting activities under SARA that may affect a listed species or its critical habitat. • Works or developments that are on, in or under the water may affect abalone and/or abalone habitat. DFO Habitat Managers require protocols for the authorization under Fisheries Act Section 35 of activities that may affect abalone habitat. • Recovery activities for northern abalone are being implemented under the ‘National Recovery Strategy for the Northern Abalone in Canada’ (finalized under the Accord for the Protection of Species at Risk) and its draft action plan, and may affect northern abalone in the wild.

99 110

Objectives of Working Paper: • To provide an ‘allowable harm assessment’ for northern abalone following the nationally developed framework (http://www.dfo- mpo.gc.ca/csas/Csas/Proceedings/2004/PRO2004_040_B.pdf).

Question(s) to be addressed in the Working Paper: Follow the nationally developed framework. 1. What is present/recent species trajectory? 2. What is present/recent species status? 3. What is expected order of magnitude / target for recovery? 4. What is expected general time frame for recovery to the target? 5. What is the maximum human-induced mortality that the species can sustain and not jeopardize survival or recovery of the species? 6. What are the major potential sources of mortality/harm? More specifically for northern abalone, these may include: • Illegal harvest • Possible future directed fishing for food, social ceremonial harvest by First Nations • Predation by sea otters • Detrimental alteration of habitat by permitted activities (e.g. nearshore works or developments, finfish aquaculture farms) • Direct mortality of abalone by permitted habitat alterations • Abalone aquaculture • Population rebuilding under the recovery strategy − out-planting hatchery-raised northern abalone to the wild − aggregating mature (wild) reproductive adults • Research recommended under the recovery strategy − tagging − population surveys 7. Quantify to the extent possible the amount of mortality or harm caused for each activity. 8. Aggregate total mortality/harm attributable to all human causes and contrast with that determined in question 5. 9. Are there reasonable alternatives to the activity with the potential for less impact? 10. Are there feasible measures that could be taken to minimize impacts? 11. Is the expected level of harm below that determined in question 5? Does the projected population trajectory indicate the activity will jeopardize survival or recovery? 12. Prepare options (where justified) and recommendations for the permitting of activities, including rationales and relevant conditions.

Stakeholders Affected: • Works or developments that may affect abalone or abalone habitat and aquaculture farms. (Will be diverse and difficult to specify). • First Nations and stewardship groups involved in abalone rebuilding or research. • Abalone researchers. • First Nations.

How Advice May Impact the Development of a Fishing Plan: • There is no fishing plan for northern abalone, all harvest is closed.

100 111

• The recovery strategy could be used to permit activities pursuant to SARA Section 83(4).

Timing issues related to when Advice is necessary: • Works and development on, in and under water are ongoing and repairs may at times be urgent for human safety reasons. Applications for aquaculture leases and tenures are referred regularly to DFO from the Province of BC. DFO Habitat Managers urgently require protocols for the authorization of these activities under the Fisheries Act Section 35. • Abalone rebuilding and research activities have been well underway since 1999. Previous PSARC papers are available to provide guidance to DFO in the issuance of permits for many of these activities, however, these have not yet been compiled under the nationally adopted framework for an ‘allowable harm assessment’.

101 112 Available online at www.sciencedirect.com

Marine Pollution Bulletin 56 (2008) 903–912 www.elsevier.com/locate/marpolbul

Assessment of chemical dispersant eﬀectiveness in a wave tank under regular non-breaking and breaking wave conditions

Zhengkai Li a,*, Kenneth Lee a, Thomas King a, Michel C. Boufadel b, Albert D. Venosa c

a Center for Oﬀshore Oil and Gas Environmental Research, Bedford Institute of Oceanography, Fisheries and Oceans (DFO) Canada, One Challenger Drive, P.O. Box 1006, Dartmouth, NS, Canada B2Y 4A2 b Department of Civil and Environmental Engineering, Temple University, Philadelphia, PA 19122, USA c National Risk Management Research Laboratory, US EPA, Cincinnati, OH 45268, USA

Abstract

Current chemical dispersant effectiveness tests for product selection are commonly performed with bench-scale testing apparatus. How- ever, for the assessment of oil dispersant effectiveness under real sea state conditions, test protocols are required to have hydrodynamic conditions closer to the natural environment, including transport and dilution effects. To achieve this goal, Fisheries and Oceans Canada and the US Environmental Protection Agency (EPA) designed and constructed a wave tank system to study chemical dispersant effectiveness under controlled mixing energy conditions (regular non-breaking, spilling breaking, and plunging breaking waves). Quantification of oil dispersant effectiveness was based on observed changes in dispersed oil concentrations and oil-droplet size distribution. The study results quantitatively demonstrated that total dispersed oil concentration and breakup kinetics of oil droplets in the water column were strongly dependent on the presence of chemical dispersants and the influence of breaking waves. These data on the effectiveness of dispersants as a function of sea state will have significant implications in the drafting of future operational guidelines for dispersant use at sea. Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Oil spill; Droplet size distribution; Energy dissipation rate; Breaking waves; Dispersants

1. Introduction formulations that are less toxic and more effective for the treatment of viscous oils, the application of chemical dis- Oil spills in the open ocean are subjected to natural dispersants has recently gained popularity as one of the pri- persion processes where wave action results in the forma- mary oil spill countermeasures for reducing the overall tion of oil-in-water emulsions of small oil droplets that adverse impact of marine oil spills on the environment are eventually diluted to concentrations below toxic thresh- (NRC, 1989, 2005). In addition to operational conve- old limits (Lee, 2002; Li and Garrett, 1998; Shaw, 2003; nience, the application of dispersants to oil slicks on the Tkalich and Chan, 2002). Breaking waves, in particular, sea surface minimizes the harmful effects of floating oil play a crucial role in the dispersion of an oil slick by gen- on aquatic wildlife such as birds and marine mammals that erating velocity shear to break up and transport oil in their frequent the water surface, and potentially mitigates the turbulent flows (Li and Garrett, 1998; Shaw, 2003; Tkalich risk of oil slicks contaminating coastal and shoreline envi- and Chan, 2002). The dispersion of oil slicks is significantly ronments (NRC, 2005). enhanced in the presence of chemical dispersants, which Oil dispersion effectiveness depends on the chemical reduce the interfacial tension between oil and water (Les- properties of both dispersant and oil and on various envi- sard and Demarco, 2000). With the development of new ronmental factors (Fingas, 2000; NRC, 2005). Standard- ized bench-scale tests are frequently conducted to evaluate the effectiveness of particular chemical oil disper- * Corresponding author. Tel.: +1 902 426 3442; fax: +1 902 426 1440. sants (References in NRC, 2005) to provide data for prod- E-mail address: [email protected] (Z. Li). uct selection. However, in the context of actual at sea

113 904 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 operations, the utility of these tests is limited due primarily the tank). The water depth during the experiments was to failure to account for the transport and dilution effect in maintained at 1.25 m. Different waves are generated by a the water column (NRC, 2005). During the physical and computer-controlled flap-type wave maker situated at one chemical dispersion of oil spills, the initial break-up and end of the tank linked to an adjustable cam that controlled submergence of a surface oil slick (as well as the secondary its stroke length to alter wave-height characteristics. This break-up of the oil into smaller droplets) depends on the type of wave generator produces ‘‘deep water” waves, turbulent structures, which also play an important role in which occur when the ratio of water depth to wave length the vertical transport of oil droplets in the water column. is greater than 0.5 (Dean and Dalrymple, 1984). The wave To overcome the restrictions inherent in bench-scale tests, frequency (and thus the wavelength) is controlled by the larger facilities are required to enable a more comprehen- rotational speed of the cam. The computer-controlled sive evaluation of chemical dispersant effectiveness under wave-generator can produce both regular non-breaking a more realistic setting. In this context, DFO’s Center for waves and breaking waves with designated length, height, Offshore Oil and Gas Environmental Research (COOGER) and frequency. The system is useful for dispersion studies and US EPA’s National Risk Management Research Lab- because recurrent breaking of waves can be generated oratory designed and constructed a wave tank facility reproducibly at the same location. This is accomplished (NRC, 2005) to conduct research on the effects of wave- using the frequency sweep technique, wherein a wave of energy on the efficacy of chemical dispersants on crude oils one frequency is superimposed on another wave of a differ- under a range of controlled regular wave and breaking ent frequency, causing the wave to increase in height until it wave hydrodynamic regimes. To achieve this goal, it is eventually breaks. important to characterize the hydrodynamics of different The energy dissipation rate per unit mass (e) was used to wave conditions. Friction associated with velocity shear characterize the intensity of the breaker. It was evaluated causes the dissipation of kinetic energy of the fluid, result- by the correlation function method (Kaku et al., 2006b) ing in a temporal and spatial variation of the energy dissi- using a time series of velocity measurements at select loca- pation rate per unit mass of water (e, in watts/kg or m2/s3). tions in the tank. These measurements were obtained by an The operational hypothesis is that e plays a major role in Acoustic Doppler Velocimeter (SonTec/YSI, Inc. San the effectiveness of a dispersant, and hence it may serve Diego, CA). as an important scalable parameter to characterize chemical dispersant effectiveness under different wave-energy 2.2. Experimental procedures conditions in the field. To evaluate dispersant effectiveness as a function of e in The effects of mixing energy and the presence of a dis- our experimental wave tank, experiments were conducted persant on dispersion of crude oil were investigated using over three different energy dissipation rates (a regular a representative crude oil and a reference dispersant under non-breaking wave and two breaking waves) similar to three different mixing energy conditions. The experimental those found in the open sea (Delvigne and Sweeney, 1988). design was a two-factor mixed-level full factorial design. The test oil was MESA crude oil (Petro-Canada, Montreal, 2. Materials and methods QC) with an API (American Petroleum Institute) gravity of 29.7°; the oil was artificially weathered by aeration to 2.1. Wave tank facility 86.2% of original weight. Corexit EC9500A (Nalco Energy Service, L.P. Sugar Land, TX), a commercially available Fig. 1 is a schematic representation of the first version of product frequently stockpiled for use by oil spill response the wave tank facility located at the Bedford Institute of agencies in Canada and the United States, was used as Oceanography (Dartmouth, NS) with geometric dimen- the reference oil dispersant. For each run, 150 ml of weath- sions of 16 m long, 0.6 m wide, and 2 m high (it has subse- ered MESA crude oil was first released onto the seawater quently been doubled in length to enable more energetic surface in the middle of the tank within a rigid square waves and to reduce wave interference from the end of frame (30 cm 40 cm and 30 cm high) constructed of oleo- phobic material. Dispersant (or seawater for the control) was immediately sprayed onto the oil slick at a disper- 0.75 Sampling Dispersant sant-to-oil ratio (DOR) of 1:25 through a 0.635-mm nozzle. Oil slick LISST The frame was promptly removed from the water immediately (within less than one second) prior to the incoming 1.25 Wave paddle Wave absorbers breaking wave (or regular wave). Samples were taken from the tank at two horizontal locations (L = 1.5 and 4 m downstream from oil application), five depths (5, 20, 40, 1.0 7.0 1.5 2.5 0.5 3.5 60, and 110 cm from the surface), and four time points Fig. 1. Schematic representation (all dimensions in m) of the wave tank (1, 10, 30, and 60 min). Dispersant effectiveness was deter- facility. Note the sampling sites were 1.5 m and 4.0 m downstream from mined by comparing control (water alone) to experimental oil, respectively. (dispersant) runs. Evidence for enhanced oil dispersion

114 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 905 included deeper penetration of dispersed oil into the water In this study, the LISST-100X was situated in the wave column and the formation of smaller oil droplets. tank 4.5 m downstream from the center of the oil application area with the detection windows 0.6 m from the sur- 2.3. Analytical methods for measuring dispersed oil face of the water. It was operated in real time mode so concentration that the dynamic oil-droplet size distribution as a function of time was acquired every 3 s. To compare the effects of e Dispersed oil in aqueous samples was extracted with and the presence of dispersant on the droplet size distribu- dichloromethane (DCM) according to EPA Method tion, a mass mean diameters (MMD) of the measured 3510 C (liquid–liquid partitioning) and measured with a droplet size distribution was calculated as a function of Genesys 20 ultraviolet–visible spectrophotometer (Thermo time. The MMD is a weighted average droplet size by mass Fisher Scientific; Calgary, Canada) following an estab- or volumetric fraction P lished protocol (Chandrasekar et al., 2005; Venosa et al., mi di 2002). Standard solutions of oil for calibration were pre- MMD ¼ ; ð4Þ M pared with the oil and dispersant (or oil alone for the control) for each particular set of experimental runs. A stock where M is total mass or volume concentration, and mi is solution of dispersant–oil mixture (or oil alone for the con- mass or volume concentration of oil droplets in a size inter- trol) was prepared by adding 800 ll of the dispersant to val of average size di. 2 ml of oil and then brought up to a final volume of 20 ml with DCM. An eight-point calibration curve was pre- 2.4. Oil droplet breakup kinetic model pared from serial volumetric dilutions of the stock solution, generating the following working standards: 15, 25, 50, The droplet size distribution and its rate of evolution are 100, 250, 500, 1000, and 2500 mg l1. The samples collected determined by the process of breakage and coalescence from the wave tank were transferred to a 125 ml separatory (Baldyga and Podgorska, 1998). In our system, we deal funnel and extracted three times with 40 ml fresh DCM. with a diluted dispersed phase (crude oil), and hence it is Sample volumes were measured using a graduated cylinder legitimate to neglect the re-coalescence of the dispersed that was pre-calibrated by mass. The extracts were then oil droplets. The kinetics of oil dispersion in regard to oil adjusted to a final volume of 10 ml and transferred to a droplet breakup can be described by a modified semi- graduated 15-ml glass tube with a Teflon-lined screw cap. empirical model based on the assumption that oil droplet The vials were stored at 4 °C prior to analysis. Extracts breakage rate is proportional to the generalized Weber were analyzed for absorbance at three different wave- number defined by Hinze (1955). The droplet breakage rate lengths: 340, 370, and 400 nm. The area was determined is inversely proportional to dispersion time (Polat et al., by applying the trapezoidal rule according to the equation 1999) ðA þ A Þ30 ðA þ A Þ30 dD sD 1 Area ¼ 340 370 þ 370 400 ð1Þ ¼k0 ; ð5Þ 2 2 dt c t The concentration of oil in the extract was determined by where D is the mass mean diameter of droplet size (instead Area of mass median diameter that was used by Polat et al. C ðmg=lÞ¼ ð2Þ DCM Slope of Calibration Curve (1999)) at time t; s is the external stress (force per unit area) at time t; c is the oil and water interfacial tension; and k0 is The concentration of the sample that was collected from the rate constant. Assuming that s can be replaced with a the wave tank was calculated by time-averaged stress per unit mass (Tatterson, 1991), k0, s, and c can be collected as a new dimensionless rate con- CDCM V DCM Csample ðmg=lÞ¼ ð3Þ stant k and Eq. (5) can be integrated to give: V sample LnðD Þ¼LnðD ÞkLnðtÞð6Þ Oil-droplet size distribution in the wave tank was deter- t 1 mined with a Type C LISST-100X particle counter (Se- where D1 is the mass mean diameter of droplet size distribu- quoia Sci. Inc., Seattle, WA). This instrument is tion at 1 min. Parameter estimation was conducted by forc- frequently used by geologists to measure the particle size ing an equal value of D1 for all treatment conditions and distribution of suspended particulates (Gartner et al., performing a linear least-squares regression of the sum of 2001; Serra et al., 2002; Traykovski et al., 1999) and has residuals. Forcing an equal D1 for each treatment is a rea- been applied to determine the droplet size distribution of sonable assumption considering that oil dispersion starts dispersed oil (Sterling et al., 2004, 2003). There are 32 par- with a floating oil slick, and the measured initial size distri- ticle size intervals logarithmically placed from 2.5–500 lm bution has a similar MMD at the initial readings. Eq. (6) in diameter, with the upper size in each bin 1.18 times was fit to observed data from t = 1.0 to t = 60 min. the lower. Particle size distribution is expressed as the aver- MMD at t = 60 min (D60) were computed to compare the age volumetric concentration of oil droplets within each treatment effects on the extent of oil dispersion into interval of the size range. droplets.

115 906 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912

2.5. Statistical data analysis face. The spilling breakers were created by generating alternating trains of high frequency waves (1.0 Hz, wave height The effects of the presence/absence of chemical disper- 8 cm, duration 20 s) and low frequency waves (0.4 Hz, sant and e under regular waves, spilling breakers, and wave height 4 cm, duration 20 s). The plunging break- plunging breakers on the effectiveness of oil dispersion ing waves were produced similarly, where the high (as measured by the integrated dispersed oil concentration frequency waves had a wave height of 12 cm and the low in the water column) were investigated using a two-factor frequency waves 6 cm. The average energy dissipation rates analysis of variance (ANOVA). A two-way ANOVA was at the surface were approximately 0.02 and 0.05 watts/kg also used to determine the effects of presence/absence of for the spillers and the plungers, respectively. The e of chemical dispersant and mixing turbulence under the three the two breaking wave types decayed almost exponentially wave conditions on the dispersed oil-droplet size distribu- as a function of depth to approximately 0.001 watts/kg at tion and dispersion kinetics. When significant differences the 40 cm depth. Details of the wave tank hydrodynamic were identified, the least-significant difference (LSD) test characterization have been reported elsewhere (Wickley- was conducted to determine which treatments significantly Olsen et al., 2007). affected the oil dispersion effectiveness as determined by the criteria of dispersed oil concentration and droplet size distribution. 3.2. Dispersed oil concentration as a function of energy dissipation rate

3. Results The dispersed oil concentrations at two locations downstream from the oil addition in the presence and absence of 3.1. Characterization of energy levels chemical dispersant were plotted as a function of depth and time. The results are shown in Figs. 2–4. The average energy dissipation rate (e) of the regular Under the regular non-breaking wave conditions, the oil non-breaking wave with a constant frequency of 0.4 Hz was transported as a non-dispersed surface slick to the end and a wave height of about 6 cm was computed to be of the tank within 5 min, and the surface slick was persis- 0.0005 watts kg1 within a depth of 10–40 cm from the sur- tent for the remainder of the experiment. The oil concen-

Fig. 2. Dispersed oil concentration as a function of time and space in the presence and absence of dispersant under regular non-breaking waves: (A) 1.5-m downstream, no dispersant; (B) 1.5-m downstream, with dispersant; (C) 4.0-m downstream, no dispersant; and (D) 4.0-m downstream, with dispersant.

116 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 907

20 20

15 15

10 10

5 5 20 20 Total oil concentration ( μ l/l) Total oil concentration ( μ l/l) 0 40 0 40 20 20 40 60 40 60 60 80 60 80 Depth (cm) 100 120 Time (min) Depth (cm) 100 120 Time (min)

20 20

15 15

10 10

5 5 20 20 Total oil concentration ( μ l/l) Total oil concentration ( μ l/l) 0 40 0 40 20 20 40 60 40 60 60 80 60 80 Depth (cm) 100 120 Time (min) Depth (cm) 100 120 Time (min)

Fig. 3. Dispersed oil concentration as a function of time and space under spilling breaking waves: (A) 1.5-m downstream, no dispersant; (B) 1.5-m downstream, with dispersant; (C) 4.0-m downstream, no dispersant; and (D) 4.0-m downstream, with dispersant.

20 20

15 15

10 10

5 5 20 20 Total oil concentration ( μ l/l) Total oil concentration ( μ l/l) 0 40 0 40 20 20 40 60 60 40 60 60 80 100 80 100 Depth (cm) 120 Time (min) Depth (cm) 120 Time (min)

20 20

15 15

10 10

5 5 20 20 0 40 0 40 Total oil concentration ( μ l/l) Total oil concentration ( μ l/l) 20 20 40 60 60 40 60 60 80 100 80 100 Depth (cm) 120 Time (min) Depth (cm) 120 Time (min)

Fig. 4. Dispersed oil concentration as a function of time and space under plunging breaking waves: (A) 1.5-m downstream, no dispersant; (B) 1.5-m downstream, with dispersant; (C) 4.0-m downstream, no dispersant; and (D) 4.0-m downstream, with dispersant.

117 908 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 trations in the water column were consistently less than Table 1 a 1 mg/l at the 1.5-m (Fig. 2A) and 4-m (Fig. 2C) sampling Summary of intrusion depth and bulk water column oil concentration locations downstream, except that the near-surface oil con- Wave Dispersantb Intrusion Oil concentration c d centration was more than 2.5 mg/l at the 4-m sampling site (cm) (mg/l) (Fig. 2C). This indicates that natural oil dispersion was lim- Regular non- D 3 ± 3 (A) 0.64 ± 0.30 (A) ited under non-breaking wave conditions. The presence of breaking +D 14 ± 18 (A) 0.67 ± 0.32 (A) chemical dispersant increased oil concentrations in the Spilling breaking D 82 ± 32 (B) 2.14 ± 0.68 (B) +D 60 ± 0 (B) 3.06 ± 1.46 (B) water column at 1.5 m downstream (Fig. 2B) and at the Plunging breaking D 110 ± 0 (C) 2.60 ± 0.82 (B) surface at 4 m downstream (Fig. 2D). The surface spread- +D 110 ± 0 (C) 4.71 ± 0.46 (C) ing of oil was enhanced by the effect of chemical dispersant. a Data reported as the average ± one standard deviation of independent However, the concentrations of oil in the bulk aqueous duplicate runs; letters in parentheses indicate whether the differences phase were always less than 1 mg/l. These results indicate between each pair of treatments are statistically significant. b that at an e of 0.0005 watts/kg, very little dispersion occurs Dispersant conditions: ‘‘D”, absence of dispersant; ‘‘+D”, presence at any depth below the surface, and application of disper- of dispersant. c The intrusion depth was determined by the average depth from the sants under calm seas is likely to be ineffective at these low surface of water at which the measured oil concentration was P1 mg/l mixing energies. (approximately 10% of the nominal oil concentration for a uniform dis- The spilling breakers increased the surface e by an order tribution of oil in the water). d of magnitude and thereby increased the dispersed oil con- The average concentrations at a depth of 60 cm from the surface of the centrations in the water column (Fig. 3). In the absence water samples collected from both 1.5 m and 4.0 m downstream of the duplicate runs. of dispersant, the concentrations of the dispersed oil increased gradually over time in the upper portion (in samples taken from depths equal to 20, 40, and 60 cm of the not have a significant effect on intrusion depth (p = 0.83) water column), but were still small (Figs. 3A and C). Sim- compared to no dispersant. A further least-significant dif- ilar to the regular wave condition, the near-surface oil conference test (LSD) revealed that the intrusion depth of oil centrations at 4 m downstream were much higher in the was strongly correlated to the energy dissipation rate. absence of dispersant, suggesting weak dispersion. The dis- The two-way ANOVA was also used to analyze wave persion of oil into the water column was enhanced by the and dispersant effects on the bulk water column oil concen- presence of dispersant at this wave condition (Figs. 3B trations on the average depth of the tank (60 cm). Similar and D): the initial oil concentration was the highest at to the oil intrusion depth, there was no significant interac- the surface and decreased to a constant level after 30 min tion between wave and dispersant (p = 0.09), but both at 1.5 m downstream. The oil concentrations at 4 m down- wave condition (p = 4.3 107) and the presence of disper- stream were evenly distributed over time and approached sant (p = 0.0044) significantly increased bulk oil concentra- approximately 5 mg/l, equivalent to approximately 50% tions in the water column. The LSD test indicated that of the nominal oil concentration that would occur if one breaking waves significantly increased dispersed oil concen- assumed all the oil was dispersed uniformly throughout tration (with and without chemical dispersants), and plung- the tank. Regardless of the increased oil concentration at ing breaking waves plus dispersant resulted in the highest the upper portion of the water column, the oil concentra- oil concentration in the water column (Table 1). tions near the bottom were still low due to the insufficient penetration depth of the dispersed oil. 3.3. Dispersed oil-droplet size distribution as a function of The plunging breakers increased the energy dissipation energy dissipation rate rate by half-an-order of magnitude in comparison to the spilling breaking waves, resulting in an increased penetra- The particle size distributions from the LISST generally tion depth of oil (Fig. 4). Natural dispersion under the fit a lognormal distribution. The dispersed oil-droplet size plunging breakers slightly increased the dispersed oil con- distributions expressed as mass mean diameter (MMD) centrations in the bulk aqueous phase (Figs. 4A and C). under three different wave conditions in the presence and The presence of dispersant dramatically increased the effec- absence of chemical dispersant were plotted as a function tiveness of oil dispersion as illustrated by higher concentra- of time. The results are shown in Figs. 5–7; also shown tions of oil over the entire depth at both the 1.5-m and 4-m are the best-fits of data using the kinetic expression of downstream locations (Figs. 4B and D). Eq. (6). The effects of energy dissipation rate and the presence of Under the regular non-breaking waves, the dispersed dispersant on oil dispersion effectiveness are summarized in oil-droplet sizes were always greater than 200 lm (MMD) Table 1 by the intrusion depths and bulk oil concentrations in the absence of chemical dispersant (Fig. 5A), and the in the water column. A two-way analysis of variance presence of dispersant dramatically reduced the droplet size (ANOVA) indicated no significant interaction between especially after 30 min dispersion (Fig. 5B). The relatively wave condition and dispersant on penetration depth high variability of MMD under regular non-breaking wave (p = 0.43). Waves significantly affected intrusion depth of conditions was likely due to droplet entrainment and resur- dispersed oil (p = 1.34 1010), and the dispersant did facing of the oil. Because of the relatively large droplet size,

118 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 909

400 400

300 300

200 200 MMD ( μ m) MMD ( μ m)

100 100

0 0

300 300

200 200 MMD ( μ m) MMD ( μ m)

100 100

0 0 0 10203040506070 010203040506070 Time (min) Time (min) Fig. 6. Dispersed oil-droplet size as a function of time in the absence (A) Fig. 5. Dispersed oil-droplet size as a function of time in the absence (A) and presence (B) of dispersant under spilling breaking waves. The symbols and presence (B) of dispersant under regular non-breaking waves. The are experimental data and the lines are modeling regression of Eq. (6). symbols are experimental data and the lines are modeling regression of Eq. (6). tion and oil dispersion kinetics. The results reported in especially in the absence of chemical dispersants, buoyant Table 2 indicate a significant (p = 0.002) interaction oil particles tend to resurface, and hence overall dispersion between wave conditions and chemical dispersants on effectiveness is relatively low. droplet size distribution; the presence of chemical disper- The spilling breaking waves dispersed the oil slick into sant main effect (p = 0.0002) and the wave-energy main smaller droplets. Although MMDs lower than 50 lm were effect (p = 0.002) also significantly affected the dispersed occasionally measured within the first 10 min, relatively droplet size distribution. The ANOVA of the factorial stable droplet size distributions were detected after effects on the kinetic rate constant of the dispersion process 30 min. The droplet MMD after 60 min was lower than indicates significant interactions between wave and disper- 100 lm in the absence of dispersant (Fig. 6A) and lower sant (p = 0.008) and that the presence of chemical disper- than 50 lm in the presence of dispersant (Fig. 6B). sant significantly increased the kinetic rate constant The plunging breakers dispersed the oil into smaller (p = 0.01). Variations of e also significantly affected the rate droplet sizes similar to the spilling breaking waves, but var- constant (p = 0.0003), indicating that wave conditions were iability was much lower (Fig. 7). critical in the oil breakup kinetics. The LSD test indicated The presence of an effective chemical dispersant reduced no significant differences between the two breaking wave the average dispersed oil-droplet size, especially at lower conditions on kinetic rate coefficients or terminal dispersed mixing energies. The average oil-droplet sizes observed in oil-droplet size (Table 2). However, due to deeper penetra- this study under non-breaking waves were within a range tion of oil in the water column, plunging waves were still of 200–300 lm; the presence of chemical dispersant more effective in oil dispersion. reduced the oil-droplet size distribution by a factor of 5 to about 50 lm. Under breaking waves, however, the dif- 4. Discussion ference in oil-droplet size distribution in the presence and absence of dispersant was less pronounced, where the ter- Mixing energy at the ocean surface, in particular, the minal oil-droplet sizes were considerably reduced com- energy of the breaking waves, plays an important role in pared to regular wave conditions without dispersant. the dispersion of oil slicks (Delvigne and Sweeney, 1988; A two-way ANOVA was conducted to analyze the fac- Hinze, 1955; Li and Garrett, 1998; Shaw, 2003). The energy torial effects on the terminal dispersed droplet size distribu- dissipation rate per unit mass of water is related to the

119 910 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 energy of micro-turbulence eddies and has been proposed posed by a number of authors (Delvigne and Sweeney, as a parameter that can be scaled from laboratory to the 1988; Li and Garrett, 1998; Lunel, 1993, 1995). Lunel field, based on the conservation of energy in studies of (1995), in particular, proposed that droplets are initially chemical dispersant effectiveness (Kaku et al., 2006a,b; generated in several areas of different microscale turbu- Venosa et al., 2005). The energy levels of the different lence, but eventually all the droplets pass through the area waves that were studied in our wave tank system covered of smallest microscale turbulence so that droplet size the range of energy conditions typically observed in the becomes much smaller over time. Our observation agreed field. Terray et al. (1996) reported e to vary between 105 with Lunel’s (1995) mechanism that more than 50% of and 102 watts/kg at a wave height of about 0.25 m in the final dispersed concentration was transferred from the Lake Ontario. Drennan et al. (1996) conducted similar surface oil slick to the bulk aqueous phase within 10 min measurements in the Atlantic Ocean off the Maryland coast (Figs. 2–4), whereas further dispersion of oil into smaller and found e to vary between 1 104 and 5 104 watts/ droplets continued after an hour (Figs. 5–7). kg at wave heights of 1 m. The highest energy dissipation In this study, the kinetics of oil droplet formation have rate of the plunging breaking waves generated in our sys- been described reasonably well by the semi-empirical rate tem was near the lower end of the breaking wave-energy expression that was developed in a baffled agitated vessel dissipation level reported in the field (Delvigne and Swee- (Polat et al., 1999), and the effects of mixing energy and ney, 1988; Drennan et al., 1996; Terray et al., 1996), chemical dispersant on the rate and extent of breakup of whereas the values for regular waves were similar to those oil are clearly identified (Table 2). The small rate constant found on the sea surface layer (Delvigne and Sweeney, (k) under regular waves in the absence of dispersants indi- 1988). cates that microscale turbulence is limited so that oil dis- The results from the oil dispersion experiments support persion is ineffective and slow, but the rate coefficients our hypothesis that the dispersed oil concentration is were significantly increased under spilling and plunging strongly dependent on e (Table 1). A nonlinear regression breaking wave conditions, suggesting that the elevated of the bulk water column oil concentration (C) with e energy dissipation rates accelerated the breakup of oil reveals a strong correlation following the empirical droplets (Figs. 5–7; Table 2). The effects of chemical disper- expression sant on oil dispersion kinetics are also evident in the dispersion kinetic rate coefficients and the terminal particle size C ¼ 7:61 e0:32 ðR2 ¼ 0:95Þð7Þ D distributions (Table 2). It is expected that large e affects in the absence of chemical dispersant, and the breakup of oil droplets through its impact on the Kol- mogorov scale, g (Kolmogorov, 1949), which is an estimate C ¼ 19:21 e0:43 ðR2 ¼ 0:98Þð8Þ þD of the smallest eddy that exists prior to dissipation by vis- in the presence of dispersant. These expressions are consis- cous friction. Under regular non-breaking wave conditions, tent with the empirical relationship between oil entrain- the g is approximately 200 lm, which is the same level as ment (Q) and wave-energy per unit surface area Dba (J/ the observed terminal mass mean diameter of the oil drop- 2 0.57 m )asQ–Dba as reported by Delvigne and Sweeney lets in the absence of dispersants. Under the spilling and (1988). The observed significant effect of chemical disper- breaking waves the g was reduced to about 100 and sant and mixing energy on the increased oil concentration 70 lm, respectively, consistent with the observed level of in the water column was consistent with the chemical dis- oil-droplet sizes. Clearly, the chemical dispersant changed persant effectiveness tests that were conducted in bench- the surface properties of oil, and hence the droplets were scale flask tests (Chandrasekar et al., 2005; Sorial et al., much smaller. Indeed, a large number of dispersed oil 2004). The effect of mixing energy on total dispersed oil droplets of size ranging from 1.0 to 2.5 lm were observed concentration was also in agreement with the observation with an epifluorescence microscopic system equipped with of field trials under low and high energy regime caused motorized stage and image processing software. Since the by various wind effects (Lunel et al., 1995). The field hydro- lower detection limit of the type C LISST-100X particle dynamic parameters such as wave height and wave fre- counter used in this research was 2.5 lm, the exact size of quency as well as velocity gradients are readily the smallest droplets that were generated in the wave tank measurable to obtain energy dissipation rates. Based on systems cannot be determined. this quantitative hydrodynamic information, together with readily measurable oil physical and chemical characteristics 5. Conclusions (e.g. viscosity and interfacial tension), it is possible to predict the oil dispersability and the chemical dispersant effec- This research has demonstrated the feasibility of con- tiveness under given wave conditions. ducting chemical dispersant effectiveness tests in a meso- The observed dependence of oil-droplet size distribu- scale wave tank facility under a range of controlled wave tions on energy dissipation rates and chemical dispersants conditions. The hydrodynamic experiments have shown in this study is consistent with other reports (Byford that the regular waves and breaking waves generated in et al., 1984; Jasper et al., 1978; Lunel, 1995). Different the wave tank have similar energy dissipation rates kinetic mechanisms for droplet formation have been pro- reported in open seas. This quantitative oil dispersant

120 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912 911

400 of total dispersed oil concentrations by chemical oil dispersants is dependent on the mixing energy state at sea; applying chemical dispersant under regular non-breaking wave 300 conditions or quiescent sea states will not be nearly as eﬀec- tive as under breaking wave conditions. In almost all cir-

200 cumstances, the application of the chemical dispersant significantly stimulated the oil breakup kinetics, which in MMD ( μ m) turn, enhanced overall dispersant effectiveness, particularly 100 in breaking wave conditions. The data presented here are useful in predicting chemical dispersant effectiveness in different sea mixing 0 conditions. Correlation between total dispersed oil and oil-droplet size distribution with waves at sea may also have significance in assessing the potential biodegradation kinetics and toxicological impact of physically and 300 chemically dispersed oil at sea. Although this research presents a typical range of mixing energies and their effect on

200 the dispersion of oil, more studies are warranted to investigate the inﬂuence of various crude oils and dispersant

MMD ( μ m) formulations on dispersant effectiveness as a function of 100 energy dissipation rates. Recently, our wave tank facility has been upgraded by extending the length from 16 m to 32 m to simulate more energetic breakers and create higher 0 0 10203040506070 energy dissipation rates, and incorporating a flow-though Time (min) system in addition to batch mode to simulate commonly encountered wave- and current- driven oceanographic con- Fig. 7. Dispersed oil-droplet size as a function of time in the absence (A) ditions. Additional experimental studies are being con- and presence (B) of dispersant under plunging breaking waves. The ducted in the upgraded system. In addition, modeling of symbols are experimental data and the lines are modeling regression of Eq. (6). oil drop breakage in wave- and current-induced intermittent turbulence with computational fluid mechanics approach is currently being pursued to account for the effect of scale of the system on the break-up process includ- Table 2 Best-fit parameters for oil- droplet size as a function of time [Eq. (6)]. Data ing determining the stable and transient size of drops and shown in Figs. 5–7 the break-up rates. a b b With favorable results from net-benefit analysis during Wave Dispersant k D60 (lm) recent spill response operations, chemical oil dispersant Regular non-breaking D 0.049 ± 0.026 (A) 246 ± 26 (A) +D 0.427 ± 0.015 (C) 52 ± 3 (C) use is expected to rise. The application of chemical oil dis- Spilling breaking D 0.310 ± 0.009 (B) 85 ± 3 (B) persants must now be considered a component within +D 0.450 ± 0.010 (C) 47 ± 2 (C) future integrated ecosystem management plans. To fulfill Plunging breaking D 0.293 ± 0.104 (B) 95 ± 39 (B) this need, it is required to have additional knowledge on +D 0.374 ± 0.018 (C) 65 ± 5 (C) the effects of mixing energy and chemical dispersants on a Dispersant conditions: ‘‘D”, absence of dispersant; ‘‘+D”, presence the transport (dispersion), fate (persistence) and effects of dispersant. (acute and chronic toxicity) of dispersed oil in the water b Model parameters (rate constant k and average dispersed drop size column. D60) reported as the average ± one standard deviation of independent duplicate runs; letters in parentheses indicate whether the differences between each pair of treatment conditions are statistically significant. Acknowledgements

This research was funded by the Panel of Energy Re- effectiveness study has provided a comprehensive data set search and Development (PERD), US EPA (Contract describing dispersed oil concentrations and dispersed oil- No. 68-C-00-159), and NOAA/UNH Coastal Response droplet size distributions under different wave conditions. Research Center (NOAA Grant Number: NA04- Our findings suggest that dispersant effectiveness is directly NOS4190063 UNH Agreement No. 06-085). Essential related to energy dissipation rates of waves. Breaking technical and logistical support was provided by Jennifer waves provide orders-of-magnitude higher energy dissipa- Dixon, Xiaowei Ma, Peter Thamers, Jay Bugden, Susan tion rates than regular non-breaking waves. Thus, better Cobanli, Matt Coady, and Matt Arsenault. Views ex- effectiveness of oil dispersion can be achieved under condi- pressed by the authors do not necessarily reflect the positions that increase e, namely breaking waves. Enhancement tion of the US Environmental Protection Agency.

121 912 Z. Li et al. / Marine Pollution Bulletin 56 (2008) 903–912

References Chemicals in Oil Spill Response, ASTM STP 1252, American Society for Testing and Materials, Philadelphia, PA, pp. 240–285. Baldyga, J., Podgorska, W., 1998. Drop break-up in intermittent Lunel, T., Davies, L., Brandvik, P.J., 1995. Field trials to determine turbulence: maximum stable and transient sizes of drops. The dispersant effectiveness at sea. In: Proceedings of the 18th Arctic and Canadian Journal of Chemical Engineering 76, 456–470. Marine Oilspill Program (AMOP) Technical Seminar, Edmonton, Byford, D.C., Laskey, P.R., Lewis, A., 1984. Effect of low temperature Alberta, Canada. Environment Canada, Ottawa, Ontario, Canada, pp. and varying energy input on the droplet size distribution of oils treated 629–651. with dispersants. In: Proceedings of the Seventh Annual Arctic and NRC, 1989. National Research Council: Using Oil Spill Dispersant on the Marine Oilspill Program (AMOP) Technical Seminar. Environment Sea. The National Academies Press, Washington D.C. Canada. Ottawa, Ontario, Canada, pp. 208–228. NRC, 2005. National Research Council: Understanding Oil Spill Disper- Chandrasekar, S., Sorial, G.A., Weaver, J.W., 2005. Dispersant effective- sants: Efficacy and Effects. The National Academies Press, Washing- ness on three oils under various simulated environmental conditions. ton, DC. Journal of Environmental Engineering and Science 22 (3), 324–336. Polat, H., Polat, M., Chander, S., 1999. Kinetics of oil dispersion in the Dean, R.G., Dalrymple, R.A., 1984. Water Wave Mechanics for absence and presence of block copolymers. AICHE Journal 45 (9), Engineers and Scientists. Prentice-Hall, Inc., Englewood Cliffs, NJ. 1866–1874. Delvigne, G.A.L., Sweeney, C.E., 1988. Natural dispersion of oil. Oil and Serra, T., Casamitjana, X., Colomer, J., Granata, T.C., 2002. Observa- Chemical Pollution 4 (4), 281–310. tions of the particle size distribution and concentration in a coastal Drennan, W.M., Donelan, M.A., Terray, E.A., Katsaros, K.B., 1996. system using an in situ laser analyzer. Marine Technology Society Oceanic turbulence dissipation measurements in SWADE. Journal of Journal 36 (1), 59–69. Physical Oceanography 26 (5), 808–815. Shaw, J.M., 2003. A microscopic view of oil slick break-up and emulsion Fingas, M.F., 2000. Use of Surfactants for Environmental Applications. formation in breaking waves. Spill Science & Technology Bulletin 8 (5– Surfactants: Fundamentals and Applications to the Petroleum Indus- 6), 491–501. try. In: Schramm, L.L. (Ed.). Cambridge University Press, pp. 461– Sorial, G.A., Venosa, A.D., Koran, K.M., Holder, E., King, D.W., 2004. 539. Oil spill dispersant effectiveness protocol. II: Performance of revised Gartner, J.W., Cheng, R.T., Wang, P.F., Richter, K., 2001. Laboratory protocol. Journal of Environmental Engineering–ASCE 130 (10), and field evaluations of the LISST-100 instrument for suspended 1085–1093. particle size determinations. Marine Geology 175 (1–4), 199–219. Sterling, M.C., Bonner, J.S., Ernest, A.N.S., Page, C.A., Autenrieth, R.L., Hinze, J.O., 1955. Fundamentals of the hydrodynamic mechanism of 2004. Characterizing aquatic sediment–oil aggregates using in situ splitting in dispersion processes. Journal of AICHE 1, 289– instruments. Marine Pollution Bulletin 48 (5–6), 533–542. 295. Sterling, M.C., Bonner, J.S., Page, C.A., Fuller, C.B., Ernest, A.N.S., Jasper, W.L., Kim, T.L., Wilson, M.P., 1978. Droplet size distributions in Autenrieth, R.L., 2003. Partitioning of crude oil polycyclic aromatic a treated oil-water system. In: McCarthy, L.T.J., Lindblom, G.P.,Wal- hydrocarbons in aquatic systems. Environmental Science & Technol- ter, H.F. (Eds.), Chemical Dispersants for the Control of Oil Spills, ogy 37 (19), 4429–4434. ASTM STP 659, American Society for Testing and Materials, Tatterson, G.B., 1991. Fluid Mixing and Gas Dispersion in Agitate Tanks. Philadelphia, PA, pp. 203-216. McGraw Hill, New York, NY. Kaku, V.J., Boufadel, M.C., Venosa, A.D., 2006a. Evaluation of mixing Terray, E.A., Donelan, M.A., Agrawal, Y.C., Drennan, W.M., Kahma, energy in laboratory flasks used for dispersant effectiveness testing. K.K., Williams, A.J., Hwang, P.A., Kitaigorodskii, S.A., 1996. Journal of Environmental Engineering–ASCE 132 (1), 93–101. Estimates of kinetic energy dissipation under breaking waves. Journal Kaku, V.J., Boufadel, M.C., Venosa, A.D., Weaver, J., 2006b. Flow of Physical Oceanography 26 (5), 792–807. dynamics in eccentrically rotating flasks used for dispersant effective- Tkalich, P., Chan, E.S., 2002. Vertical mixing of oil droplets by breaking ness testing. Environmental Fluid Mechanics 6 (4), 385–406. waves. Marine Pollution Bulletin 44 (11), 1219–1229. Kolmogorov, A.N., 1949. The local structure of turbulence in an Traykovski, P., Latter, R.J., Irish, J.D., 1999. A laboratory evaluation of incompressible fluid for very large Reynolds numbers. Doklady the laser in situ scattering and transmissometery instrument using Adademii Nauk SSSR 30, 301–305. natural sediments. Marine Geology 159 (1–4), 355–367. Lee, K., 2002. Oil-particle interactions in aquatic environments: influence Venosa, A.D., Kaku, V.J., Boufadel, M.C., Lee, K., 2005. Measuring on the transport, fate, effect and remediation of oil spills. Spill Science energy dissipation rates in a wave tank. In: Proceedings of 2005 & Technology Bulletin 8 (1), 3–8. International Oil Spill Conference, Miami, FL. American Petroleum Lessard, R.R., Demarco, G., 2000. The significance of oil spill dispersants. Institute, Washington D. C. Spill Science & Technology Bulletin 6 (1), 59–68. Venosa, A.D., King, D.W., Sorial, G.A., 2002. The baffled flask test for Li, M., Garrett, C., 1998. The relationship between oil droplet size and dispersant effectiveness: a round robin evaluation of reproducibility upper ocean turbulence. Marine Pollution Bulletin 36, 961–970. and repeatability. Spill Science & Technology Bulletin 7 (5–6), 299– Lunel, T., 1993. Dispersion: Oil droplet size measurement at sea. In: 308. Proceedings of the Sixteeth Arctic and Marine Oil Spill Program Wickley-Olsen, E., Boufadel, M.C., King, T., Li, Z., Lee, K., Venosa, (AMOP) Technical Seminar, Calgary, Alberta, Canada. Environment A.D., 2007. Regular and breaking waves in wave tank for dispersion Canada, Ottawa, Ontario, Canada, pp. 1023–1056. effectiveness testing. In: Proceedings of the 30th Arctic and Marine Lunel, T., 1995. Understanding the mechanism of dispersion through oil Oilspill Program (AMOP) Technical Seminar, Edmonton, AB, Can- droplet size measurements at sea. In: Lane, P. (Ed.), The Use of ada. Environment Canada, Ottawa, Ontario, Canada, pp. 161–187.

122 ENVIRONMENTAL ENGINEERING SCIENCE Volume 26, Number 6, 2009 ª Mary Ann Liebert, Inc. DOI: 10.1089=ees.2008.0377

Evaluating Chemical Dispersant Efﬁcacy in an Experimental Wave Tank: 1, Dispersant Effectiveness as a Function of Energy Dissipation Rate

Zhengkai Li,1,* Kenneth Lee,1 Thomas King,1 Paul Kepkay,1 Michel C. Boufadel,2 and Albert D. Venosa3

1Centre for Offshore Oil and Gas Environmental Research, Bedford Institute of Oceanography, Fisheries and Oceans (DFO) Canada, Dartmouth, Canada. 2Department of Civil and Environmental Engineering, Temple University, Philadelphia, Pennsylvania. 3National Risk Management Research Laboratory, US EPA, Cincinnati, Ohio.

Received: November 24, 2008 Accepted in revised form: February 19, 2009

Abstract Numerous laboratory test systems have been developed for the comparison of efﬁcacy between various chemical oil dispersant formulations. However, for the assessment of chemical dispersant effectiveness under realistic sea state, test protocols are required to produce hydrodynamic conditions close to the mixing, transport, and dilution effects found in the natural environment. To this end, we have designed and constructed an experimental wave tank system capable of generating waves of different energy levels, ranging from regular non- breaking waves to plunging breakers. The hydrodynamics of these wave conditions were characterized using an autocorrelation function method applied to in situ velocity measurements. We report here an investigation of the effectiveness of two chemical dispersants (Corexit 9500 and SPC 1000) on two crude oils (weathered MESA and fresh ANS) under three different wave conditions in the wave tank operated in batch mode. Corexit 9500 dispersed approximately 75% of the weathered MESA and more than 90% of the fresh ANS crude, and SPC 1000 dispersed about 53 and 64%, respectively. Under control conditions (absence of chemical dispersant), only 10 to 20% of the crude oils were dispersed. Quantitative relationships were established between dispersant effectiveness and energy dissipation rate under the different simulated wave conditions. These relationships are essential for the development of accurate predictive models on dispersant effectiveness and operational guidelines for dispersant use.

Key words: oil spill; dispersant effectiveness; energy dissipation rate; breaking waves

Introduction tiveness of chemical dispersants depends on hydrodynamic conditions, chemical properties of both dispersant and oil, atural physical dispersion of oil spills through wave and a variety of environmental factors (NRC, 1989, 2005; Naction results in the formation of oil-in-water emulsions Fingas, 2000). of mm-scale oil droplets (Delvigne and Sweeney, 1988; Lee The importance of wave action for physical and chemical et al., 2001; Daling et al., 2003). The application of chemical dispersion of oil has been recognized (Delvigne and Sweeney, dispersants accelerates dispersion of oil by reducing the oil– 1988; Li and Garrett, 1998; Tkalich and Chan, 2002; Shaw, water interfacial tension, which facilitates droplet formation 2003). Oil dispersion effectiveness is determined by several and results in reduced droplet collision rates as the oil is di- related processes, including initial breakup of the oil slick into luted in the water column (Lessard and Demarco, 2000; NRC, small oil droplets, transport and dilution of oil droplets in the 2005; Chapman et al., 2007). In addition, chemical dispersants water column, and coalescence and resurfacing of oil droplets. promote the formation of smaller droplets than those gener- The formation of droplets occurs during oil breakup under the ated solely by physical dispersion, and can also change the inﬂuence of mixing energy as the turbulent structure of surface thermodynamic properties of the oil to increase the breaking waves stretch and split the oil by velocity shear. The stability of these small oil droplets in seawater. The effec- transport and dilution of oil droplets is then regulated by turbulent diffusion and wave propagation. Coalescence and resurfacing of oil is more likely to occur when the dispersed *Corresponding author: Centre for Offshore Oil and Gas Environ- phase volumetric fraction (concentration) is high. Droplet mental Research, Bedford Institute of Oceanography, Fisheries and Oceans (DFO) Canada, 1 Challenger Drive, Dartmouth, NS B2Y 4A2, coalescence kinetics are dependent on collision frequency Canada. Phone: 902-426-3442; Fax: 902-426-1440; E-mail: liz@dfo-mpo (proportional to the shear and differential surfacing=settling .gc.ca velocity) and collision efﬁciency, which is determined by

1139 123 1140 LI ET AL. droplet surface thermodynamic properties. The resurfacing of pation rate per unit mass of water (e). An autocorrelation oil is driven by the buoyancy force that is proportional to function (Kresta and Wood, 1993; Kaku et al., 2006) applied to surface area or the square of droplet size. time-series velocity measurements obtained by an Acoustic Tests of chemical dispersant effectiveness (DE) are typically Doppler Velocimeter (SonTec=YSI, Inc. San Diego, CA) was conducted on different scales, ranging from laboratory jar used to evaluate e. tests to mesoscale wave tank testing and large-scale field trials (NRC, 2005). Bench-scale testing can be useful for comparison Wave conditions of dispersant product effectiveness (Venosa et al., 2002; Sorial The hydrodynamic characteristics of three wave conditions et al., 2004) and for investigating the effect of environmental were characterized prior to oil dispersion experiments. Reg- factors such as temperature, salinity, and other variables ular nonbreaking waves were generated with the flap stroke (Chandrasekar et al., 2005; 2006). However, laboratory tests set at 12 cm to produce a constant frequency wave of 0.80 Hz, are inherently limited in simulating real field operational a wave length of 2.44 m, and a wave height of about 23 cm. performance due to space constraints that are critical for Low-energy spilling breakers were created with an 8.4-cm transport and dilution efficiency. To account for the important flap stroke by generating alternating trains of high-frequency interplay between wave-propagation and wave-breaking- waves (0.85 Hz, wave length 2.16 m, wave height 18 cm, and induced turbulence, we used the wave tank located at the duration 20 s) and low-frequency waves (0.48 Hz, wave Bedford Institute of Oceanography (BIO) in Dartmouth, Nova length 6.77 m, wave height 6 cm, and duration 5 s). Plunging Scotia, to allow for evaluation of chemical DE under a variety breaking waves were produced with a 12-cm flap stroke and of simulated wave conditions. This wave tank is capable of alternating trains of high-frequency waves (0.85 Hz, wave generating recurrent breaking waves at the same locations by length 2.16 m, wave height 26 cm, and duration 20 s) and low- using the frequency sweep technique (Funke and Mansard, frequency waves (0.5 Hz, wave length 6.24 m, wave height 1979), wherein a low-frequency, fast-moving wave is super- 9 cm, and duration 5 s). The characterization of the wave tank imposed onto a high-frequency, slow-moving wave, causing hydrodynamics has been reported in detail elsewhere the wave to increase in height until it breaks. In characterizing (Wickley-Olsen et al., 2008). the hydrodynamics of different wave conditions, since the friction associated with velocity shear causes the dissipation Dispersants of kinetic energy of the fluid and results in a temporal and spatial variation of the energy dissipation rate per unit mass of Two commercial chemical dispersants, Corexit 9500 and water (e,inW kg1 water), the intensity of microscale tur- SPC 1000, were tested. Both dispersants are listed on EPA’s bulence of regular nonbreaking and breaking waves was National Contingency Plan Product Schedule (http:== www quantified by computing their energy dissipation rate using .epa.gov=emergencies=content=ncp=index.htm). The precise an autocorrelation function approach (Kresta and Wood, formulas of the dispersants are proprietary. Corexit 9500 is a 1993; Kaku et al., 2006). A previous study on the first version of hydrocarbon-based reformulation of water-based Corexit the wave tank demonstrated the significant effects of breaking 9527 and is meant to be used on higher viscosity oils and waves on chemical DE and dispersed oil droplet size distri- emulsions. SPC 1000 is a water-based formulation. bution (Li et al., 2008a). The present experimental study has been designed to evaluate the effectiveness of chemical dis- Crude oils persants of different formulations on different crude oils un- Two types of crude oil of varying viscosities were tested: der a variety of wave conditions, and to resolve the effects of (1) medium sulphurous crude oil (MESA), with API (Ameri- breaking wave energy dissipation rate and dispersant appli- can Petroleum Institute) gravity of 29.7, and (2) Alaska North cation on oil dispersion. The quantitative relationships be- Slope (ANS), with API of 29.6. MESA oil was obtained from tween DE and energy dissipation rates under varying wave the Petro-Canada Refinery in Montreal (QB, Canada), and conditions have been sought to establish better operational was weathered by sparging with air at a pressure of 18 psi for guidelines of dispersant use and to develop improved pre- 130 h, resulting in 13.8% evaporation of the volume of oil to dictive models of DE in the field. This is the first of a series of simulate the loss of volatile components at sea shortly after a two papers; the second paper focuses on elucidating signifi- spill. ANS crude oil was provided by ExxonMobil (Fairfax, cant factors determining in situ dispersed oil droplet size VA), and was not weathered to enable the testing of DE under distribution (Li et al., submitted). ideal conditions where dispersant is applied immediately after a spill. Materials and Methods Wave tank description Experimental design Figure 1 shows a schematic of the wave tank used in this The wave tank study was conducted using a three-factor research. It measures 32 m long, 0.6 m wide, and 2 m high, (dispersant, oil, and wave) mixed-level (223) factorial ex- with an average water depth of 1.5 m. Different waves are perimental design with triplicate runs for every treatment. generated by a computer-controlled flap-type wave maker These treatments were operated in a randomized block design situated at one end of the tank. The wave maker is linked to an containing three blocks of 12 experimental units per block. adjustable cam that controls stroke length to alter wave- The treatments were randomly assigned to the units in each height characteristics; wave frequency is controlled by the block, with each treatment appearing exactly once in every rotation speed of the cam. The wave tank can produce both block. The advantage of a randomized block design over the regular nonbreaking waves and breaking waves, with their completely randomized design in this study was to minimize mixing energy levels quantified in terms of the energy dissi- the bias on the observed DE caused by variations in envi-

124 DISPERSANT EFFICACY AND ENERGY DISSIPATION RATE 1141

FIG. 1. Photograph (upper) and schematic representation (lower, all dimensions in centimeters, not to scale) of the wave tank facility at the Bedford Institute of Oceanography. Color images available online at www.liebertonline.com=ees.

ronmental conditions such as water temperature, salinity, and sampling at four horizontal locations (8, 12, 16, and 20 m wind. Experimental seawater temperature and salinity were downstream from the wave maker), three depths (5, 75, and recorded as 16.3 1.88C, and 26.7 3.2 ppt, respectively, 140 cm from the average water surface), and four time points during the experimental period. (5, 30, 60, and 120 min after initial mixing). Samples were extracted in dichloromethane (DCM) according to EPA Experimental procedures Method 3510C (liquid–liquid partitioning) and analyzed for total petroleum hydrocarbons (TPH) with a Genesys 20 ul- For each experiment, fresh seawater was pumped from the traviolet spectrophotometer (Thermo Fisher Scientific; Cal- Bedford Basin through a double-layer sock-filter (Atlantic gary Canada) as described previously (Li et al., 2008a). Purification Ltd, Dartmouth, NS, Canada) with a pore size of m 25 and 5 m for the coarse and fine filters, respectively. DE Background temperature, salinity, and particle size distribution were recorded before each experiment. To start an ex- DE is defined as the mass of dispersed oil in the water periment, 300 mL of crude oil was gently poured onto the column divided by the total mass of oil added: water surface within a 40-cm (inner diameter) ring (constructed of NSF-51 reinforced clear PVC tube) located 10 m C V % ¼ sample wt · from the wave maker. Immediately after oil addition, 12 mL of DE( ) 100 (1) qOilVOil dispersant (or water for the control) was sprayed on top of the oil through a pressurized nozzle (60 psi; 0.635 mm i.d.) re- where: Vwt ¼ total water volume in the wave tank (27,000 L); sulting in a dispersant-to-oil ratio (DOR) of 1:25. The ring was Csample represents the average concentration of oil measured 1 then lifted immediately prior to the upcoming of the first in the water column; oil is the density of the test oil (g L ), wave, and waves were generated continually during the next and Voil is the volume of oil used in the experiment (300 mL). 2 h. Wave tank water samples were collected by using four The DE of different dispersants on different crude oils sets of 100 mL syringes connected to a stainless steel manifold can be empirically related by a power law to the energy

125 1142 LI ET AL. dissipation rate, e, under different wave conditions (Delvigne baffled flasks at the standard mixing speed of 150 rpm (Kaku and Sweeney, 1988; Li et al., 2008a): et al., 2006). Disregarding other factors such as transport, dilution, and wall interference, the baffled flasks may pro- DE ¼ a · eb (2) vide representative mixing energy levels that are required for the testing of chemical DE in the field. where a and b are dimensionless regression coefficients; e is energy dissipation rate per unit mass of water in watt kg1 or Effects of dispersant and wave conditions m2 s3. Physically, a represents the ultimate maximum DE, on oil distribution in the wave tank and b indicates the dependency of DE on energy dissipation rate for different wave conditions. Equation (2) was applied to To evaluate DE, the fraction of added oil entrained in the perform a nonlinear regression of the experimental observa- water column of the wave tank must be determined. This can tions for different combinations of oil (MESA and ANS) and be accomplished either by measuring the amount of oil re- dispersant (water, Corexit 9500, and SPC 1000). The average maining on the surface after mixing in the presence of che- DE of the triplicate data was fitted by nonlinear least-squares mical dispersants or by measuring the oil concentration in the regression to obtain the kinetic parameters. water column (NRC, 2005, and refs therein). The indirect method of measuring oil at the surface has been questioned Data analysis because of incomplete recovery of oil fractions from com- partments that cannot be explicitly measured, such as those The effects of dispersant type, wave type, oil type, and evaporated into the atmosphere and irreversibly absorbed to dispersion time on the effectiveness of oil dispersion (as the walls (Fingas and Ka’aihue, 2004). In contrast, we col- determined by the dispersed oil concentration in the water lected a large number of samples with high resolution in space column) were analyzed using a four-way factorial analysis and time to directly measure the oil dispersed in the water of variance (ANOVA) using statistical data analysis package column. Figures 2 to 4 display representative contour plots of S-Plus7.0(InsightfulInc.,Seattle,WA).Whenasignificant the MESA oil concentration in the wave tank. The ANS oil factor was identified, the Tukey’s paired comparison method distributions in the wave tank (data not shown) were similar was used to clarify the significance of effect at each treatment as those of MESA oil. level. Figure 2A shows the control condition where only seawater was sprayed onto the oil slick under regular wave conditions. Results and Discussion The added oil, which remained on the surface, was rapidly transported to the end of the tank due to the absence of dis- Hydrodynamic characterization of the three persant that would reduce the oil–water interfacial tension wave conditions and the lack of sufficient mixing energy to break up the slick. The average energy dissipation rates were estimated to be The ineffective natural dispersion of oil under regular wave approximately 0.005, 0.1, and 1 W kg1 near the surface at conditions is clearly shown in Fig. 2A, with high oil concen- the mixing zone for regular nonbreaking waves, spilling tration on the surface near the wave absorbers, and the slow breaking waves and plunging breaking waves, respectively. movement of dispersed oil upstream. In contrast, dispersion The energy dissipation rates declined exponentially to ap- of MESA oil was significantly more effective over time proximately 0.001 W kg1 for the two breaking wave con- after the slick was sprayed with either Corexit 9500 or SPC ditions at a depth of 30 cm and decreased linearly under 1000 (Fig. 2B and C). Although the distribution of oil in the regular nonbreaking wave conditions (Wickley-Olsen et al., tank was similar to the control at 5 min, the oil became pro- 2008). The average energy dissipation rates of the three wave gressively more dispersed over time as its concentration conditions that were investigated in this research are similar steadily declined at the furthest downstream surface sam- to the mixing energies observed in the field. Terray et al. pling, and the oil plume dispersed upstream and deeper into (1996) reported e to vary between 105 and 102 W kg1 at a the water column. The overall dispersion was more evi- wave height of about 0.25 m in Lake Ontario. Drennan et al. dent after the addition of Corexit 9500 (Fig. 2B) than SPC 1000 (1996) conducted similar measurements in the Atlantic (Fig. 2C). Ocean off the Maryland coast and found e to vary between Under spilling breaking waves (Fig. 3), the spreading of (1 * 5)104 W kg1 at wave heights of 1 m. The high en- oil at the surface was enhanced under physical dispersion ergy dissipation rate under plunging breaking waves in our (Fig. 3A) and chemical dispersion (Fig. 3B and C). This was system was similar to the breaking wave energy dissipation probably attributable to the dissipation of total kinetic energy rate reported in the field (Delvigne and Sweeney, 1988; as microscale turbulent eddies under the breaking waves, Drennan et al., 1996; Terray et al., 1996), whereas the values dampening the downstream drift velocity of the water and for regular waves were similar to those found on the sea increasing the turbulent diffusion of dispersed oil droplets. surface layer (Delvigne and Sweeney, 1988). Interestingly, Chemical dispersion by Corexit 9500 (Fig. 3B) or SPC 1000 the average energy dissipation rates of the three wave con- (Fig. 3C) under spilling breaking waves increased oil disper- ditions measured here are in line with the energy dissipation sion significantly compared to the regular wave conditions rates in the Baffled Flask at mixing speed of 100, 150, and (Fig. 2A–C), as indicated by the movement of oil farther up- 200 rpm, respectively (Kaku et al., 2006). The Baffled Flask stream. However, the depth of penetration of the oil plume has been proposed by the U.S. EPA (Venosa et al., 2002; Sorial was limited to the upper half of the tank, similar to dispersion et al., 2004) as a replacement of the Swirling Flask Testing under regular wave conditions. (SFT) protocol (EPA, 1996). The SFT was measured to have Under plunging breaking waves (Fig. 4A–C), spreading of orders of magnitude lower energy dissipation rates than the the oil plume was even more pronounced compared to the

126 DISPERSANT EFFICACY AND ENERGY DISSIPATION RATE 1143

A A 2 -35 15 30 -35 4 6 -70 45 -70 60 8 -105 75 -105 10 5 min 5 min 12 -140 -140 7 -35 1 14 -35 2 21 -70 -70 3 28 4 -105 30 min 35 -105 5 30 min 6 -140 -140 -35 1.2 1 2.4 -35 2 -70 3.6 3 4.8 -70 4 -105 60 min 6.0 -105 5 -140 60 min 6 -140 -35 5 10 -35 1 -70 15 2 20 -70 3 -105 25 4 Depth (cm) Depth (cm) Depth (cm)120 Depth (cm) min -105 -140 Depth (cm) Depth (cm) Depth (cm)120 Depth (cm) min 5 10 12 14 16 18 20 -140 10 12 14 16 18 20 Distance from the wave maker (m) Distance (m) B B 10 -35 18 -35 20 36 30 -70 54 -70 40 72 50 -105 5 min 90 -105 5 min 60 -140 -140 4 -35 3.5 -35 7.0 8 -70 10.5 -70 12 14.0 16 -105 17.5 -105 30 min 20 30 min 24 -140 -140 -35 1.8 -35 3 3.6 6 -70 5.4 -70 9 7.2 -105 -105 12 60 min 9.0 60 min 15 -140 -140 -35 3 -35 3.0 4 4.5 -70 5 -70 6.0 6 7.5 -105 7 -105 Depth (cm) Depth (cm) Depth (cm) Depth (cm) 120 min Depth (cm) Depth (cm)120 Depth (cm) Depth (cm) min 9.0 -140 -140 10 12 14 16 18 20 10 12 14 16 18 20 Distance from the wave maker (m) Distance (m) C C -35 10 -35 2 20 4 -70 30 -70 6 40 8 -105 50 -105 5 min 5 min 10 -140 -140 12 2.5 -35 -35 5.0 4 7.5 -70 -70 6 10.0 8 -105 12.5 -105 30 min 10 30 min 15.0 12 -140 -140 1 -35 -35 2 4 -70 3 -70 6 4 -105 60 min 8 -105 5 10 60 min 6 -140 -140 1.5 -35 -35 3 3.0 4 -70 4.5 -70 5 6.0 -105 120 min

-105 Depth (cm) Depth (cm) Depth (cm) Depth (cm) 6 Depth (cm) Depth (cm) Depth (cm)120 Depth (cm) min 7.5 -140 -140 10 12 14 16 18 20 10 12 14 16 18 20 Distance (m) Distance (m)

FIG. 2. Dispersed oil concentration (mg L1) as a function FIG. 3. Dispersed oil concentration (mg L1) as a function of time and space under regular nonbreaking waves: (A)no of time and space under spilling breaking waves: (A)no dispersant; (B) with corexit 9500; (C) with SPC 1000. Color dispersant; (B) with corexit 9500; (C) with SPC 1000. Color images available online at www.liebertonline.com=ees. images available online at www.liebertonline.com=ees.

127 1144 LI ET AL.

A nonbreaking and spilling breaking waves. In the no-dispersant -35 1 control condition, spreading at the surface was higher, as 2 -70 3 shown by the reduced net drifting of oil at the surface (Fig. 4A). -105 4 5 min 5 In the presence of chemical dispersants, the oil plume appeared Depth (cm) -140 virtually homogenized in the wave tank at all depths due to the -35 1 combined effect of more vigorous turbulent diffusion created 2 -70 3 by the plunging breaking waves and the presence of chemical 4 -105 dispersants (Fig. 4B and C). 30 min 5 Depth (cm) -140 The effect of breaking waves on the oil distribution (in -35 1 2 particular, the penetration depth of dispersed oil) is related to -70 3 a number of contributing factors. As waves break, it is esti- 4 -105 5 mated that 30 to 50% of the dissipated wave energy entrains

Depth (cm) 60 min -140 oil droplets into the water column (Lamarre and Melville, -35 1 1991; Tkalich and Chan, 2002), and effectively determines the 2 -70 3 ﬁrst-order oil entrainment rate (Tkalich and Chan, 2002). 4 -105 Breaking waves develop a mixing layer in the upper water 120 min 5 Depth (cm) -140 column, and the penetration of oil results in a uniform mixing 10 12 14 16 18 20 of the droplets, with the mixing layer proportional to the Distance (m) height of breaking waves (Delvigne and Sweeney, 1988; B Tkalich and Chan, 2002). Moreover, breaking waves generate -35 8 microscale turbulence with the smallest eddies having the 16 -70 24 greatest velocity gradients, leading to deformation, elonga- -105 32 tion, and eventual breakup of larger droplets, forming a large 40

Depth (cm) 5 min -140 number of small droplets that have lower buoyancy and more 2 rapid diffusion efﬁciency (Delvigne et al., 1987; Li and Garrett, -35 4 -70 6 1998). 8 -105 10 12 Depth (cm) 30 min -140 DE as a function of energy dissipation rate 2 -35 4 -70 6 DE was determined by estimating the average dispersed oil 8 -105 10 concentration in the water column for the two chemical dis-

Depth (cm) 12 -140 60 min persant treatments and the no-dispersant (physically dis- 2 persed) control as a function of e. This was performed by -35 4 calculating the average oil concentrations of the eight samples -70 6 8 recovered from the four horizontal locations and two lower -105 10

Depth (cm) 120 min 12 depths of the water column in the wave tank. It is reasonable -140 10 12 14 16 18 20 to exclude the surface samples from the calculation of the Distance (m) average dispersed oil concentration in the water column because the high oil concentrations on the surface of the water at C the wave absorber end of the tank were largely controlled by -35 5 10 surface drift, and would skew the calculations of DE. In- -70 15 20 tuitively, the extremely high oil concentration of the samples -105 5 min 25 Depth (cm) recovered from the surface in front of wave absorbers under -140 regular waves in the absence of dispersants is a clear indica- -35 2.5 5.0 tion of poor oil dispersion efﬁciency. -70 7.5 10.0 Figures 5 and 6 present the estimated DE of the three dis- -105 12.5 persant types (including water as the control) on the two Depth (cm) 30 min 15.0 -140 crude oils as a function of e. Different degrees of physical -35 2 4 dispersion of the MESA and ANS crude were measured at -70 6 each energy dissipation rate. The physical DE of MESA crude 8 -105 60 min ranged from 4 to 9% at 5 min (Fig. 5A), and then steadily Depth (cm) -140 increased with time, approaching 12 to 24% after 2 h (Fig. 5D). -35 3 Similarly, the physical DE of ANS crude was between 3.5 and 4 -70 5 9% at 5 min (Fig. 6A) and increased to between 10 and 19% at -105 120 min 6 2 h (Fig. 6D). Delvigne and Sweeney (1988) have reported that 7 Depth (cm) -140 the physical dispersion of oil in a grid column generated 10 12 14 16 18 20 droplets that were mostly larger than 50 mm in turbulence at Distance (m) energy dissipation rates of up to 3.5 W kg1, but they ob- FIG. 4. Dispersed oil concentration (mg L1) as a function served only a very small fraction of oil dispersed by breaking of time and space under plunging breaking waves: (A)no waves under a surface slick in their ﬂume experiments. Lunel dispersant; (B) with corexit 9500; (C) with SPC 1000. Color (1993, 1995) found that the dispersion of Forties crude oil with images available online at www.liebertonline.com=ees. and without dispersant applied at sea generated a similar

128 100 100 (A) 5 min (B) 30 min 80 80 water 60 corexit 60 spc

DE (%) DE 40 (%) DE 40

20 20

0 0 0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10

(m2/s3) (m2/s3)

100 100 (C) 60 min (D) 120 min 80 80

60 60

DE (%) DE 40 (%) DE 40

20 20

0 0 0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10

(m2/s3) (m2/s3)

FIG. 5. Dispersant effectiveness (DE) on MESA crude oil as a function of energy dissipation rate at: (A) 5 min, (B) 30 min, (C) 60 min, and (D) 120 min. Data shown are mean and one standard deviation of independent triplicate runs. Lines are best-ﬁt regressions of Equation (2).

100 100 (A) 5 min (B) 30 min 80 80 water 60 corexit 60 spc

DE (%) 40 DE (%) 40

20 20

0 0 0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10 (m2/s3) (m2/s3)

100 100 (C) 60 min (D) 120 min 80 80

60 60

DE (%) 40 DE (%) 40

20 20

0 0 0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10 (m2/s3) (m2/s3)

FIG. 6. Dispersant effectiveness (DE) on ANS crude oil as a function of energy dissipation rate at: (A) 5 min, (B) 30 min, (C) 60 min, and (D) 120 min. Data shown are mean and one standard deviation of independent triplicate runs. Lines are best- ﬁt regressions of Equation (2).

1145 129 1146 LI ET AL.

Table 1. Four-Way Factorial Analysis of Variance of the Main Factor and Multifactor Interaction Effects on the Dispersant Effectiveness

Degree of freedom Sum of square Mean square F value Pr (F)

Dispersant (Disp) 2 627.8180 313.9090 140.2928 0.0000000* Wave 2 163.0560 81.5280 36.4366 0.0000000* Time 3 100.6297 33.5432 14.9912 0.0000000* Oil 1 16.5033 16.5033 7.3757 0.0074208* Disp=oil 2 19.0702 9.5351 4.2614 0.0159212* Disp=wave 4 69.9299 17.4825 7.8133 0.0000100* Oil=wave 2 1.9648 0.9824 0.4391 0.6455039 Disp=time 6 45.2753 7.5459 3.3724 0.0038509* Oil=time 3 0.4255 0.1418 0.0634 0.9790765 Wave=time 6 7.1573 1.1929 0.5331 0.7824024 Disp=oil=wave 4 28.0072 7.0018 3.1293 0.0167294* Disp=oil=time 6 1.9656 0.3276 0.1464 0.9895068 Disp=wave=time 12 6.7733 0.5644 0.2523 0.9948054 Oil=wave=time 6 8.4322 1.4054 0.6281 0.7075827 Disp=wave=oil=time 12 5.4511 0.4543 0.2030 0.9981657 Residuals 144 322.2041 2.2375

*Signiﬁcant factors.

number of large oil droplets (> 70 mm), but dispersion of the was dispersed for 2 h at a high-energy dissipation rate. A oil with dispersant produced much greater number of small similar trend was observed for the dispersion of ANS crude. droplets (<70 mm). However, because none of these authors At all three energy dissipation rates, though, the DE of Corexit have reported physical DE explicitly, it is impossible to 9500 was consistently higher than that of SPC 1000, and both compare their results with our current work, beyond the were significantly higher than the control. The DE of the consistency that physical dispersion of oil was expected. It control was significantly increased as a function of time at a was even more difficult to obtain physical DE of oil when high-energy dissipation rate. The SPC 1000 was less effective bench-scale testing apparatus was used, due to coalescence at the two lower energy dissipation rates. In contrast, the and resurfacing of oil from the relatively large water to oil Corexit 9500 appeared equally effective at the two higher ratio and the wall effect of confined space (Chandrasekar et al. energy dissipation rates but less effective at the low-energy 2005, 2006). dissipation rate. Chemical dispersants were effective under all three wave To delineate the effect of each treatment factor on the oil conditions (Figs. 5 and 6), with oil dispersion most effective dispersion, a four-way factorial ANOVA was conducted under plunging breaking wave conditions and least effective to determine the effects of main factors and multifactor under regular nonbreaking wave conditions. The dispersant interactions on the DE. The main factors tested included dis- Corexit 9500 appeared twice as effective as SPC 1000 for the persant type (three levels), wave condition (three levels), oil dispersion of MESA crude at the two lower energy dissipation type (two levels), and oil dispersion time (four levels). The oil rates, but the DE of the two dispersants was closer after oil dispersion time (5, 30, 60, and 120 min) was analyzed sepa-

Table 2. Tukey’s Paired Comparison of the Different Effects Between Treatment Levels on Disperscent Effectiveness (%) in the Water Column

Treatment Level Estimates Standard error Lower limit Upper limit

Dispersant Corexit—SPC 21.6* 2.98 14.6 28.7 Corexit—Water 47.7* 2.98 40.6 54.7 SPC—Water 26.1* 2.98 19.0 33.0 Wave Plunging—Regular 24.3* 2.98 17.3 31.4 Plunging—Spilling 12.3* 2.98 5.5 19.5 Spilling—Regular 11.8* 2.98 4.7 18.9 Time 30 min–5 min 17.8* 3.29 9.3 26.4 60 min–5 min 18.5* 3.29 10.0 27.1 120 min–5 min 17.6* 3.29 9.1 26.2 60 min–30 min 0.73 3.29 7.8 9.3 120 min–30 min 0.19 3.29 8.7 8.4 120 min–60 min 0.92 3.29 9.5 7.6

*Significant differences, based on 95% simultaneous confidence intervals for specified linear combinations.

130 DISPERSANT EFFICACY AND ENERGY DISSIPATION RATE 1147 rately to identify the time required to achieve the maximum Author Disclosure Statement overall DE for the three dispersant conditions on the two oils The authors declare that no conflicting financial interests under the three imposed wave conditions. The results of the exist. ANOVA are summarized in Table 1. The ANOVA revealed that there was one significant three-factor interaction (dispersant by oil by wave, p ¼ 0.017) and three significant two- References factor interactions (dispersant by oil, p ¼ 0.016; dispersant by wave, p ¼ 0.00001; and dispersant by time, p ¼ 0.00385). In- Chandrasekar, S., Sorial, G.A., and Weaver, J.W. (2005). Dis- persant effectiveness on three oils under various simulated teractions of these factors indicate that they had opposite ef- environmental conditions. Environ. Eng. Sci. 22, 324. fects on the DE. There was no two-way time-by-wave Chandrasekar, S., Sorial, G.A., and Weaver, J.W. (2006). Dis- interaction. All four main factors showed very strong signif- persant effectiveness on oil spills—Impact of salinity. ICES icant effects on the DE (oil, p ¼ 0.0074; wave, dispersant type, J. Marine Sci. 63, 1418. and dispersion time, p < 0.0000001). Chapman, H., Purnell, K., Law, R.J., and Kirby, M.F. (2007). The The Tukey’s paired comparison method was used to use of chemical dispersants to combat oil spills at sea: A re- compare the effects of different levels of the main factors that view of practice and research needs in Europe. Marine Polluti. have been identified to significantly affect the DE; the results Bull. 54, 827. are listed in Table 2. The two tested dispersants both have Daling, P.S., Moldestad, M.O., Johansen, O., Lewis, A., and significantly higher DE than the water control (increasing DE Rodal, J. (2003). Norwegian testing of emulsion properties at by an average margin of 48 and 26%, respectively, for Corexit sea—The importance of oil type and release conditions. Spill 9500 and SPC 1000); Corexit 9500 was more effective than Sci. Technol. Bull. 8, 123. SPC 1000 by a 22% higher DE. As for the effect of the wave Delvigne, G.A.L., and Sweeney, C.E. (1988). Natural dispersion condition, spilling breaking waves significantly increased of oil. Oil Chem. Pollut. 4, 281. DE by 12% compared to the regular nonbreaking waves; Delvigne, G.A.L., Van del Stel, J.A., and Sweeney, C.E. (1987). plunging breaking waves significantly increased the DE by an Measurements of Vertical Turbulent Dispersion and Diffusion of Oil additional 13% compared to the spilling breaking waves. Droplets and Oil Particles. MMS 87-111. Anchorage, Alaska: U.S. With regard to dispersion time, the extent of oil dispersion Department of the Interior, Minerals Management Service. increased significantly from 5 min to 30 min, and leveled off Drennan, W.M., Donelan, M.A., Terray, E.A., and Katsaros, K.B. thereafter. (1996). Oceanic turbulence dissipation measurements in SWADE. J. Phys. Oceanogr. 26, 808. Conclusion Fingas, M.F. (2000). Use of surfactants for environmental applications. In L.L. Schramm, Ed. Surfactants: Fundamentals and This study demonstrated the quantitative relationship be- Applications to the Petroleum Industry, New York: Cambridge tween chemical DE and energy dissipation rate when the ef- University Press, p. 461. fects of two dispersants on two crude oils were evaluated Fingas, M., and Ka’aihue, L. (2004). Dispersant tank testing—A under three different wave conditions in an experimental review of procedures and considerations. Environment Canada wave tank. The mixing energies of these wave conditions Arctic and Marine Oil Spill Program Technical Seminar (AMOP) were close to those encountered in the field when oil spills are Proceedings, p. 1003. treated with chemical dispersants. The data reported here Funke, E.R., and Mansard, E.P. (1979). SPLSH A Program for the support the importance of breaking wave conditions in fa- Synthesis of Episodic Waves. Hydraulics Laboratory Technical cilitating chemical DE (Nilson et al., 1985; Delvigne and Report LTR-HY-65. Ottawa, Canada: National Research Council. Sweeney, 1988; Tkalich and Chan, 2002; Shaw, 2003; Li et al., Kaku, V.J., Boufadel, M.C., Venosa, A.D., and Weaver, J. (2006). 2008a). The oil-based dispersant, Corexit 9500, was more ef- Flow dynamics in eccentrically rotating flasks used for dis- fective than the water-soluble dispersant, SPC 1000, in dispersant effectiveness testing. Environ. Fluid Mech. 6, 385. persing the two crude oils under all three wave types and Kresta, S.M., and Wood, P.E. (1993). The flow field produced by a energy levels. More studies are needed to evaluate dilution pitched blade turbine: Characterization of the turbulence and effect on the performance of these dispersants. These results, estimation of the dissipation. Chem. Engi. Sci. 48, 1761. together with the oil droplet size distribution data (Li et al., Lamarre, E., and Melville, W.K. (1991). Air entrainment and 2008b), will have significant implications in developing better dissipation in breaking waves. Nature 351, 469. operational guidelines for dispersant use and improved pre- Lee, K., Stoffyn-Egli, P., and Owens, E.H. (2001). Natural dis- dictive models of dispersant effectiveness in the field. persion of oil in a freshwater ecosystem: Desaguadero Pipeline Spill, Bolivia. In Proceedings of the 2001 International Oil Spill Acknowledgments Conference. Publication No. 14710B. Washington, DC: Ameri- can Petroleum Institute, p. 1445. This research was funded by the Program of Energy Re- Lessard, R.R., and Demarco, G. (2000). The significance of oil search and Development (PERD), U.S. EPA (contract No. 68- spill dispersants. Spill Sci. Technol. Bull. 6, 59. C-00-159), and NOAA=UNH Coastal Response Research Li, M., and Garrett, C. (1998). The relationship between oil droplet Center (NOAA Grant Number: NA04NOS4190063 UNH size and upper ocean turbulence. Marine Pollut. Bull. 36, 961. Agreement No. 06-085). The authors wish to thank Susan Li, Z., Lee, K., King, T., Boufadel, M.C., and Venosa, A.D. Cobanli, Xiaowei Ma, Brian Robinson, Peter Thamers, and (2008a). Assessment of chemical dispersant effectiveness in a William Yeung for their contributions and logistic support. wave tank under regular non-breaking and breaking wave The findings, opinions, and recommendations expressed in conditions. Marine Pollut. Bull. 56, 903. this report are those of the authors and do not necessarily Li, Z., Lee, K., King, T., Boufadel, M.C., and Venosa, A.D. reflect those of the sponsors. (2008b). Oil droplet size distribution as a function of energy

131 1148 LI ET AL.

dissipation rate in an experimental wave tank, 2008 Interna- Shaw, J. M. (2003). A microscopic view of oil slick break-up and tional Oil Spill Conference, Savannah, GA. emulsion formation in breaking waves. Spill Sci. Technol. Bull. Li, Z., Lee, K., King, T., Boufadel, M.C., and Venosa, A.D. 8, 491. (submitted). Oil spill chemical dispersion efficacy in an Sorial, G.A., Venosa, A.D., Koran, K.M., Holder, E., and King, D.W. experimental wave tank: 2, significant factors determining in- (2004). Oil spill dispersant effectiveness protocol. II: Perfor- situ oil droplet size distribution Environ. Eng. Sci. mance of revised protocol. J. Environ. Eng. ASCE 130, 1085. Lunel, T. (1993). Dispersion: Oil droplet size measurement at sea. Terray, E.A., Donelan, M.A., Agrawal, Y.C., Drennan, W.M., Environment Canada Arctic and Marine Oil Spill Program Kahma, K.K., Williams, A.J., Hwang, P.A., and Kitaigorodskii, (AMOP) Technical Seminar Proceedings, p. 1023. S.A. (1996). Estimates of kinetic energy dissipation under Lunel, T. (1995). Understanding the mechanism of dispersion breaking waves. J. Phys. Oceanogr. 26, 792. through oil droplet size measurements at sea. In P. Lane, Ed., Tkalich, P., and Chan, E.S. (2002). Vertical mixing of oil droplets The Use of Chemicals in Oil Spill Response, ASTM STP 1252, by breaking waves. Marine Pollut. Bull. 44, 1219. American Society for Testing and Materials, Philadelphia, PA, U.S. EPA (U.S. Environmental Protection Agency) (1996). Swir- p. 240. ling flask dispersant effectiveness test. Title 40 code of feral Nilson, J., Naess, A., and Volent, Z. (1985). Measurements of Oil regulations, part 300, Appendix C, Naragansett, R.I., 245–250. Concentrations in the Water Column under Breaking Waves. Re- Venosa, A.D., King, D.W., and Sorial, G.A. (2002). The baffled port STF 60 A 85079. Trondheim, Norway: Norwegian Hy- flask test for dispersant effectiveness: A round robin evalua- drotechnical Laboratory. tion of reproducibility and repeatability. Spill Sci. Technol. Bull. NRC. (1989). National Research Council: Using Oil Spill Dispersant 7, 299. on the Sea. Washington, DC: National Academy Press. Wickley-Olsen, E., Boufadel, M.C., King, T., Li, Z., Lee, K., and NRC. (2005). National Research Council: Understanding Oil Spill Venosa, A.D. (2008). Regular and breaking waves in a wave Dispersants: Efficacy and Effects. Washington, DC: National tank for dispersion effectiveness testing. In 2008 International Academies Press. Oil Spill Conference, Savannah, GA.

132 This article has been cited by:

1. Zhengkai Li , Kenneth Lee , Thomas King , Michel C. Boufadel , Albert D. Venosa . 2009. Evaluating Chemical Dispersant Efficacy in an Experimental Wave Tank: 2—Significant Factors Determining In Situ Oil Droplet Size DistributionEvaluating Chemical Dispersant Efficacy in an Experimental Wave Tank: 2—Significant Factors Determining In Situ Oil Droplet Size Distribution. Environmental Engineering Science 26:9, 1407-1418. [Abstract] [PDF] [PDF Plus]

133 ENVIRONMENTAL ENGINEERING SCIENCE ORIGINAL ARTICLE Volume 26, Number 9, 2009 ª Mary Ann Liebert, Inc. DOI: 10.1089=ees.2008.0408

Evaluating Chemical Dispersant Efﬁcacy in an Experimental Wave Tank: 2—Signiﬁcant Factors Determining In Situ Oil Droplet Size Distribution

Zhengkai Li,1,* Kenneth Lee,1 Thomas King,1 Michel C. Boufadel,2 and Albert D. Venosa3

1Centre for Offshore Oil and Gas Environmental Research, Bedford Institute of Oceanography, Fisheries and Oceans Canada, Dartmouth, Canada. 2Department of Civil and Environmental Engineering, Temple University, Philadelphia, Pennsylvania. 3National Risk Management Research Laboratory, U.S. EPA, Cincinnati, Ohio.

Received: December 18, 2008 Accepted in revised form: June 15, 2009

Abstract Chemical dispersion is one of the most cost-effective options to remediate oil spill at open sea. Identifying signiﬁcant factors that determine in situ droplet size distributions facilitates mechanistic understanding of dispersant effectiveness. In this work, in situ dispersed oil droplet size distributions were characterized during testing of chemical dispersant effectiveness of two dispersants (Corexit 9500 and SPC 1000) on two oils [Medium South American (MESA) and Alaska North Slope (ANS)] under three wave conditions (regular nonbreaking, spilling breaking, and plunging breaking waves) in an experimental wave tank. Results showed that physical dispersion generated monomodal lognormal oil droplet size distributions of larger median diameters, whereas chemical dispersion produced bi- or trimodal lognormal oil droplet size distributions of smaller median diameters over a wider range. Factorial analysis of variance (ANOVA) followed by Tukey’s paired comparison statistical data analysis indicated that the volume mean diameters of dispersed oil droplets were reduced by 36 mm (from 122 to 86 mm) by plunging breaking conditions. Volume mean diameters were decreased by 92 mm (from 153 to 61 mm) and 37 mm (from 153 to 116 mm), respectively, by Corexit 9500 and SPC 1000. These results are useful in optimizing operational guidelines for dispersant use, and providing input for modeling transport, fate, and biological effects of dispersed oil.

Key words: oil spill; chemical dispersant; droplet size distribution; breaking waves; wave energy

Introduction from 10 m in France to 60 m in Malta (Chapman et al., 2007). In the United States, many states have defined preapproval hemical dispersants are used to reduce interfacial zones for dispersant use outside three nautical miles from Ctension between oil and water, so as to enhance the shore and=or in depths greater than 10 m (NRC, 2005). natural process of dispersion by generating larger numbers of In dispersant effectiveness testing, it is important for the small droplets of oil that are entrained into the water column testing system to have hydrodynamics and dilution [com- by wave energy (Lessard and Demarco, 2000; NRC, 2005; monly expressed as oil-to-water ratio (OWR)] approach field Chapman et al., 2007; Kirby and Law, 2008). The premise for conditions (Fingas et al., 1989; NRC, 2005). Fingas et al. (1989) the use of dispersants as a cost-effective oil spill counter- showed that the dispersant effectiveness values were rela- measure is the presence of large volumes of seawater that can tively constant over the OWR ranging from 1:1,000 to render rapid dilution of dispersed oil to concentrations lower 1:120,000, but had a large variation when the ratio was 1:500 than the toxicity threshold of biota. Presently, dispersant-use or smaller. In that study, however, the effect of OWR, in policies vary from region to region, but the general consensus the range from 1:4,000 to 1:120,000, was evaluated in a is that sufficient dilution capacity is essential. In Europe, recirculation-flow cylinder, which had no control over resur- minimum permitted depths for dispersant spraying range facing and=or settling of the dispersed oil. NRC (2005) reit- erated the importance of OWR in the dispersant effectiveness testing. For example, when OWR is low, anionic and nonionic *Corresponding author: Centre for Offshore Oil and Gas Envir- surfactants with a high hydrophilic–lipophilic balance tend to onmental Research, Bedford Institute of Oceanography, Fisheries and Oceans Canada, 1 Challenger Drive, Dartmouth, NS B2Y 4A2, partition into the aqueous phase, where they cannot effec- Canada. Phone: 902-426-3442; Fax: 902-426-1440; E-mail: liz@dfo- tively promote formation of small oil droplets. When the OWR mpo.gc.ca is high, however, more of the surfactant can be associated

1407 134 1408 LI ET AL. with the oil phase where it can facilitate droplet formation. small droplets in the water column was reported in laboratory Conversely, high OWR could reduce the observed dispersion tests (Jasper et al., 1978; Byford et al., 1984; Lewis et al., 1985) efficiency by increasing the rate of droplet coalescence, which and field trials (Lunel, 1993, 1995). Jasper et al. (1978) observed is proportional to the number concentration of oil droplets that the volume mean diameter of dispersed oil was reduced (NRC, 1989). Droplet coalescence will produce larger oil by 30–40% by the presence of a chemical dispersant. Lunel droplets that can resurface more quickly and reduce the mass (1995) reported that dispersants increased the number of of oil entrained in the aqueous phase. small droplets (<50 mm) by 5- to 30-fold in a sea trial, but the Due to the constraints of volume, laboratory tests have number of larger droplets (>50 mm) between the dispersant- limited capacity to accommodate transport and dilution ef- treated and untreated oil were the same. fects. For a typical bench-scale laboratory dispersant effec- The NRC (2005) report emphasized that existing databases tiveness test (such as Fingas et al., 1995; U.S. EPA, 1996; ASTM, for the droplet-size distributions of physically and chemically 2002), 120 mL of seawater in a testing flask is spiked with dispersed oil must be expanded to determine whether and 0.1 mL testing oil, resulting an OWR of 1:1,200. Further re- how factors such as energy dissipation rate, oil type, disper- ducing the dosage of oil can jeopardize the accuracy of the sant characteristics, and dispersant use influence the droplet amount of oil that can be measured. As a result, dispersant size distribution and formation kinetics under hydrodynamic effectiveness measured as a percentage oil dispersed into the conditions approaching those existing in the field. To this end, water column in a bench-scale test is sensitive to the settling the BIO wave tank was used to characterize and compare time that precedes collection of samples, largely due to coa- in situ droplet size distributions of physically and chemically lescence and resurfacing of oil droplets promoted by high dispersed oil under a variety of nonbreaking and breaking OWR (Fingas et al., 2002; Fingas, 2005). Conversely, expensive wave conditions. The data generated in this study will be and logistically challenging sea trials often lead to results that useful in optimizing operational guidelines, modeling trans- are inconclusive due to limitations in the level of replication port and fate, and potentially evaluating biological effects of and control of experimental variables. chemically dispersed oil. This is the second of a series of two Wave tank testing allows for evaluating dispersant effec- articles; the first article focuses on elucidating dispersant tiveness under more realistic (i.e., field) conditions, including effectiveness as a function of energy dissipation rate (Li et al., the mixing energy generated by various wave types, and the accepted). dilution effect (OWR) that is similar to what is observed in the field. The misleading role of coalescence in small systems may Materials and Methods be significantly reduced in large wave tanks where dilution Experimental procedures more closely approximates natural conditions. In response to an identified need for testing the performance of chemical Figure 1 shows the top and side view schematic represen- dispersants under more realistic oceanographic and envi- tations of the wave tank facility with sampling locations at ronmental conditions including wave-induced mixing ener- BIO. The tank measures 32 m long, 0.6 m wide, and 2 m high, gies (NRC, 2005), a wave tank facility has been constructed at with an average water depth of 1.5 m. Three different wave the Bedford Institute of Oceanography (BIO, Dartmouth, conditions were generated by a computer-controlled flap- NS, Canada) for reproducible, quantitative evaluation of type wave maker situated at one end of the tank. Table 1 chemical dispersant effectiveness under a variety of wave summarizes the conditions that were used to create the vari- conditions. ous wave types. The energy dissipation rate per unit mass of To understand the intrinsic mechanisms of dispersant ef- water (e) was evaluated by the autocorrelation function fectiveness obtained in a testing system, it is important to method (Kresta and Wood, 1993) using time-series velocity measure the dispersed oil droplet size distributions and measurements obtained by an Acoustic Doppler Velocimeter compare the data with those observed at sea (Lunel, 1995). (SonTec=YSI, Inc. San Diego, CA) at select locations in the In situ dispersed oil droplet size distributions are controlled tank. Details of the wave generation and the hydrodynamic by a variety of hydrodynamic and environmental variables. characterization of the wave tank have been reported else- For example, the intensity of the turbulent mixing energy where (Wickley-Olsen et al., 2008; Li et al., accepted). The dictates the breakup of large oil droplets into smaller droplets dispersant effectiveness of two dispersants (Corexit 9500 and and the depth of submergence of the droplets. Breaking SPC 1000) on two oils [Medium South American (MESA) and waves, in particular, have been documented to play a crucial Alaska North Slope (ANS)] under the above three wave role in dispersion of oil slicks by generating velocity shear to conditions was assessed using a three-factor, mixed-level break up and transport oil in turbulent flow (Li and Garrett. factorial experiment with three replications. The design was a 1998; Tkalich and Chan, 2002; Shaw, 2003). The effects of noise-reducing randomized block design (Mendenhall et al., mixing energy on dispersed oil droplet size distributions, and 1981) to minimize the impacts of confounding factors such as consequently dispersant effectiveness, have been reported in temperature, salinity, and wind variation. For each experi- laboratory tests (Byford et al., 1984; Lewis et al., 1985; Sorial ment, seawater was pumped from the Bedford Basin (NS, et al., 2004; Chandrasekar et al., 2005; Ma et al., 2008) and in Canada) through a double layer sock-filter (Atlantic Pur- field trials under low- and high-energy regimes caused by ification Ltd, Dartmouth, NS, Canada) with pore sizes of 25 various wind effects (Lunel, 1993, 1995). The droplet size dis- and 5 mm for the coarse and fine filters, respectively. The tributions are also affected by the collision frequency, which background temperature, salinity, and particle size distribu- has been considered a function of system hydrodynamics, and tions were recorded before each experiment. The seawa- collision efficiency, which is generally believed to represent ter temperature and salinity were recorded as 16.3 1.88C, the chemistry involved in the coalescence reactions. The in- and 26.7 3.2 ppt, respectively, during the experimental fluence of chemical dispersants in increasing the number of period.

135 DISPERSANT EFFICACY AND DROPLET SIZE DISTRIBUTION 1409

Flap-type wave maker Initial oil slick Wave absorbers 60 30

125 800 200 200 400 200 200

3200

Flap-type wave maker Wave absorbers 5 70 200 65

125 800 200 200 400 200 200

3200

Locations for water samplers Locations for laser particle counter

FIG. 1. Top (upper) and side view (lower) of the schematic representation (all dimensions in centimeters, not to scale) of the Bedford Institute of Oceanography (BIO) wave tank facility.

To start an experiment, 300 mL of crude oil was gently 8, 12, 16, and 20 m downstream from the wave maker. The poured onto the water surface within a 40 cm (inner diameter) three syringes at each station collected water samples at three ring (constructed of NSF-51 reinforced clear, flexible PVC depths (5, 75, and 140 cm below the average water surface) in tube) located 10 m from the wave-maker. The volume of oil in the water column of the wave tank at four time points (5, 30, each experiment was intended to achieve an OWR (1:100,000) 60, and 120 min after the initial mixing). The total oil con- and a nominal oil concentration (10 mg=L) close to the field centrations of these samples were determined by extracting (ITOPF, 2005). Subsequently, 12 mL of dispersant (or water the total petroleum hydrocarbons (TPH) from the water for the control) was sprayed onto the oil slick through a samples with dichloromethane and then analyzing the TPH pressurized nozzle (60 psi; 0.635 mm i.d.), resulting in a dis- concentrations in a UV-Visible spectrophotometer as de- persant-to-oil ratio (DOR) of 1:25, a typical dosage as re- scribed elsewhere (Venosa et al., 2002; Li et al., 2008). commended for field application (NRC, 2005). The ring was then lifted immediately prior to the approaching first wave. In situ droplet size distribution The specific wave condition selected for each experimental run was maintained continuously during the 2-h experiment. The dispersed oil droplet size distribution was measured To measure the total dispersed oil concentration, water sam- using an in situ laser light scattering and transmissometer ples were collected through a set of three 100-mL syringes (LISST-100X, Type C; Sequoia Scientific, Seattle, WA). The connected to each of four stainless steel manifolds located at LISST-100X records 32 particle size intervals logarithmically

Table 1. Conditions Used to Generate Various Wave Types in the Wave Tank

Wave type Stroke (cm) Frequency (Hz) Wavelength (m) Height (cm) Duration (s)

Regular 12 0.80 2.44 23 — Spiller 8.4 0.85 2.16 18 20 8.4 0.48 6.77 6 5 Plunger 12 0.85 2.16 26 20 12 0.50 6.24 9 5

136 1410 LI ET AL. placed from 2.5–500 mm in diameter, with the upper size in Data analysis each bin 1.18 times the lower. The measured particle size dis- A five-way factorial analysis of variance (ANOVA) was tribution is expressed as the average volumetric concentration performed to test the effects of factors, including dispersants, of oil within each size interval. The droplet size distributions wave conditions, oil type, dispersion time, and sampling were measured at three different depths (45, 80, and 125 cm depths, on the average VMD of the dispersed oil droplets. from the average water surface) at one horizontal location When significant effects were identified, the Tukey’s paired (18 m downstream from the wave-maker) over four half-an- comparison was conducted to isolate the factorial effects at hour time periods (0–30, 30–60, 60–90, 90–120 min). For each each treatment level. Data analyses were conducted using a depth at each time period, four 10-min continuous records of statistical data analysis package S-PLUS 7.0 (Insightful Inc., droplet size distribution data were obtained every 3 s. Seattle, WA). A significance alpha level of 0.05 was adopted To compare the effects of dispersant application, wave for statistical tests. A Bonferroni adjustment was performed conditions, and other factors on dispersed oil droplet size, the on the alpha level to control for the family-wise alpha rate to volume mean diameters (VMD) of the measured droplet size reach an adjusted and more strict alpha level at 0.0016. distribution was calculated as a weighted mean according to its individual volume contribution to the total volume of the Results droplets: P Physically and chemically dispersed (C d ) oil droplet size distributions VMD ¼ P i i (1) Ci Figures 2 to 4 display representative dispersed MESA crude oil droplet size distributions measured near the surface of the where Ci (i ¼ 1–32) is the volume concentration of oil droplets wave tank for physical and chemical dispersion of oil under in a size interval with average size di, which is the geometric three different wave conditions. Similar droplet size distri- mean of the lower and upper limit of every size range within butions were obtained in the middle and near the bottom of the range 2.5–500 mm. the wave tank (data not shown). The dispersed oil droplet size

2.00 100 (A) (B) 10 min 1.50 75 40 min 70 min 1.00 100 min 50

0.50 25

0.00 0 µ l/L) 1.00 100 (C) (D)

0.75 75

0.50 50

0.25 25 Cumulative fraction (%) 0.00 0

Volumetric particle sizedistribution ( 1.00 100 (E) (F)

0.75 75

0.50 50

0.25 25

0.00 0 10 100 10 100 Diameter (µm) Diameter (µm)

FIG. 2. Volumetric (left) and cumulative (right) Medium South American (MESA) oil droplet size distributions dispersed by water (A, B), Corexit 9500 (C, D), and SPC 1000 (E, F) under regular wave conditions.

137 DISPERSANT EFFICACY AND DROPLET SIZE DISTRIBUTION 1411

4.0 100 (A) (B)

3.0 10 min 75 40 min 2.0 70 min 50 100 min

1.0 25

0.0 0 µ l/L) 100 (C) (D) 12.0 75

8.0 50

4.0 25 Cumulative fraction (%) 0.0 0

Volumetric particle size distribution ( 100 (E) (F) 12.0 75

8.0 50

4.0 25

0.0 0 10 100 10 100 Diameter (µm) Diameter (µm)

FIG. 3. Volumetric (left) and cumulative (right) MESA oil droplet size distributions dispersed by water (A, B), Corexit 9500 (C, D), and SPC 1000 (E, F) under spilling breaking wave conditions. distribution data were also recorded for ANS crude from all 2–4). However, the droplet size distribution patterns of the three depths under different experimental conditions (data chemically dispersed oil droplets were essentially the same not shown). These data were similar to those for MESA. after 30 min under all three wave conditions, as indicated by Tables 2 and 3 summarize the droplet size distribution sta- the percentage cumulative droplet size distributions (Fig. 2D tistics for MESA and ANS oil, respectively, at the surface. and F, 3D and F, and 4D and F). Physical dispersion (i.e., dispersion in absence of a chemical dispersant) of MESA oil created monomodal lognormal drop- Significant factors determining the average let size distributions under regular wave conditions through- dispersed oil droplet sizes out the entire experiment (Fig. 2A and B). Bi- or trimodal lognormal distributions were generated initially (first 10 min), The full spectrum of particle size distribution at each which were further dispersed to monomodal distributions, sampling time was converted to VMD to compare the influ- under spilling and plunging breakers (Fig. 3A and B and ence of different treatment conditions, including wave con- 4A and B). In the presence of chemical dispersants, how- ditions, dispersant type, and oil type, on the average droplet ever, multimodal lognormal size distributions were produced size. The VMD from physical dispersion (absence of disper- under all three wave conditions throughout the entire dura- sants) under all wave conditions started with wide fluctuation of the dispersion experiment (Fig. 2C and E–4C and E). A tions but generally decreased over time. The time-series VMD large number of oil droplets with less than 10 mm in size were were in good agreement at all three depths under regular created in the presence of chemical dispersants, especially by waves and plunging breaking waves, but were smaller at the Corexit 9500. Application of dispersants expanded the range bottom than in the middle and near the surface under spilling of size distributions as indicated by larger geometric stan- breaking wave conditions, probably caused by less penetra- dard deviations (GSD) associated with chemical dispersants tion depth of the spilling breaking waves. The VMD from the (Tables 2 and 3). Chemical dispersants also caused much chemical dispersants were larger near the bottom than near higher dispersed phase volume concentrations (areas under the surface and in the middle of the wave tank under regular the droplet size distributions) than physical dispersion (Figs. waves, but had wider fluctuations in the middle of the wave

138 1412 LI ET AL.

2.0 100 (A) (B) µ l/L)

1.5 75 10 min 40 min 1.0 70 min 50 100 min 0.5 25 Cumulative fraction (%) Cumulative fraction

Particle size distribution ( 0.0 0 10 100 10 100 Diameter (µm) Diameter (µm)

4.0 100

µ l/L) (C) (D)

3.0 75

2.0 50

1.0 25 Cumulative fraction (%) fraction Cumulative

Particle size distribution ( 0.0 0 10 100 10 100 Diameter (µm) Diameter (µm)

4.0 100 (E) (F) µ l/L)

3.0 75

2.0 50

1.0 25 Cumulative fraction (%)

Particle size distribution ( Particle size 0.0 0 10 100 10 100 Diameter (µm) Diameter (µm)

FIG. 4. Volumetric (left) and cumulative (right) MESA oil droplet size distributions dispersed by water (A, B), Corexit 9500 (C, D), and SPC 1000 (E, F) under plunging breaking wave conditions.

tank than near the surface and the bottom of the wave tank dispersant type and wave conditions. Physical dispersion of under spilling and plunging breaking waves conditions. both crude oils appears to have required the longest time In Figures 5 and 6 the average dispersed oil droplet VMD at (100 min) to reach the ultimate stable VMD under regular the surface of the wave tank are illustrated as a function of nonbreaking wave conditions and the shortest time (40 min) time at the end of each measurement period for MESA and under plunging breaking wave conditions. As expected, ANS crude, respectively. Similar trends of the time-series chemical dispersants facilitated better dispersion at lower average VMD under different dispersant and wave condi- energy dissipation rates under regular nonbreaking and tions were observed at the other two depths (in the middle spilling breaking wave conditions. Corexit 9500 reduced the and near the bottom) of the wave tank (data not shown). average VMD for both MESA and ANS under all three wave After 10 min, physically dispersed oil droplets were large conditions at all four time points (Fig. 5A–C). SPC1000, (VMD > 160 mm) under all three wave conditions (Figs. 5 however, decreased the average VMD markedly under spill- and 6). In contrast, chemical dispersants (both Corexit 9500 ing and plunging breaking waves for MESA, but reduced the and SPC 1000) created the largest oil droplets (VMDMESA > average VMD even more dramatically under regular wave 140 mm, VMDANS > 200 mm) under regular nonbreaking and spilling breaking waves for ANS. The effect of dispersant waves and the smallest droplets (VMDMESA < 80 mm was clearly illustrated by the ultimate VMD for each combi- VMDANS < 120 mm) under plunging breaking waves, with the nation of dispersant, wave, and oil. Physically dispersed average VMD more dependent on dispersant and oil type MESA and ANS oil droplets had VMD > 100 mm under all under spilling breaking wave conditions. As oil dispersion three wave conditions. SPC 1000 dispersed MESA to 150 mm progressed, the average VMD declined at a rate dependent on under regular nonbreaking waves and less than 70 mm under

139 DISPERSANT EFFICACY AND DROPLET SIZE DISTRIBUTION 1413

Table 2. Droplet Size Distribution Statistics for Medium South American at the Surface

a a a Wave Dispersant Time (min) d16% (mm) d50% (mm) d84% (mm) GSD Modes % < 70 (mm)

Regular Water 10 53 104 143 1.64 1 22% 40 46 80 116 1.59 1 34% 70 45 80 114 1.59 1 35% 100 28 55 74 1.63 1 75% Corexit 10 8 33 250 5.59 4 62% 40 6 19 55 3.03 2 90% 70 3.5 16 45 3.59 2 94% 100 3 14 38 3.56 2 96% SPC 10 11 45 186 4.11 3 65% 40 15 50 120 2.83 3 65% 70 13 39 100 2.77 3 72% 100 12 35 90 2.74 3 77% Spilling Water 10 230 280 360 1.25 2 0% 40 85 310 380 2.11 2 13% 70 29 50 75 1.61 1 78% 100 25 48 80 1.79 1 75% Corexit 10 160 350 430 1.64 1 95% 40 2.5 3 25 3.16 2 99% 70 2.5 5 25 3.16 2 99% 100 2.5 7 25 3.16 2 99% SPC 10 9 28 74 2.87 2 82% 40 10 27 58 2.41 2 90% 70 10 25 50 2.24 2 93% 100 10.5 24 50 2.18 2 93% Plunging Water 10 225 330 430 1.38 4 5% 40 45 70 96 1.46 2 45% 70 31 50 70 1.50 2 82% 100 29 48 68 1.53 2 92% Corexit 10 2.8 22 70 5.00 2 88% 40 3 19 43 3.79 2 100% 70 3 20 52 4.16 2 98% 100 3 18 53 4.20 2 98% SPC 10 11 64 170 3.93 2 50% 40 8 30 80 3.16 2 77% 70 7 24 70 3.16 2 84% 100 4 18 52 3.61 2 92%

a d16%, d50%, d84% represent the 16, 50, or 84% of total mass of the droplets that are smaller than this diameter (mm); d50% is the mass median diameter. GSD, geometric standard deviation. spilling and plunging breaking wave conditions, but it dis- p ¼ 0.027), indicating that average dispersed oil droplet size persed ANS crude to around 100 mm under all three wave was affected nonuniformly by changes in the interacting conditions. Corexit 9500, however, dispersed both MESA and variables. Besides the significant multifactor interactions, ANS to small sizes (<70 mm) under all three wave conditions. three of the five main factors (wave type, dispersant type, A five-way ANOVA was performed to test the factorial and dispersion time, p < 0.0001) were identified to have effects on the average VMD by: (1) wave condition, (2) dis- strong, statistically significant effects on the average dispersant type, (3) oil type, (4) dispersion time, and (5) mea- persed oil droplet sizes. Two other tested main factors did suring depth. Among these factors, testing of dispersion not affect average droplet sizes significantly: oil ( p ¼ 0.917) time allows for the identification of a minimum duration and sampling depth ( p ¼ 0.865). The significant effects of that is required for dispersion of oil into relatively stable various factors were further compared at each treatment average droplet sizes. The effect of water depth was tested level with the Tukey’s paired comparison test, and the re- to evaluate the spatial variation in the dispersed droplet sults are summarized in Table 5. The two dispersants sig- sizes. The ANOVA results are presented in Table 4. As ex- nificantly reduced the dispersed oil droplet size, with the pected, the high-order interactions (all four- and five-factor average VMD being reduced by 91.5 mm by Corexit 9500 and and all but one three-factor interactions) are insignificant 36.6 mm by SPC 1000. The plunging wave conditions signif- ( p > 0.25). There was one significant three-factor interac- icantly reduced the average dispersed droplet size by tion (dispersantoilwave, p < 0.0001), two significant two- 35.7 mm. The average dispersed oil droplet sizes declined factor interactions at Bonferroni-adjusted level (dispersant significantly by 42 mm between the first and second disper- wave, p ¼ 0.0003; and dispersantoil, p ¼ 0.0001), and two sion period. The differences in the average dispersed oil additional significant two-factor interactions at a less strict droplet sizes after 30 min were statistically insignificant significant level (depthtime, p ¼ 0.041; and oilwave, ( p > 0.05).

140 1414 LI ET AL.

Table 3. Droplet Size Distribution Statistics for Alaska North Slope at the Surface

a a a Wave Dispersant Time (min) d16% (mm) d50% (mm) d84% (mm) GSD Modes % < 70 (mm)

Regular Water 10 219 390 420 1.38 2 7% 40 70 250 400 2.39 1 15% 70 46 150 340 2.72 1 25% 100 46 140 300 2.55 1 27% Corexit 10 2.5 27 400 12.65 3 55% 40 2.5 3 19 2.76 2 99% 70 2.5 8 36 3.79 2 99% 100 2.5 4 8 1.79 2 99% SPC 10 10 36 85 2.92 2 78% 40 14 42 80 2.39 2 75% 70 12 36 70 2.42 2 85% 100 13 50 74 2.39 2 80% Spilling Water 10 280 310 370 1.15 1 0% 40 200 210 220 1.05 2 0% 70 110 295 300 1.65 2 4% 100 35 38 43 1.11 1 98% Corexit 10 2.5 19 125 7.07 3 5% 40 2.5 11 38 3.90 2 95% 70 2.5 11 38 3.90 2 96% 100 2.5 10 25 3.16 2 99% SPC 10 11 36 74 2.59 2 80% 40 13 36 60 2.15 2 95% 70 10 40 45 2.12 2 99% 100 9 22 36 2.00 2 100% Plunging Water 10 60 120 170 1.68 4 20% 40 50 85 120 1.55 2 35% 70 40 70 100 1.58 2 45% 100 36 65 96 1.63 2 62% Corexit 10 3 33 190 7.96 2 66% 40 2.5 12 25 3.16 2 100% 70 2.5 15 32 3.58 2 100% 100 2.5 11 22 2.97 2 100% SPC 10 10 25 400 6.32 2 81% 40 7 20 35 2.24 2 98% 70 4 18 35 2.96 2 99% 100 3 16 34 3.37 2 100%

a d16%, d50%, d84% represent the 16, 50, or 84% of total mass of the droplets that are smaller than this diameter (mm); d50% is the mass median diameter.

Discussion droplet size (Figs. 5 and 6 and Tables 2 and 3). High-energy dissipation rates associated with plunging breaking waves The effect of wave-generated mixing energy on the dis- reduced the droplet size distributions and increased the per- persed oil droplet size distribution was examined under three centage fraction of oil dispersed in the water column (Li et al., wave conditions with different energy dissipation rates. The accepted). In a closed system with little dilution potential, average energy dissipation rates were estimated to be ap- however, coalescence could be promoted by providing mix- proximately 0.005, 0.1, and 1 W kg1 near the surface at the ing energy over a prolonged period of time (NRC, 2005). mixing zone for regular nonbreaking waves, spilling breaking Sterling et al. (2004) has clearly demonstrated that increased waves, and plunging breaking waves, respectively. The en- mixing energy in an enclosed vessel promoted coalescence ergy dissipation rates declined exponentially to approxima- and resurfacing of dispersed oil droplets over time at a tely 0.001 W kg1 for the two breaking wave conditions at a nominal OWR of 1:6,000, resulting in an incrementally re- depth of 30 cm and decreased linearly under regular non- duced dispersed oil fraction in the aqueous phase. breaking wave conditions (Wickley-Olsen et al., 2008). The Application of chemical dispersants substantially altered energy dissipation rate of plunging breaking waves was the dispersed oil droplet size distributions in the wave tank by similar to those measured for breaking waves in the field creating wider range multimodal size distributions including (Drennan et al., 1996; Terray et al., 1996; Delvigne and Swee- large number of small droplets (<10 mm in diameter) (Figs. ney, 1988). Regular nonbreaking wave energy was also sim- 2–4 and Tables 2 and 3) and significantly reduced the average ilar to that reported on the sea surface by Delvigne and VMD (Figs. 5 and 6). A significant difference in the average Sweeney (1988), and energy level of spilling breakers was in dispersed droplet sizes was identified between the two tested between. These wave conditions had significant effects on the chemical dispersants. The apparent superiority of the oil- droplet size distributions (Figs. 2–4 and Table 1) and average based dispersant (Corexit 9500) is probably due to its stronger

141 DISPERSANT EFFICACY AND DROPLET SIZE DISTRIBUTION 1415

500 400 (A) (A) Water Water 400 SPC 300 SPC Corexit Corexit 300 200 200

100 100

0 0 (B) (B)

400 µ m) 300

300 200 200

100 100 Mass mean diameter ( µ m) Volume mean diameter ( Volume

0 0 (C) (C)

400 300

300

200 200

100 100

0 0 0306090 0306090 Time (min) Time (min)

FIG. 5. Dispersed MESA oil droplet size (volume mean FIG. 6. Dispersed Alaska North Slope (ANS) oil droplet diameter) as a function of time under: (A) regular non- size (volume mean diameter) as a function of time under: (A) breaking, (B) spilling breaking, and (C) plunging breaking regular nonbreaking, (B) spilling breaking, and (C) plunging wave conditions. breaking wave conditions. affiliation with oil in the course of oil dispersion, whereas the distributions within a short period (*30 min), whereas water-based dispersant (SPC 1000) tends to be washed away physical dispersion took much longer to reach the ultimate from the surface of oil droplets and partitioned into the droplet size distributions (Figs. 2–6 and Tables 2 and 3). In- aqueous phase over a longer period of time, resulting in re- deed, the oil dispersion kinetic rate coefficients of chemical coalescence and resurfacing of the dispersed oil droplets in dispersion are several-fold higher than physical dispersion less vigorously mixed areas. The rearrangement of surfactant- (data not shown). In addition, the length of time required to stabilized oil-in-water emulsions has been reported and was disperse oil into relatively stable droplet size distributions, explained as a result of the development of a surfactant either in the presence or absence of chemical dispersants, is depletion–flocculation process (Sanchez et al., 2001). dependent on the system hydrodynamics, especially the in- Not surprisingly, time is a significant factor in dispersion of tensity and frequency of occurrence of breaking waves. This is crude oils in the wave tank. In these experiments chemical also related to the droplet population dynamics of the dis- dispersion broke up oil into relatively stable droplet size persed oil as a balance of breakup and coalescence kinetics.

142 1416 LI ET AL.

Table 4. Factorial Analysis of Variance of the Effects on the Average Water Column Dispersed Oil Droplet Sizes

Df Sum of Sq Mean Sq F Value Pr (F)

Dispersant 2 915,915 45,7957.5 118.6329 <0.0001*** Oil 1 42 42.0 0.0109 0.9169 Wave 2 185,989 92,994.7 24.0901 <0.0001*** Depth 2 1,124 561.8 0.1455 0.8650 Time 3 338,033 112,677.7 29.1889 <0.0001*** Dispersant : Oil 2 75,480 37,740.2 9.7765 0.0001*** Dispersant : Wave 4 84,215 21,053.8 5.4539 0.0003*** Oil : Wave 2 28,265 14,132.4 3.6610 0.0265* Dispersant : Depth 4 12,944 3,235.9 0.8383 0.5014 Oil : Depth 2 5,011 2,505.5 0.6491 0.5231 Wave : Depth 4 18,572 4,643.0 1.2028 0.3089 Dispersant : Time 6 47,433 7,905.6 2.0479 0.0582 Oil : Time 3 1,777 592.5 0.1535 0.9274 Wave : Time 6 47,863 7,977.1 2.0665 0.0560 Depth : Time 6 51,246 8,541.0 2.2125 0.0409* Dispersant : Oil : Wave 4 306,795 76,698.9 19.8687 <0.0001*** Dispersant : Oil : Depth 4 20,321 5,080.3 1.3160 0.2631 Dispersant : Wave : Depth 8 25,722 3,215.3 0.8329 0.5739 Oil : Wave : Depth 4 10,791 2,697.8 0.6989 0.5931 Dispersant : Oil : Time 6 10,436 1,739.3 0.4506 0.8446 Dispersant : Wave : Time 12 29,459 2,455.0 0.6359 0.8117 Oil : Wave : Time 6 1,642 273.7 0.0709 0.9986 Dispersant : Depth : Time 12 15,634 1,302.8 0.3375 0.9820 Oil : Depth : Time 6 3,360 559.9 0.1450 0.9900 Wave : Depth : Time 12 9,548 795.7 0.2061 0.9982 Dispersant : Oil : Wave : Depth 8 11,042 1,380.3 0.3576 0.9422 Dispersant : Oil : Wave : Time 12 19,691 1,640.9 0.4251 0.9535 Dispersant : Oil : Depth : Time 12 6,975 581.3 0.1506 0.9996 Dispersant : Wave : Depth : Time 24 45,060 1,877.5 0.4864 0.9820 Oil : Wave : Depth : Time 12 17,553 1,462.8 0.3789 0.9707 Dispersant : Oil : Wave : Depth : Time 24 20,595 858.1 0.2223 1.0000 Residuals 432 1,667,645 3,860.3

Statistically significant factors determined by more rigorous Bonferroni-adjusted level, p < 0.0016 are flagged with ***. Significant factors determined by less strict level, p < 0.05, are labeled with *.

Table 5. Tukey’s Paired Comparison of the Different Effects Between Treatment Levels on the Average Dispersed Oil Droplet Size in the Water Column

Factor Level mean (mm) Difference estimates Standard error Lower limit Upper limit

Dispersant Water 152.9 Corexit SPC ¼54.9* 5.98 69.0 40.9 Corexit 61.4 Corexit Water ¼91.5* 5.98 106.0 77.4 SPC 116.3 SPC Water ¼36.6* 5.98 50.6 22.5 Wave Regular 121.9 Plunger Regular ¼35.7* 5.98 49.7 21.6 Spiller 122.4 Plunger Spiller ¼36.2* 5.98 50.3 22.2 Plunger 86.2 Spiller Regular ¼ 0.566 5.98 14.6 13.5 Timea P1 148.5 P2 P1 ¼41.9* 6.9 59.7 24.1 P2 106.6 P3 P1 ¼53.5* 6.9 71.3 35.7 P3 94.9 P4 P1 ¼57.7* 6.9 75.5 39.9 P4 90.8 P3 P2 ¼11.7 6.9 29.5 6.2 P4 P2 ¼15.8 6.9 33.6 2.0 P4 P3 ¼4.13 6.9 21.9 13.7 Depth Surface 112.0 Middle Surface ¼3.00 5.98 17.1 11.1 Middle 109.0 Bottom Surface ¼2.53 5.98 16.6 11.5 Bottom 109.5 Bottom Middle ¼ 0.47 5.98 13.6 14.5 Oil MESA 109.9 ANS MESA ¼ 0.51 4.88 9.1 10.1 ANS 110.4

aTime factor is expressed as P1, P2, P3, or P4 to denote the first, second, third, and fourth half-an-hour of measurement period. Within the same time period, the measurements in the middle and the bottom of the tank are 10 and 20 min later than those at the surface, respectively. Significant differences are flagged by * based on 95% simultaneous confidence intervals for specified linear combinations.

143 DISPERSANT EFFICACY AND DROPLET SIZE DISTRIBUTION 1417

A previous wave tank study conducted with less frequently Agreement No. 06-085). The authors thank Susan Cobanli, occurring plunging breaking waves with lower wave energies Rod Doane, Paul Kepkay, Xiaowei Ma, Brian Robinson, Peter (Li et al., 2008) showed that active breakup of oil into smaller Thamer, and William Yeung for their contributions and lo- droplets continued beyond 1 h. gistical support. The findings and opinions expressed in this The significant factors affecting the average VMD are in report are those of the authors and do not necessarily reflect good agreement with those having significant effects on dis- those of the funding agencies. persant effectiveness that has been evaluated by measurement of the dispersed oil concentration in the water column Author Disclosure Statement (Li et al., accepted). For instance, the effects of Corexit 9500 and SPC 1000 on reducing the average VMD from 153 to 61 (or The authors declare that no conflicting financial interests by 92) and 116 (or by 37) mm, respectively, were corre- exist. spondingly correlated with their significant effects on increasing the dispersant effectiveness by 48 and 26%, respectively. The significant effect of the plunging breaking References wave conditions on reducing the average VMD from 122 to 86 ASTM. (2002). American Society for Testing and Materials. F2059-00 (or by 36) mm matched the significant effect of increasing the Standard Test Method for Laboratory Oil Spill Dispersant Effec- dispersant effectiveness by 25%. In addition, dispersion time tiveness Using the Swirling Flask. ASTM 11.04. West Con- was found to significantly decrease the average VMD from shohocken, PA: ASTM, p .1536. 148 to 106 (or by 42) mm and increase the dispersant effec- Byford, D.C., Laskey, P.R., and Lewis, A. (1984). Effect of low tiveness by 18% during the first 30 min of dispersion. Such temperature and varying energy input on the droplet size good correlation can be explained by the enhanced dispersant distribution of oils treated with dispersants. The 7th Annual effectiveness by dispersants and breaking waves through in- Arctic and Marine Oilspill Program (AMOP) Technical Seminar, creasing the fraction of droplets that are more ‘‘permanently’’ Ottawa, ON, Canada, p. 208. dispersed (e.g., <70 mm) in the water column (Tables 2 and 3). Chandrasekar, S., Sorial, G.A., and Weaver, J.W. (2005). Dis- Although reduction of oil–water interfacial tension by che- persant effectiveness on three oils under various simulated mical dispersants allows for turbulent shears to form droplets environmental conditions. Environ. Eng. Sci. 22, 324. on the order of 10 mm, the plunging breaking waves are as- Chapman, H., Purnell, K., Law, R.J., and Kirby, M.F. (2007). The sociated with elevated energy dissipation rates and a lower use of chemical dispersants to combat oil spills at sea: A re- Marine Pollut. Kolmogorov scale that reduces the droplet sizes. view of practice and research needs in Europe. Bull. 54, 827. Delvigne, G.A.L., and Sweeney, C.E. (1988). Natural dispersion Conclusions of oil. Oil Chem. Pollut. 4, 281. Understanding in situ dispersed oil droplet size distri- Drennan, W.M., Donelan, M.A., Terray, E.A., and Katsaros, K.B. bution has significant implications in optimizing dispersant (1996). Oceanic turbulence dissipation measurements in effectiveness testing protocols in systems ranging from small- SWADE. J. Phys. Oceanogr. 26, 808. scale laboratory testing apparatuses to large-scale field trials. Fingas, M. (2005). Stability and Resurfacing of Dispersed Oil. Pre- This research has identified several significant factors that pared for Prince William Sound Regional Citizens’ Advisory determine the in situ dispersed oil droplet size distributions Council (PWSRCAC). Fredericton, Canada: Environmental during dispersant effectiveness testing in an experimental Technology Center, Environment Canada. wave tank. The type of chemical dispersant used is the most Fingas, M.F., Debra, L.M., White, B., Stoodley, R.G., and Crerar, I.D. (1989). Laboratory testing of dispersant effectiveness: The important factor in controlling the dispersed oil droplet size importance of oil-to-water ratio and settling time. 1989 Oil distributions. Application of dispersant generates a large Spill Conference, San Antonio, TX, p. 365. number of small droplets and reduces the average droplet Fingas, M.F., Kyle, D., and Tennyson, E. (1995). Dispersant ef- size. The tested oil-based dispersant Corexit 9500 was ap- fectiveness: studies into the causes of effectiveness variations. parently more efficient in reducing oil droplet size than the In P. Lane, Ed., The Use of Chemicals in Oil Spill Response, ASTM tested water-based dispersant SPC 1000 for the two tested oil STP 1252. Philadelphia, PA: American Society for Testing and types in the batch wave tank testing system. Wave conditions Materials, p. 92. also significantly influence the dispersed oil droplet size dis- Fingas, M.F., Sigouin, L., Wang, Z., and Thouin, G. (2002). tributions, with the experimental plunging breaking waves Resurfacing of oil with time in the swirling flask. The 25th generating the smallest oil droplets. The data linking disper- Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, sant effectiveness and droplet size distributions with the wave p. 773. mixing energy conditions can provide valuable information to ITOPF. (2005). The Use of Chemical Dispersants to Treat Oil Spills. support the formulation of the operational guidelines for London: The International Tanker Owners Pollution Federa- dispersant use. These results are also useful for modeling tion Limited, Technical information paper. No. 4, p. 1. transport, persistence and potentially biological effects of Jasper, W.L., Kim, T.L., and Wilson, M.P. (1978). Droplet dispersed oil in the aquatic environment. size distributions in a treated oil–water system. In: L.T.J. McCarthy, G.P. Lindblom, and H.F. Walter, Eds., Chemical Acknowledgments Dispersants for the Control of Oil Spills, ASTM STP 659. Phila- delphia, PA: American Society for Testing and Materials, This research was funded by the Program of Energy Re- p. 203. search and Development (PERD), U.S. EPA (contract No. 68- Kirby, M.F., and Law, R.J. (2008). Oil spill treatment products C-00-159), and NOAA=UNH Coastal Response Research approval: The UK approach and potential application to the Center (NOAA Grant Number: NA04NOS4190063 UNH Gulf region. Marine Pollut. Bull. 56, 1243.

144 1418 LI ET AL.

Kresta, S.M., and Wood, P.E. (1993). The flow field produced by NRC. (1989). National Research Council: Using Oil Spill Dispersant a pitched blade turbine: Characterization of the turbulence on the Sea. Washington, DC: National Academias Press. and estimation of the dissipation. Chem Eng Sci 48, 1761. NRC. (2005). National Research Council: Understanding Oil Spill Lessard, R.R., and Demarco, G. (2000). The significance of oil Dispersants: Efficacy and Effects. Washington, DC: National spill dispersants. Spill Sci. Technol. Bull. 6, 59. Academies Press. Lewis, A., Byford, D.C., and Laskey, P.R. (1985). The significance Sanchez, M.C., Berjano, M., Guerrero, A., and Gallegos, C. of dispersed oil droplet size in determining dispersant effec- (2001). Emulsification rheokinetics of nonionic surfactant- tiveness under various conditions. International Oil Spill Con- stabilized oil-in-water emulsions. Langmuir 17, 5410. ference. Shaw, J.M. (2003). A microscopic view of oil slick break-up and Li, M., and Garrett, C. (1998). The relationship between oil emulsion formation in breaking waves. Spill Sci. Technol. Bull. droplet size and upper ocean turbulence. Marine Pollut. Bull. 8, 491. 36, 961. Sorial, G.A., Venosa, A.D., Koran, K.M., Holder, E., and King, Li, Z., Lee, K., King, T., Boufadel, M.C., and Venosa, A.D. (2008). D.W. (2004). Oil spill dispersant effectiveness protocol. I: Assessment of chemical dispersant effectiveness in a wave Impact of operational variables. J. Environ. Eng. ASCE, 130, tank under regular non-breaking and breaking wave condi- 1073. tions. Marine Pollut. Bull. 56, 903. Sterling, M.C., Bonner, J.S., Ernest, A.N.S., Page, C.A., and Li, Z., Lee, K., King, T., Kepkay, P., Boufadel, M.C., and Venosa, Autenrieth, R.L. (2004). Chemical dispersant effectiveness test- A.D. (2009). Evaluating chemical dispersant efficacy in ing: influence of droplet coalescence. Marine Pollut. Bull. 48, 969. an experimental wave tank: 1, dispersant effectiveness as a Terray, E.A., Donelan, M.A., Agrawal, Y.C., Drennan, W.M., function of energy dissipation rate. Environ. Eng. Sci. 26, 1139. Kahma, K.K., Williams, A.J., Hwang, P.A., and Kitaigorodskii, Lunel, T. (1993). Dispersion: Oil droplet size measurement at sea. S.A. (1996). Estimates of kinetic energy dissipation under In: Proceedings of the Sixteeth Arctic and Marine Oil Spill Program breaking waves. J. Phys. Oceanogr. 26, 792. (AMOP) Technical Seminar, Calgary, Alberta, Canada. Ottawa: Tkalich, P., and Chan, E.S. (2002). Vertical mixing of oil droplets Environment Canada, p. 1023. by breaking waves. Marine Pollut. Bull. 44, 1219. Lunel, T. (1995). Understanding the mechanism of dispersion U.S. EPA. (1996). Swirling Flask Dispersant Effectiveness Test. Title through oil droplet size measurements at sea. In P. Lane, Ed., The 40 code of feral regulations, part 300, Appendix C. Naragansett, Use of Chemicals in Oil Spill Response, ASTM STP 1252. Phila- RI: U.S. EPA, p. 245. delphia, PA: American Society for Testing and Materials, p. 240. Venosa, A.D., King, D.W., and Sorial, G.A. (2002). The baffled Ma, X., Cogswell, A., Li, Z., and Lee, K. (2008). Particle size flask test for dispersant effectiveness: A round robin evalua- analysis of dispersed oil and oil–mineral aggregates with an tion of reproducibility and repeatability. Spill Sci. Technol. Bull. automated ultraviolet epi-fluorescence microscopy system. 7, 299. Environ. Technol. 29, 739. Wickley-Olsen, E., Boufadel, M.C., King, T., Li, Z., Lee, K., and Mendenhall, W., Scheaffer, R.L., and Wackerly, D.D. (1981). Venosa, A.D. (2008). Regular and breaking waves in wave Mathematical Statistics with Applications. Boston, MA: Duxbury tank for dispersion effectiveness testing. 2008 International Oil Press. Spill Conference, Savannah, GA, p. 499.

145 Marine Pollution Bulletin 58 (2009) 735–744

Contents lists available at ScienceDirect

Marine Pollution Bulletin

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

Evaluating crude oil chemical dispersion efﬁcacy in a ﬂow-through wave tank under regular non-breaking wave and breaking wave conditions

Zhengkai Li a,*, Kenneth Lee a, Thomas King a, Michel C. Boufadel b, Albert D. Venosa c a Centre for Offshore Oil and Gas Environmental Research, Bedford Institute of Oceanography, Fisheries and Oceans (DFO) Canada, One Challenger Drive, Dartmouth, NS, Canada B2Y 4A2 b Department of Civil and Environmental Engineering, Temple University, Philadelphia, PA 19122, USA c National Risk Management Research Laboratory, US EPA, Cincinnati, OH 45268, USA article info abstract

Keywords: Testing dispersant effectiveness under conditions similar to that of the open environment is required for Oil spill improvements in operational procedures and the formulation of regulatory guidelines. To this end, a Dispersant novel wave tank facility was fabricated to study the dispersion of crude oil under regular non-breaking Waves and irregular breaking wave conditions. This wave tank facility was designed for operation in a ﬂow- Currents through mode to simulate both wave- and current-driven hydrodynamic conditions. We report here Dynamic dispersant effectiveness an evaluation of the effectiveness of chemical dispersants (CorexitÒ EC9500A and SPC 1000TM) on two Particle size distribution crude oils (Medium South American [MESA] and Alaska North Slope [ANS]) under two different wave conditions (regular non-breaking and plunging breaking waves) in this wave tank. The dispersant effectiveness was assessed by measuring the water column oil concentration and dispersed oil droplet size distribution. In the absence of dispersants, nearly 8–19% of the test crude oils were dispersed and diluted under regular wave and breaking wave conditions. In the presence of dispersants, about 21–36% of the crude oils were dispersed and diluted under regular waves, and 42–62% under breaking waves. Consis- tently, physical dispersion under regular waves produced large oil droplets (volumetric mean diameter or VMD P 300 lm), whereas chemical dispersion under breaking waves created small droplets (VMD 6 50 lm). The data can provide useful information for developing better operational guidelines for dispersant use and improved predictive models on dispersant effectiveness in the ﬁeld. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction effect by diluting the dispersed oil droplets through advection and spreading. While advection moves them away from the source The use of chemical dispersants can be an effective means to (i.e., the oil slick at the surface), spreading, which is caused by var- combat oil spills at sea (NRC, 1989, 2005). There has been renewed iation of velocity over space, causes the distance between droplets interest for the use of chemical dispersants due to escalated oil to increase. Therefore, currents have the tendency to increase the spill incidents, logistic constraints of traditional spill response op- apparent dispersion effectiveness through dissipating the formed tions, and the development of new generation, low-toxicity, high oil-in-water emulsion droplets away from the treated zone or efficiency dispersant formulations for potential use on oils cover- curbing recoalescence of dispersed oil droplets by reducing colli- ing a greater viscosity range (Chapman et al., 2007; Kirby and sion frequency of dispersed oil droplets. Law, 2008; Lessard and Demarco, 2000). Dispersant effectiveness Bench-scale dispersant effectiveness tests (ASTM, 2002; EPA, depends on the chemical properties of both the dispersant and 1996; Fingas et al., 1987) in the laboratory have been used for com- the oil and mixing energy from wave action (Fingas, 2000). Mixing parison of dispersant product effectiveness (Sorial et al., 2004a,b; results from shear forces in the water body due to both spatial and Venosa et al., 2002) and for testing the effects of temperature, salin- temporal variations in velocities. Velocity shear with its associated ity, and other environmental factors (Chandrasekar et al., 2005, friction also causes the dissipation of kinetic energy of the fluid, 2006; Srinivasan et al., 2007). However, laboratory tests for product which results in the breakup of an oil slick into tiny droplets and selection suffer from the inherent limitation that, regardless of how dispersion of the spilled oil into the water column, especially in closely flow fields are able to mimic mixing conditions at sea, cur- the presence of a chemical dispersant. Sea currents add to the rent effects cannot be accommodated due to space constraints that influence transport and dilution effects. At the other extreme, field * Corresponding author. Tel.: +1 902 426 3442; fax: +1 902 426 1440. tests at sea are expensive and difficult to manage, and results are of- E-mail address: [email protected] (Z. Li). ten inconclusive and non-repeatable due to lack of control of

0025-326X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.marpolbul.2008.12.014 146 736 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744 experimental and climatic conditions. Thus, wave tank studies have was 1.50 m. Different waves were generated by a computer-con- been suggested to generate needed operational data on both mixing trolled flap-type wave maker situated at one end of the tank. The energy and current effects so that chemical dispersion can be stud- wave maker is linked to an adjustable cam that controls its stroke ied and evaluated reproducibly and repeatably in a setting that best to alter wave-heights. The wave frequency is controlled by the simulates conditions at sea (NRC, 2005). rotation speed of the cam. The computer-controlled wave genera- In anticipation of and in response to the requirements for test- tor can produce both regular non-breaking waves and breaking ing the performance of chemical dispersants in more realistic waves. The breaking waves are generated using the frequency oceanographic and environmental settings, the Department of sweep technique (Funke and Mansard, 1979), wherein a wave of Fisheries and Oceans Canada (DFO) and the US Environmental Pro- one frequency is superimposed on another wave of a different fre- tection Agency (EPA) built a wave tank facility. This wave tank was quency, causing the wave to increase in height until it breaks. The originally developed to evaluate dispersant effectiveness under dif- energy dissipation rate per unit mass (e) was evaluated by the ferent reproducible wave energy conditions with energy dissipa- autocorrelation function method (Kresta and Wood, 1993) using tion rates similar to those that are encountered in the field. The a time series of velocity measurements obtained by an Acoustic main goal was to relate quantitatively dispersant effectiveness Doppler Velocimeter (SonTec/YSI Inc., San Diego, CA) at select loca- with energy dissipation rate for varying dispersant formulae, oil tions in the tank. types, and the weathering status of oil. Our wave tank experiments initially conducted in a batch mode configuration demonstrated 2.2. Wave conditions the significance of wave conditions to chemical dispersant effectiveness (Li et al., 2008a,b; Venosa et al., 2008). However, hydrody- Two wave conditions, namely regular non-breaking waves and namic characterization of the wave tank operated in the batch plunging breaking waves, were generated and their hydrodynam- mode also revealed the presence of back-flowing underwater cur- ics characterized. The regular non-breaking waves were generated rents counter to the direction of the progressive waves generated with 12 cm stroke, 0.80 Hz frequency, 2.44 m wave length, and by the wave maker. This recirculation mechanism is caused by 23 cm wave height. The plunging breaking waves were produced the surface Stoke’s drift of the progressive waves (Wickley-Olsen with a 12 cm stroke and alternating trains of high-frequency waves et al., 2008) and is a necessary condition applicable to the conser- (0.85 Hz, wave length 2.16 m, wave height 26 cm, and duration vation of water mass. To counteract the backward underwater cur- 20 s) and low-frequency waves (0.5 Hz, wave length 6.24 m, wave rent flow and to allow for simulation of natural exposure levels height 9 cm, and duration 5 s). that result from dilution of dispersed oil in an open environment influenced by waves, tides, and currents, the wave tank was mod- 2.3. Current flow ified for operation in flow-through mode to simulate the influence of ocean currents. In this work, we studied dispersant effectiveness A uniform current was introduced to the wave tank at a flow subjected to the combined actions of waves and currents. Specifi- rate of 60 ± 2 gallon per min. This rate was selected to counteract cally, we investigated the chemical dispersant effectiveness of the surface Stoke’s drift velocity of the high-frequency (0.85 Hz) two dispersants on two crude oils under regular non-breaking regular wave conditions. The component influent system includes waves and plunging breaking waves while a current velocity equal uptake of seawater from the Bedford Basin (Dartmouth, NS, Can- to the Stokes drift of the progressing wave was applied to the sys- ada), holding tank, electric pump, sediment trap and water filtra- tem. Such an experimental system allows for dilution caused by tion, flow meter, distribution pipes, control valves, and a water the undersea current carrying away the dispersed oil plume. bypass for flow adjustment. The effluent system consists of outlets and valves, flow meter, electric pump, and wastewater treatment 2. Materials and methods facility.

2.1. Wave tank description 2.4. Dispersants

Fig. 1 shows the schematic representation of the wave tank Two commercial chemical dispersants were tested, Corexit facility that was used in this research. The geometric dimensions 9500 and SPC 1000. Both dispersants are listed in EPA oil spill con- are 32 m long, 0.6 m wide, and 2 m high. The average water depth tingency plan. The precise formulae of the dispersants are proprie-

Wave maker A B C D Wave absorbers 5 70 200

LISST 65

200 200 125 800 400 200 200 Effluents Influents 3200

Fig. 1. Schematic representation (all dimensions in cm) of the wave tank facility. Larger circles represent four horizontal sampling locations: (A) 2 m upstream, (B) 2 m downstream, (C) 6 m downstream, and (D) 10 m downstream from the center of the spiked oil slick. 147 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744 737

20 20 A A

15 15 Water Water Corexit Corexit SPC SPC 10 10

5 5 Oil concentration (mg/l) Oil concentration (mg/l)

0 0 B B

15 15

10 10

5 5 Oil concentration (mg/l) Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

Fig. 2. MESA oil concentration as a function of time in the middle of the wave tank Fig. 3. ANS oil concentration as a function of time in the middle of the wave tank at at 10 m downstream location under: (A) regular wave and (B) breaking wave 10 m downstream location under: (A) regular wave and (B) breaking wave conditions. conditions.

tary. Corexit 9500 is a hydrocarbon-based reformulation of water- experiment, 300 ml of crude oil was gently poured onto the water based Corexit 9527 and is meant to be used on higher viscosity oils surface within a 40 cm (inner diameter) ring (constructed of NSF- and emulsions. SPC 1000 is a water-based formulation. 51 reinforced clear PVC tube) located 10 m downstream from the wave maker, and 12 ml of dispersant (or water for the control) 2.5. Crude oils was sprayed onto the surface of the oil slick through a pressurized nozzle (60 psi; 0.635 mm i.d.). This resulted in a dispersant-to-oil Two types of crude oil of varying viscosities were tested: (1) ratio (DOR) of 1:25. The ring was then lifted prior to the upcoming Medium South American (MESA) sour crude (viscosity of 42.3 cP) of the first wave. All of this occurred within several seconds before and (2) Alaska North Slope (viscosity of 50.1 cP) at 21 °C. MESA arrival of the first wave. The design wave conditions were operated oil was weathered by evaporation (sparging with air for 130 h) to continually during the next hour of the experiment. simulate the loss (approximately 14%) of volatile components at sea shortly after a spill. ANS oil was fresh and not weathered to test 2.7. Sample collection the dispersant effectiveness assuming an idea oil spill response scenario where dispersant application is available immediately To collect samples for measuring the total dispersed oil concen- available in the incident. tration, four stainless steel manifolds each connected to three 100- ml syringes were located at four horizontal locations (8, 12, 16, and 2.6. Experimental design 20 m downstream from the wave maker), collecting water samples at three depths (5, 75, and 140 cm from the average water surface) The chemical dispersant effectiveness of the two dispersants on in the water column of the wave tank. Samples were collected at the two oils under the two wave conditions was investigated using five time points (2, 5, 15, 30, and 60 min after start up). These sam- a three-factor mixed-level factorial experimental design with trip- ples were extracted in dichloromethane and then followed by licate runs. The treatments were operated in random order to min- ultraviolet spectrophotometer measurement as described previ- imize the impacts of confounding factors such as temperature, ously (Li et al., 2008a). salinity, and wind. For each experiment, seawater was pumped from the Bedford Basin (Dartmouth, NS, Canada) through a double 2.8. Droplet size distribution layer sock-filter (Atlantic Purification Ltd., Dartmouth, NS, Canada) with pore sizes of 25 and 5 lm for the coarse and fine filters, The dispersed oil droplet size distribution was measured using a respectively. The temperature, salinity, and background particle laser in situ scattering and transmissometry (LISST-100X, Type C, size distribution were recorded before each experiment. For every Sequoia Scientific, Seattle, WA). The LISST was suspended vertically 148 738 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744

25 A 20 MESA ANS

15 Water

10 Corexit 9500 C (mg/l) SPC 1000 5

0 B 20

10 C (mg/l)

0 C 20

10 C (mg/l)

250 D

200

150

C (mg/l) 100

0 M. Regular M. Breaking A. Regular A. Breaking

Wave conditions

Fig. 4. Average dispersed oil concentrations at the surface of the wave tank at four different horizontal locations for the two tested crude oils: (A) 2 m upstream; (B) 2 m downstream; (C) 6 m downstream; and (D) 10 m downstream. Note the different Y-axis scale for (D).

in the water column with the detection window submerged tion of oil droplets in each size interval with average size di, which is around 60 cm lower than the average water surface and approxi- the geometric mean of the lower and upper limit of every size mately 8 m downstream from the center of the initial oil slick. range. Volumetric mean diameter was weighted according to its The in situ dispersed oil droplet size distribution was measured individual volumetric contribution to the total volume of the continuously by the LISST over the entire experimental duration particles. of 1 h. The LISST-100X recorded 32 particle size intervals logarithmically spaced from 2.5–500 lm in diameter, with the upper size in each bin 1.18 times the lower. The measured particle size distri- 2.9. Data analysis bution is expressed as the average volumetric concentration of oil droplets with each interval of the size range. To compare the dis- The effects of treatment factors, including two wave conditions persed oil droplet size distribution as a function of time at each and three dispersant types, on the average dispersed oil concentra- treatment condition, the volumetric mean diameters (VMD) of tions at different locations and depths were identified by conduct- the measured droplet size distribution was calculated: ing two-factor with replication analysis of variance (ANOVA) using P Microsoft ExcelÒ Data Analysis Tool. When significant effects were V d VMD ¼ i i ð1Þ identified, the Tukey’s paired comparison test was conducted to V isolate the factorial effects at each treatment level. The effects of in which V is total volumetric concentration of the particles within these treatment factors on the dynamic dispersant effectiveness the range 2.5–500 lm, and Vi (i = 1–32) is the volumetric concentra- (see definition below) were compared by conducting two-factor 149 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744 739

16 14 MASE ASN A 12 10 Water 8 Corexit 9500

C (mg/l) 6 SPC 1000 4 2 0 14 B 12 10 8

C (mg/l) 6 4 2 0 14 C 12 10 8

C (mg/l) 6 4 2 0 14 D 12 10 8

C (mg/l) 6 4 2 0 M.Regular M.Breaking A. Regular A. Breaking Wave conditions

Fig. 5. Average dispersed oil concentrations in the middle of the wave tank at four different horizontal locations for the two tested crude oils: (A) 2 m upstream; (B) 2 m downstream; (C) 6 m downstream; and (D) 10 m downstream.

with replication ANOVA. A significance level of 0.05 was adopted centrations several-fold under both regular and breaking wave for all statistical tests. conditions, with the breaking waves creating much higher oil concentrations than regular non-breaking waves. Similar oil distribu- 3. Results and discussion tion profiles were observed and recorded for the other sampling locations and depths within the test facility, with variability in 3.1. Effects of dispersant and wave conditions on oil dispersion oil concentrations and the time for the peak oil concentrations to effectiveness occur (data not shown). To compare the effects of dispersant type and wave conditions on oil dispersion in this dynamic environ- The effectiveness of dispersants under different wave condi- ment, the time-series oil concentrations at each sampling position tions with currents was evaluated by monitoring spatial and (horizontal location and depth) were converted to an equivalent oil temporal oil distribution in the flow-through wave tank. Figs. 2 concentration (Ceq, mg/l) that had the same flux of oil as the time- and 3 show the representative dispersed MESA and ANS oil con- dependent oil concentrations over the experimental period (1 h): P centrations, respectively, as a function of time in the middle depth ðDti ðPCi þ Ci1Þ=2Þ (75 cm below the surface) at a location 10 m downstream from the Ceq ¼ ð2Þ Dti initial slick. The dispersed oil concentration at mid-depth increased rapidly (within 5–10 min) and then decreased steadily due to the in which Dti (i = 5) is the time interval between the sampling dilution effect of the current. The effects of wave conditions and events, and Ci is the measured oil concentration at the specific time dispersants are also clearly evident: dispersants increased oil con- point. C0 = 0, since no oil was dispersed at time zero. 150 740 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744

6 A 5 MESA ANS

4 Water Corexit 9500 3 SPC 1000 C (mg/l) 2

0 B 5

C (mg/l) 2

0 C 5

C (mg/l) 2

0 D 5

C (mg/l) 2

0 M.Regular M.Breaking A.Regular A.Breaking Wave conditions

Fig. 6. Average dispersed oil concentrations at the bottom of the wave tank at four different horizontal locations for the two tested crude oils: (A) 2 m upstream; (B) 2 m downstream; (C) 6 m downstream; and (D) 10 m downstream.

Figs. 4–6 summarize the equivalent oil concentration at differ- D). Dispersants significantly (p < 0.05) increased oil concentration ent horizontal locations and depths. The upstream (2 m from the in the middle of the wave tank (Fig. 5C and D), indicating that dis- oil addition) location had consistently low oil (MESA or ANS) con- persants enhanced penetration of oil into the water column. Dis- centrations (about 1 mg/l) at all three depths for all dispersant persants also increased (p = 0.05) the surface oil concentration at types and wave conditions (Figs. 4–6, A), where the effects of wave 6 m downstream location (Fig. 4C), indicating that dispersants conditions and dispersants were insignificant for the dispersed oil stimulated horizontal spreading of oil. These effects are observed concentrations (p > 0.05). Low oil concentrations (about 1–2 mg/l) under both regular non-breaking and breaking wave conditions, were also observed at the bottom of the wave tank at three down- but the extent appeared stronger under breaking waves. This is stream locations (Fig. 6B and C), and the effects of wave conditions consistent with the batch system wave tank experimental results, and dispersant type on oil concentrations at these spots were also showing that Corexit 9500 was effective in low, moderate, and insignificant (p > 0.05). The insignificant effects of dispersant type higher energy conditions, whereas SPC 1000 worked better in and wave conditions were likely due to the strong current flow, moderate and higher energy conditions (Venosa et al., 2008). which counteracted turbulent diffusion of oil, resulting in lower Although the effectiveness of the two dispersants was significantly water column oil concentrations at all depths upstream and the different in the batch system, the difference was insignificant bottom of the downstream locations. (p > 0.05) in the flow-through system (Figs. 4–6). This may be Dramatic effects of dispersants on oil distribution are evident due to the fast dilution of dispersed oil in the flow-through system from oil concentrations at the surface and in the middle of the to prevent recoalescence of the dispersed oil droplets, particularly wave tank at the downstream locations (Fig. 4C–D and Fig. 5C– for the water-soluble dispersant SPC 1000. 151 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744 741

The effect of wave conditions on oil distribution is clearly dem- 100 onstrated by the measured oil concentrations at the surface of the (A) MESA 10 m downstream location (Fig. 4D). The measured oil concentra- Regular tions under regular waves were substantially higher than under 80 Plunger plunging breaking wave conditions, indicating that breaking waves are more effective in transferring oil from the surface deeper into the water column. The effect of breaking waves is related to a ser- 60 ies of contributing factors. During the breaking of waves, it has been estimated that 30–50% of the dissipated wave energy entrains 40 the oil droplets in the water column (Lamarre and Melville, 1991; Tkalich and Chan, 2002). This energy determines the first-order oil entrainment rate (Tkalich and Chan, 2002). Breaking waves devel- 20 op a mixing layer in the upper part of the water column, and the penetration of oil results in a uniform mixing of the droplets, with Dynamic dispersant effectiveness (%) the mixing layer proportional to the height of breaking waves 0 (Delvigne and Sweeney, 1988; Tkalich and Chan, 2002). Moreover, Water Corexit SPC breaking waves generate micro-scale turbulence with the smallest Dispersant type eddies that have the greatest velocity gradients, resulting in deformation, elongation, and eventually breakup of larger droplets 100 (Delvigne et al., 1987; Li and Garrett, 1998). (B) ANS When the oil slick was dispersed into small droplets and con- veyed into the water column, the plume was consequently carried 80 Regular away from the mixing zone through current movement and dilu- Plunger tion in the water column. In the field, the dispersed oil droplets are eventually removed via accelerated biodegradation and other 60 routes of natural attenuation. The rapid dilution of oil from the central mixing zone is desired for minimizing exposure of pelagic 40 species, for their biological effects from exposure to petroleum hydrocarbon compounds are proportional to the intensity and the duration of exposure time. The effects of dispersants and wave 20 conditions on the dynamic dispersion effectiveness (DDE) of the dispersed oil within the experimental duration can be evaluated Dynamic dispersant effectiveness (%) by computing the fraction of dispersed oil flowing out of the wave 0 tank with the effluent current plume and the residual dispersed oil Water Corexit SPC in the water column of the water tank at the end of each experi- Dispersant type ment. In this regard, the flow-through wave tank can be viewed as a vessel in which mixing by the surface regular non-breaking Fig. 7. Dynamic dispersion effectiveness of oil through the water column as a function of dispersant and wave conditions on: (A) MESA and (B) ANS. Data shown waves or irregular breaking waves is coexistent with the plug flow are average with one standard deviation of three independent replicate of the uniform currents. Mixing by waves caused the deviation of experiments. the flow pattern from an ideal plug flow of currents. Fig. 7 presents the estimated DDE of the MESA and ANS oils as a result of physical and chemical dispersion under the two wave the higher dispersibility of fresh ANS versus weathered MESA conditions. For physical dispersion of MESA (Fig. 7A) under regular crude oil. The inhibition of dispersant effectiveness by increased wave conditions, the DDE in the water column was 8% under reg- weathering status of crude oil was reported in many studies ular waves but increased to 19% under breaking waves. The appli- (Chandrasekar et al., 2005; Moles et al., 2002; Nordvik, 1995; cation of Corexit 9500 or SPC 1000, respectively, increased the DDE White et al., 2002). to 22% and 30% under regular wave conditions. The combination of chemical dispersants and breaking wave conditions increased DDE 3.2. Effects of dispersant and wave conditions on droplet size to more than 56% and 46%, respectively, with Corexit 9500 and SPC distribution 1000. Statistical ANOVA indicated both chemical dispersants (p = 0.02) and breaking waves (p = 0.01) significantly increased Dispersant effectiveness is ultimately determined by the dis- the DDE of MESA dispersion, but there was no significant differ- persed oil droplet size distribution (Darling et al., 1990; Lewis ence between Corexit 9500 and SPC 1000 (p = 0.96). Similarly, dis- et al., 1985). Small droplets with sizes of tens of microns have small persants (p = 0.02) significantly increased the DDE of ANS in the rise speeds and tend to remain suspended in the water column and water column (Fig. 7B) but no significant difference was found be- become widely dispersed in the water column by turbulent diffu- tween Corexit 9500 and SPC 1000 (p = 0.76). Physical dispersion of sion and can be potentially removed more rapidly by biodegrada- ANS under regular wave and breaking wave conditions resulted in tion (Li and Garrett, 1998). Conversely, large oil drops with radii DDE in the water column to be 10% and 12%, respectively. Chemical of hundreds of microns tend to recoalesce and resurface unless very dispersion by Corexit 9500 significantly increased DDE to 36% un- strong mixing energy exists to overcome their buoyancy. Therefore, der regular wave conditions, and to 42% under breaking wave con- smaller droplets are much more favorable from the prospective of ditions. Dispersion by SPC 1000 increased the DDE to 25% under oil spill mitigation. Laboratory and field measurements suggest that regular wave conditions and to 62% under plunging breaking wave for an effective dispersion of oil in which the dispersed oil droplets conditions. Although breaking wave conditions were always asso- remain suspended in the water column, average droplet sizes have ciated with higher DDE of ANS, ANOVA indicated that the wave ef- to be less than 50–70 lm(Lunel, 1993, 1995). fect was insignificant (p = 0.13), due primarily to the relatively In this work, the dispersed oil droplet size distribution was large error bars of triplicate runs but could also be attributed to measured by a laser particle counter (LISST-100X) that was sus- 152 742 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744

500 500 A A 400 400 m) 300 m) 300 µ

200 200 VMD ( µ VMD (

100 100

0 0 B B 400 400 m) 300 m) 300 µ

200 200 VMD ( VMD ( µ

100 100

0 0 C C 400 400 m) 300 m) 300 µ

200 200 VMD ( µ VMD (

100 100

0 0 0 102030405060 0 102030405060

Time (min) Time (min)

Fig. 8. MESA oil droplet size subjected to: (A) physical dispersion; (B) dispersion by Fig. 9. Dispersed ANS oil droplet size subjected to: (A) physical dispersion; (B) Corexit 9500; and (C) dispersion by SPC 1000. Open circles are for droplets dispersion by Corexit 9500; and (C) dispersion by SPC 1000. Open circles are for dispersed under regular waves and solid circles are under breaking waves. droplets dispersed under regular waves and solid circles are under breaking wave conditions.

pended in the water column at the end of the flow-through wave at this small size for the rest of the experiment. SPC 1000 rapidly tank. The dispersed oil droplet size distribution was recorded con- dispersed oil droplets (VMD of approximately 75–100 lm) under tinuously as a function of time in real-time mode. The recorded both regular and breaking wave conditions (Fig. 8C and 9C). The size distribution at each snapshot was in the mono-modal logarith- sizes persisted or slightly increased afterwards under regular wave mic normal distribution for the physically dispersed oil and multi- conditions, probably due to recoalescence and resurfacing after the modal log-normal distribution for the chemically dispersed oil depletion of water-soluble SPC 1000 surfactants by current flow (data not shown). These oil distribution patterns are consistent over time. Depletion of surfactants with prolonged mixing of oil- with the dispersed particle size distributions that were observed in-water emulsion stabilized by surfactants was reported in mixing in the batch system wave tank experiments on chemical dispersant tank system (Sanchez et al., 2001). However, the VMD continually effectiveness testing (Li et al., 2008b). To compare the effects of decreased to about 50 lm under breaking wave conditions due to wave conditions and dispersant type on the average dispersed oil the high energy dissipation rate and turbulent diffusion. droplet size distribution in the water column, we calculated the The average size of the physically and chemically dispersed oil volumetric mean diameter (VMD) of the dispersed oil. agree well with the results of our previous batch experimental Figs. 8 and 9 summarize the effect of dispersant type and wave studies (Li et al., 2008a,b) and those of the literature (Byford energy on the average dispersed oil droplet VMD for MESA and et al., 1984; Darling et al., 1990; Lunel, 1995). The introduction ANS, respectively. The droplets started at the same VMD level of current in the flow-through wave tank, however, did not further regardless of dispersant and wave conditions. In the absence of reduce oil droplet size from what had been observed in the batch chemical dispersant (Fig. 8A and 9A), the oil droplet sizes remained system (Li et al., 2008b). This can be attributed to the similarity large and highly variable (VMD 150–400 lm) under the regular of micro-scale turbulence, particularly energy dissipation rates wave condition, but were rapidly reduced in size and variability measured during the hydrodynamic characterization of the batch (VMD 150–200 lm) under breaking wave conditions. In the system (Wickley-Olsen et al., 2008) and the flow-through system presence of chemical dispersant Corexit 9500 (Fig. 8B and 9B), (unpublished data). The flow pattern, however, was clearly differ- the dispersed oil droplet sizes remained large but considerably re- ent when the wave tank was changed from the batch to the duced in variability (VMD 300 lm) under regular wave condi- flow-through mode. In particular, the backflow near the bottom tions; these sizes were dramatically reduced (VMD 50 lm) of the wave tank in the batch system was overcome by the forward under breaking wave conditions within 10 min, and maintained current in the flow-through system, which purged the smaller dis- 153 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744 743 persed oil droplets that were suspended in the water column out of References the wave tank. This reduced the inter-drop collision frequency that would cause recoalescence and resurfacing of the smaller dis- ASTM (2002). F2059-00 standard test method for laboratory oil spill dispersant effectiveness using the swirling flask. ASTM 11.04, American Society for Testing persed oil droplets, while retaining the larger oil droplets floating and Materials, West Conshohocken, PA, pp. 1536–1539. at the surface to maintain high drop-eddy collision frequency for Byford, D.C., Laskey, P.R., Lewis, A., 1984. Effect of low temperature and varying the breakage of droplets into small particles (Tsouris and Tavla- energy input on the droplet size distribution of oils treated with dispersants. rides, 1994). Under breaking waves, however, although the larger In: Proceedings of the Seventh Annual Arctic and Marine Oilspill Program (AMOP) Technical Seminar. Environment Canada. Ottawa, Ontario, Canada, pp. droplets surfaced, they were continually broken into smaller drop- 208–228. lets because of the high energy dissipation rate, and therefore these Chandrasekar, S., Sorial, G.A., Weaver, J.W., 2005. Dispersant effectiveness on three droplets were eventually purged downstream. oils under various simulated environmental conditions. Environmental Engineering Science 22 (3), 324–336. Chandrasekar, S., Sorial, G.A., Weaver, J.W., 2006. Dispersant effectiveness on 4. Conclusion oil spills – impact of salinity. ICES Journal of Marine Science 63 (8), 1418– 1430. Chapman, H., Purnell, K., Law, R.J., Kirby, M.F., 2007. The use of chemical dispersants The effectiveness of two chemical dispersants in dispersing to combat oil spills at sea: a review of practice and research needs in Europe. crude oil into the water column as a function of low and high en- Marine Pollution Bulletin 54 (7), 827–838. ergy waves within an experimental wave tank operated under dy- Darling, P.S., Mackay, D., Mackay, N., Brandvik, P.J., 1990. Droplet size distributions in chemical dispersion of oil spills: toward a mathematical model. Oil and namic flow-through conditions was evaluated in this study. The Chemical Pollution 7 (3), 173–198. data presented demonstrated that the presence of a chemical dis- Delvigne, G.A.L., Sweeney, C.E., 1988. Natural dispersion of oil. Oil and Chemical persant under moderately high wave energy conditions signifi- Pollution 4 (4), 281–310. Delvigne, G.A.L., Van del Stel, J.A., Sweeney, C.E., 1987. Measurements of vertical cantly increased oil concentration in the water column, reduced turbulent dispersion and diffusion of oil droplets and oil particles. MMS 87-111, dispersed oil droplet size distribution, and accelerated dilution rate US Department of the Interior, Minerals Management Service, Anchorage, of the dispersed oil facilitated by current activity. While breaking Alaska. EPA, 1996. Swirling flask dispersant effectiveness test. Title 40 code of feral wave conditions contribute to the breakup of oil into small drop- regulations, part 300, Appendix C, US Environmental Protection Agency, lets due to high energy dissipation rate and enhanced droplet-eddy Naragansett, R.I., pp. 245–250. collision frequency, dispersants may increase the oil breakage effi- Fingas, M.F., 2000. Use of surfactants for environmental applications. In: Schramm, L.L. (Ed.), Surfactants: Fundamentals and Applications to the Petroleum ciency under both regular and breaking wave conditions. We ob- Industry. Cambridge University Press, pp. 461–539. served initial breakup of the oil slick into small droplets, Fingas, M.F., Hughes, K.A., Schweitzer, M.A., 1987. Dispersant testing at the penetration of the droplets into the water column, and the conse- environmental emergencies technology division. In: The 10th Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, Edmonton, Alberta, pp. quent dilution effect of the current flow during the quantification 343–356. of dynamic dispersant effectiveness in the water column of the Funke, E.R., Mansard, E.P., 1979. SPLSH A program for the synthesis of episodic flow-through wave tank. Therefore, the concept of dynamic disper- waves. Ottawa, Canada, Hydraulics Laboratory Technical Report LTR-HY-65, sant effectiveness (DDE) reported here reflects both dispersion of National Research Council, Ottawa, Canada. Kirby, M.F., Law, R.J., 2008. Oil spill treatment products approval: the UK approach oil into water column and transport and dilution of the dispersed and potential application to the Gulf region. Marine Pollution Bulletin 56 (7), oil droplets through water column. The DDE reported here might 1243–1247. be different from the dispersant effectiveness (DE) obtained in Kresta, S.M., Wood, P.E., 1993. The flow field produced by a pitched blade turbine: characterization of the turbulence and estimation of the dissipation. Chemical bench-scale jar tests, where only the contact efficiency between Engineering Science 48 (10), 1761–1774. oil and dispersants is measured in small enclosed surroundings Lamarre, E., Melville, W.K., 1991. Air entrainment and dissipation in breaking and unlimited collision frequency between oil droplets and eddies waves. Nature 351, 469–472. Lessard, R.R., Demarco, G., 2000. The significance of oil spill dispersants. Spill may occur, or batch wave tank tests, where recirculation of the Science & Technology Bulletin 6 (1), 59–68. flow and dispersed oil droplets to the less rigorous mixing zone Lewis, A., Byford, D.C., Laskey, P.R., 1985. The significance of dispersed oil droplet may induce recoalescence and resurfacing of the dispersed oil size in determining dispersant effectiveness under various conditions. In: International Oil Spill Conference. droplets. Indeed, evaluating the dispersant effectiveness under dy- Li, M., Garrett, C., 1998. The relationship between oil droplet size and upper ocean namic flow-through conditions provides a more realistic setting turbulence. Marine Pollution Bulletin 36 (12), 961–970. that may be encountered in the field. The established experimental Li, Z., Lee, K., King, T., Boufadel, M.C., Venosa, A.D., 2008a. Assessment of chemical dispersant effectiveness in a wave tank under regular non- protocol under flow-through conditions in this study will prove breaking and breaking wave conditions. Marine Pollution Bulletin 56 (5), useful for evaluating dispersant effectiveness of different chemical 903–912. dispersant formulations for different oil types under specified Li, Z., Lee, K., King, T., Boufadel, M.C., Venosa, A.D., 2008b. Oil droplet size wave energy conditions. The obtained data of dispersant effective- distribution as a function of energy dissipation rate in an experimental wave tank. In: International Oil Spill Conference, Savannah, GA. ness and particle size distribution as a function of energy dissipa- Lunel, T., 1993. Dispersion: oil droplet size measurement at sea. In: Proceedings of tion rate provide useful information for developing better the Sixteenth Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, operational guidelines for dispersant use and improved predictive Calgary, Alberta, Canada. Environment Canada, Ottawa, Ontario, Canada, pp. 1023–1056. models on dispersant effectiveness in the field. The flow-through Lunel, T., 1995. Understanding the mechanism of dispersion through oil droplet size wave tank system can also be used to conduct environmentally rel- measurements at sea. In: Lane, P. (Ed.), The Use of Chemicals in Oil Spill evant exposure studies on the toxicity of dispersed oil on sensitive Response, ASTM STP 1252. American Society for Testing and Materials, Philadelphia, PA, pp. 240–270. marine species. Moles, A., Holland, L., Short, J., 2002. Effectiveness in the laboratory of Corexit 9527 and 9500 in dispersing fresh, weathered, and emulsion of Alaska North Slope Acknowledgements Crude Oil under subarctic conditions. Spill Science and Technology Bulletin 7 (5–6), 241–247. Nordvik, A.B., 1995. The technology windows-of-opportunity for marine oil-spill This research was funded by the Program of Energy Research response as related to oil weathering and operations. Spill Science & Technology and Development (PERD), US EPA (Contract No.: 68-C-00-159), Bulletin 2 (1), 17–46. NRC, 1989. National Research Council: Using Oil Spill Dispersant on the Sea. and NOAA/UNH Coastal Response Research Center (NOAA Grant National Academy Press, Washington, DC. No.: NA04NOS4190063 UNH Agreement No.: 06-085). The find- NRC, 2005. National Research Council: Understanding Oil Spill Dispersants: Efficacy ings, opinions, and recommendations expressed in this report are and Effects. National Academies Press, Washington, DC. those of the authors and do not necessarily reflect those of the Sanchez, M.C., Berjano, M., Guerrero, A., Gallegos, C., 2001. Emulsification rheokinetics of nonionic surfactant-stabilized oil-in-water emulsions. funding agencies. Langmuir 17 (18), 5410–5416.

154 744 Z. Li et al. / Marine Pollution Bulletin 58 (2009) 735–744

Sorial, G.A., Venosa, A.D., Koran, K.M., Holder, E., King, D.W., 2004a. Oil spill Venosa, A.D., King, D.W., Sorial, G.A., 2002. The baffled flask test for dispersant dispersant effectiveness protocol. I: Impact of operational variables. Journal of effectiveness: a round robin evaluation of reproducibility and repeatability. Environmental Engineering – ASCE 130 (10), 1073–1084. Spill Science & Technology Bulletin 7 (5–6), 299–308. Sorial, G.A., Venosa, A.D., Koran, K.M., Holder, E., King, D.W., 2004b. Oil spill Venosa, A.D., Lee, K., Boufadel, M.C., Li, Z., King, T., Wickley-Olsen, E., 2008. dispersant effectiveness protocol. II: Performance of revised protocol. Journal of Dispersant effectiveness as a function of energy dissipation rate in an Environmental Engineering – ASCE 130 (10), 1085–1093. experimental wave tank. In: International Oil Spill Conference, Savannah, GA. Srinivasan, R., Lu, Q., Sorial, G.A., Venosa, A.D., Mullin, J., 2007. Dispersant White, D.M., Ask, I., Behr-Andres, C., 2002. Laboratory study on dispersant effectiveness of heavy fuel oils using baffled flask test. Environmental effectiveness in Alaskan seawater. Journal of Cold Regions Engineering 16 (1), Engineering Science 24 (9), 1307–1320. 17–27. Tkalich, P., Chan, E.S., 2002. Vertical mixing of oil droplets by breaking waves. Wickley-Olsen, E., Boufadel, M.C., King, T., Li, Z., Lee, K., Venosa, A.D., 2008. Regular Marine Pollution Bulletin 44 (11), 1219–1229. and breaking waves in wave tank for dispersion effectiveness testing. In: Tsouris, C., Tavlarides, L.L., 1994. Breakage and coalescence models for drops in International Oil Spill Conference, Savannah, GA. turbulent dispersions. AICHE Journal 40 (3), 395–406.

155 Marine Pollution Bulletin 60 (2010) 1550–1559

Contents lists available at ScienceDirect

Marine Pollution Bulletin

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

Effects of temperature and wave conditions on chemical dispersion efﬁcacy of heavy fuel oil in an experimental ﬂow-through wave tank

Zhengkai Li a,*, Kenneth Lee a, Thomas King a, Michel C. Boufadel b, Albert D. Venosa c a COOGER, Bedford Institute of Oceanography, Fisheries and Oceans (DFO) Canada, Dartmouth, NS, Canada B2Y 4A2 b Department of Civil and Environmental Engineering, Temple University, Philadelphia, PA 19122, USA c Land Remediation and Pollution Control Division, U.S. EPA, Cincinnati, OH 45268, USA article info abstract

Keywords: The effectiveness of chemical dispersants (Corexit 9500 and SPC 1000) on heavy fuel oil (IFO180 as test Chemical dispersion oil) has been evaluated under different wave conditions in a ﬂow-through wave tank. The dispersant Heavy fuel oil effectiveness was determined by measuring oil concentrations and droplet size distributions. An analysis Temperature effect of covariance (ANCOVA) model indicated that wave type and temperature signiﬁcantly (p < 0.05) affected Wave effect the dynamic dispersant effectiveness (DDE). At higher temperatures (16 °C), the test IFO180 was effec- Dynamic dispersant effectiveness tively dispersed under breaking waves with a DDE of 90% and 50% for Corexit 9500 and SPC 1000, respec- Droplet size distribution tively. The dispersion was ineffective under breaking waves at lower temperature (10 °C), and under regular wave conditions at all temperatures (10–17 °C), with DDE < 15%. Effective chemical dispersion was associated with formation of smaller droplets (with volumetric mean diameters or VMD 6 200 lm), whereas ineffective dispersion produced large oil droplets (with VMD P 400 lm). Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction effort has been directed toward evaluating the feasibility of using chemical dispersants to break up and dissipate heavy fuel oils Marine fuel oil spills occur as a result of leaks during commer- and oils with relatively high viscosities under various mixing cial shipping operations and many other types of maritime acci- energies and more realistic hydrodynamic regimes that may be dents. Heavy fuel oils, which are typically characterized as encountered in the field. Several laboratory studies have demon- having elevated polar fractions and diminished evaporative frac- strated that heavy fuel oils may be dispersed. Fiocco et al. (1999) tions, are usually far more viscous than unrefined crude oils and have shown that dispersion can be enhanced given sufficient mix- tend to be more difficult to disperse in the aquatic environment. ing energy over an extended period of time, and Guyomarch et al. Therefore, field responses to heavy fuel oil spills are more challeng- (1999) have found that dispersion efficacy can be increased ing than responses to light and median crude oil spills. Until two through controlled multiple applications of chemical dispersants. decades ago, oils and emulsions with viscosities greater than By comparing the application of three dispersants on IFO180 and 2000 cP were considered impossible to disperse chemically (NRC, IFO380 using four different bench-scale testing protocols, namely 1989). Indeed, the present guidelines on oil spill dispersant the Swirling Flask Test (SFT), the Baffled Flask Test (BFT), the Exxon application of the International Maritime Organization and United Dispersant Effectiveness Test (EXDET), and the Warren Spring Lab- Nations Environmental Program state, ‘‘Most oils can be success- oratory (WSL) test, Clark et al. (2005) pointed out that mixing en- fully treated with dispersants in the first 4–6 h of a spill; one very ergy is a predominant factor affecting dispersant effectiveness of important exception is heavy fuel oil, which generally cannot be fuel oil dispersion. The high mixing energy of the BFT and WSL pro- dispersed” (IMO/UNEP, 1995). duced orders-of-magnitude higher dispersant effectiveness than Conventional mechanically-based oil spill response technolo- the low mixing energy SFT, whereas application of the EXDET pro- gies are always constrained by weather conditions and sea state duced values between the two extremes. and are inefficient for use in the open sea. With the advent of Because mixing energy is a critical variable in the generation of new generation low toxicity and high efficiency chemical disper- well dispersed lm-size oil droplets, elucidation of the linkage be- sants, interest has been renewed in using chemical dispersants as tween mixing energies of various wave conditions with dispersant a cost-effective response to oil spills at sea. Over the years, much efficacy of heavy fuel oil continues to be of interest. In addition, temperature is a critical environmental variable that must be taken

* Corresponding author. Tel.: +1 902 426 3442; fax: +1 902 426 1440. into account. Srinivasan et al. (2007) used the BFT to investigate E-mail address: [email protected] (Z. Li). different variables, including dispersant-to-oil ratio, mixing speed,

0025-326X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.marpolbul.2010.04.012 156 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 1551 and temperature, on the effectiveness of three dispersants on hea- et al., 2005, 2008; Wickley-Olsen et al., 2008). The new flow- vy fuel oils. They found that chemical dispersion of IFO180 and through system has been used to evaluate dispersant effectiveness IFO380 by three dispersants in the BFT was almost two times more of fresh and weathered crude oil through measurement of oil con- effective at 16 °C than at 5 °C when mixing energy was sufficiently centrations in the water column and the in situ dispersed oil drop- high (Srinivasan et al., 2007). let size distributions (Li et al., 2009). Lab tests are easy to use, low in cost, and valuable for screening The objective of this study was to evaluate the dynamic disper- dispersants for product selection. However, extrapolation of lab sant effectiveness (DDE) on a heavy IFO180 fuel oil in the flow- tests results to the field needs to be done with caution because through wave tank under hydrodynamic conditions similar to wave and current effects, which influence the transport and dilu- those found in the field, including the simulation of mixing energy tion of oil in the open sea, cannot be accommodated in the confines produced by the combined action of different wave conditions and of small-scale testing apparatuses, regardless of how closely flow underwater current flow. Special emphasis was also placed on the fields generated in a bench-scale test mimic mixing energy at correlation of dispersant effectiveness of the test IFO180 oil to the sea. Caution must be exercised when making decisions on real- seawater temperature under regular non-breaking wave and world oil spill responses based solely on the results from labora- breaking wave conditions. tory dispersant tests. A combination of laboratory, wave tank, and field studies have been recommended to establish the best degree of confidence for understanding the limits of dispersants in 2. Materials and methods treating fuel oil spills (Clark et al., 2005; NRC, 2005). In 2003, a series of sea trials off the coast line of the United Kingdom (conducted 2.1. Wave tank facility and testing wave and current conditions at a sea surface temperature of 15 °C and wind speed of 7–14 knots) revealed (from visual observation) that a heavy fuel oil Fig. 1 shows the schematic representation of the wave tank IFO180 could be satisfactorily dispersed under appropriate treat- facility that was used in this research. The dimensions are 32 m ment conditions (Colcomb et al., 2005; Lewis, 2004). Subsequently, long, 0.6 m wide, and 2 m high. The average water depth was large-scale wave tank tests were conducted in the OHMSETT facil- 1.50 m. Different regular and breaking waves were generated by ity to evaluate dispersant performance on IFO180 and IFO380 (Tru- a computer-controlled flap-type wave maker situated at one end del et al., 2005). These experiments were carried out at two distinct of the tank. Two wave conditions, namely regular non-breaking wave frequencies (30 and 33 cycles per min) to relate chemical dis- waves and plunging breaking waves, were tested in this study. persion efficacy to mixing energy, with temperature and salinity The regular non-breaking waves of wavelength 2.44 m and height maintained at the same level as the 2003 UK sea trials. By taking of 23 cm were generated with 12 cm stroke, 0.80 Hz frequency set- a mass balance approach (where the amount of oil dispersed in tings. The plunging breaking waves were produced with a 12 cm the water was estimated by subtracting residual oil recovered at stroke and alternating trains of high-frequency waves (0.85 Hz the surface from the initial oil volume) and a visual observation producing a wave length of 2.16 m and wave height 26 cm with estimation (using the ranking scale used in the 2003 UK sea trials), a duration of 20 s) and low-frequency waves (0.5 Hz producing a Trudel et al. (2005) found the chemical dispersant effectiveness on wave length of 6.24 m and a wave height of 9 cm with a duration IFO180 varied from 17% to 84%, depending on the dispersant-to-oil of 5 s). A uniform current at a flow rate of 3.8 ± 0.2 L s1 was intro- ratio, wave frequency, and dispersant type. In comparison, the dis- duced into the wave tank through a manifold system that allows persion efficiency was <26% in a control experiment with no dis- passing a current through it at the average speed of 0.43 cm s1. persant at high wave frequency. The current flow rate was selected to counteract the previously To better simulate transport and dilution effects of ocean cur- measured surface Stoke’s drift velocity of the high frequency rents in the field, the flow-through wave tank at the Bedford Insti- (0.85 Hz) regular wave conditions. The presence of current allows tute of Oceanography in Dartmouth, Nova Scotia, Canada was used. for dilution and transport of dispersed oil away from the slick, The tank is used for the evaluation of dynamic dispersant effective- which simulates prevailing currents in the sea to achieve more ness under different simulated wave conditions with underwater realistic field conditions. The hydrodynamics of the wave tank currents. Regular and breaking waves are generated in the wave were characterized by four wave gauges that measured water level tank to simulate various energy dissipation rates in the field. The at a frequency of 100 Hz, and an Acoustic Doppler Velocimeter that hydrodynamics of the various wave types generated in the wave measured three-dimensional velocities at a frequency of 50 Hz. The tank facility have been characterized and documented (Venosa breaking waves were generated using the frequency sweep tech-

Wave maker A B C D Wave absorbers 5 70 200

LISST 65

200 200 125 800 400 200 200 Effluents Influents 3200

Fig. 1. Schematic representation (all dimensions in cm, not to scale) of the wave tank. Larger circles represent four horizontal sampling locations: (A) 2 m upstream, (B) 2 m downstream, (C) 6 m downstream, and (D) 10 m downstream from the center of the initial oil slick. 157 1552 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 nique (Funke and Mansard, 1979), wherein a wave of one fre- respectively. The contents of the specific alkanes and aromatic quency is superimposed on another wave of a different frequency, compounds of the test oil are reported in Tables 1 and 2. causing the wave to increase in height until it breaks. In this work, during wave breaking, the wave height of the breaking wave was about 26 cm and increased to 33 cm, while the velocity at the sur- 2.3. Experimental design and procedure face increased from 0.3 m s1 to 0.5 m s1. The energy dissipation rate at the breaking point was at least two orders-of-magnitude The effectiveness of the two dispersants on the test IFO180 un- higher than that of regular waves. The high energy dissipation rate der the two wave conditions was investigated using a randomized under plunging breaking waves was similar to the breaking wave block experimental design. In the first 3 experimental blocks, 6 energy dissipation rate reported in the field (Delvigne and combinations of the 2 wave conditions and 3 dispersant types Sweeney, 1988; Drennan et al., 1996; Terray et al., 1996), whereas (including water as no-dispersant control) were performed once. the values for regular waves were similar to those found on the sea In the fourth block, only the breaking wave conditions with three surface layer (Delvigne and Sweeney, 1988). dispersant types were conducted for an additional experiment in each treatment. In each block, the treatments were applied in random order to minimize the impacts of confounding factors such as 2.2. Testing dispersants and heavy fuel oil variations in salinity and wind effects. The experimental and sampling procedures were consistent Two commercial chemical dispersants were tested, Corexit with the crude oil dispersant efficacy testing in the flow-through 9500 and SPC 1000; both are listed on EPA’s National Contingency wave tank reported previously (Li et al., 2009). Briefly, for each Plan Product Schedule, and their precise compositions are proprie- experiment, 300 ml of fuel oil was gently poured onto the filtered tary. Corexit 9500 is a hydrocarbon-based reformulation of Corexit seawater surface within a 40 cm ring located 10 m downstream 9527 and is meant to be applicable for higher viscosity oils and from the wave maker, and 12 ml of dispersant (or seawater for emulsions. SPC 1000 is a water-based formulation. A heavy IFO180 fuel oil was selected as the test oil in this study, with a specific gravity of 0.960 at 15 °C and an API gravity of 12.5°. The pour Table 2 point and flash point of the test IFO180 was 9 and 75 °C, respec- Aromatic components of the test oil IFO180 characterized by GC–MS. tively. The dynamic viscosity and oil–seawater interfacial tension Aromatic component Abbreviation Content (ng mg1) of IFO180 (at 15 °C) was 2471 mPa s and 21 mN/m, respectively. Naphthalene nap 2005 The test oil was characterized by thin-layer chromatography Methylnaphthalene C1nap 10,005 equipped with a flame-ionization detector (TLC-FID) and gas chro- Dimethylnaphthalene C2-nap 18,619 matography (HP 6890 series II) equipped with a mass selective Trimethylnaphthalene C3-nap 16,072 Tetramethylnaphthalene C4-nap 8508 detector (HP 5971A) operating in the selected ion monitoring Acenaphthene Acenap 615 mode (GC–MS). TLC-FID analysis showed that the tested IFO180 Acenaphthylene Acenapy <25 oil contained 27.1% alkanes (saturates), 41.7% aromatics, 27.0% res- Fluorene Fl 694 ins, and 4.2% asphaltenes. The GC–MS characterization indicated Methylfluorene C1-Fl 2398 that the total alkanes, polycyclic aromatic hydrocarbons (PAHs), Dimethylfluorene C2-Fl 4013 1 Trimethylfluorene C2-Fl 4798 and alkylated PAHs were 223,067, 14,763, and 190,118 ng mg , Dibenzothiophene dibthio 297 Methyldibenzothiophene C1-dibthio 1037 Dimethyldibenzothiophene C2-dibthio 1643 Table 1 Trimethyldibenzothiophene C3-dibthio 1552 Alkane components of the test oil IFO180 characterized by GC–MS. Tetramethyldibenzothiophene C4-dibthio 849 Phenanthrene phen 2885 Alkane component Abbreviation Content (ng mg1) Anthracene Anthr 435

n-Decane C10 1977 Methylphenanthrene C1-phen 2918

Undecane C11 2838 Dimethylphenanthrene C2-phen 15,138

Dodecane C12 3607 Trimethylphenanthrene C3-phen 14,324

Tridecane C13 5161 Tetramethylphenanthrene C4-phen 8070

Tetradecane C14 7931 Fluoranthene ﬂuor 278

Pentadecane C15 10,127 Pyrene pyr 1281

Hexadecane C16 10,254 Methylpyrene C1-pyr 6562

Heptadecane C17 11,293 Dimethylpyrene C2-pyr 10,776 2,6,10,14-TMPdecane (pristane) pri 6327 Trimethylpyrene C3-pyr 8452

Octadecane C18 10,706 Tetramethylpyrene C4-pyr 7038 2,6,10,14-TMHdecane (phytane) phy 4284 Naphthobenzothiophene napbthio 386

Nonadecane C19 10,967 Methylnaphthobenzothiophene C1-napbthio 4150

Eicosane C20 10,332 DimethylNBenzothiophene C2-napbthio 5452

Heneicosane C21 12,419 TrimethylNbenzothiophene C3-napbthio 1198

Docosane C22 14,113 TetramethylNbenzothiophene C4-napbthio 589

Tricosane C23 16,716 Benz[a]anthracene b[a]anth 1310

Tetracosane C24 16,454 Chrysene Chry 2168

Pentacosane C25 15,786 Methylchrysene C1-Chry 7898

Hexacosane C26 13,098 Dimethylchrysene C2-Chry 16,949

Heptacosane C27 11,077 Trimethylchrysene C3-Chry 7038

Octacosane C28 9115 Tetramethylchrysene C4-Chry 4070

n-Nonacosane C29 6394 Benzo[b]ﬂuoranthene b[b]ﬂuor 332

Tricontane C30 2816 Benzo[k]ﬂuoranthene b[f]ﬂour 81

N-heneicontane C31 2480 Benzo[e]pyrene b[e]pyr 611

Dotriacontane C32 2487 Benzo[a]pyrene b[a]pyr 549

Tritriacontane C33 1720 Perylene peryl 365

Tetratriacontane C34 1310 Indeno[1,2,3-cd]pyrene I[1,2,3-cd]pyr 93

n-Pentatriacontane C35 1124 Dibenz[a,h]anthracene dib[a,h]anthr 197 17Beta(H), 21 alpha (H)-Hopane 155 Benzo[ghi]perylene b[ghi]peryl 180

158 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 1553 the control) was sprayed onto the surface of the oil slick through a LISST was suspended vertically in the water column with the pressurized nozzle (60 psi, 0.635 mm i.d.). This resulted in a dis- detection window submerged about 60 cm beneath the average persant-to-oil ratio (DOR) of 1:25. The ring was then promptly water surface and approximately 8 m downstream from the center lifted prior to the arrival of the first wave, and the wave conditions of the initial oil slick. The in situ dispersed oil droplet size distribu- were maintained throughout the 1-hour duration of the tion was measured continuously by the LISST over the entire experiment. experimental duration of 1 h. The LISST-100X recorded 32 particle To collect samples for measuring total dispersed oil concentra- size intervals logarithmically spaced from 2.5–500 lm in diameter, tion, four stainless steel manifolds, each connected to three 100-ml with the upper size in each bin 1.18 times the lower. The measured syringes, were located at four horizontal locations (8, 12, 16, and particle size distribution is expressed as the average volumetric 20 m) downstream from the wave maker. Water samples were col- concentration of oil droplets for each interval of the size range. lected simultaneously in the water column of the wave tank at To compare the dispersed oil droplet size distribution as a function three depths (5, 75, and 140 cm) from the average water surface. of time at each treatment condition, the volumetric mean diame- In addition, water samples were collected from the effluent ters (VMD) of the measured droplet size distribution were calcu- through the effluent pipe sampling port. Samples were collected lated from: at five time points (2, 5, 15, 30, and 60 min after start-up). The P samples were extracted in dichloromethane and then analyzed V d VMD ¼ i i ð1Þ by ultraviolet spectrophotometer adsorption for total petroleum V hydrocarbon (TPH) concentrations as described previously (Li et al., 2008; Sorial et al., 2004; Venosa et al., 2002). where V is total volumetric concentration of the particles within the

The dispersed oil droplet size distribution was measured using a range of 2.5–500 lm, and Vi (i = 1–32) is the volumetric concentra- laser in situ scattering and transmissometry (LISST-100X, Type C, tion of oil droplets in each size interval with average size di, which is Sequoia Scientiﬁc, Seattle, WA) following the same procedure used the geometric mean of the lower and upper limit of every size for previous crude oil dispersion experiments (Li et al., 2009). The range.

(A) 4 (B) 4 Water Water R, 11.6 B, 9.9 3 R, 13.5 3 B, 12.1 R, 16.1 B, 13.8 B, 17.6 2 2

1 1 Oil concentration (mg/l) Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

(C) 4 (D) 40 Corexit 9500 Corexit 9500 3 30 R, 10.1 B, 9.8 R, 10.5 B, 10.2 2 20 R, 15.5 B, 14.2 B, 15.6 1 10 Oil concentration (mg/l) Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

(E) 4 (F) 40 SPC 1000 SPC 1000 R, 12.1 3 30 R, 12.7 R, 16.8 B, 9.9 2 20 B, 13.1 B, 15.6 B, 16.9 1 10 Oil concentration (mg/l) Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

Fig. 2. Effects wave conditions [regular (A, C, and E), breaking (B, D, and F)], and water temperature (numbers in the legend) on dispersed IFO180 oil concentration in the middle of the water column and 10 m downstream from the initial oil slick. 159 1554 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559

2.4. Data analysis the dispersed IFO180 oil concentrations as a function of time at the mid-depth of 75 cm below the surface and located at 10 m down- The effects of the treatment factors on dynamic dispersant stream from the initial slick. Under regular wave conditions, the effectiveness (DDE) were analyzed by fitting an analysis of covari- measured dispersed IFO180 oil concentrations were consistently ance (ANCOVA) model, where the response variable was DDE and low, irrespective of temperature and dispersant (Fig. 2A, C, and the explanatory variables were dispersant (categorical, three E). Under breaking wave conditions, the dispersed oil concentra- levels) and wave energy (categorical, two levels), and covariate tions were also very low without dispersant (Fig. 2B). In the pres- temperature (continuous). The data analysis was performed using ence of dispersant (Fig. 2D and F), however, the dispersed IFO180 S-Plus 7.0 (Insightful Co., Seattle, WA). A significance level of 0.05 oil concentrations increased rapidly within the first 5 min and then was adopted for statistical tests. A Bonferroni adjustment was per- decreased steadily during the remaining 55 min of the experiment formed in analysis of variance (ANOVA) to control for the family- due to dilution by the flow-through current. The effect of temper- wise alpha rate to reach an adjusted and more strict alpha level ature (increasing from 9.8 to 16.9 °C) is clearly shown for disper- at 0.007. sion of the test IFO180 under breaking wave conditions (Fig. 2D and F) where the peak oil concentrations at higher temperatures 3. Results and discussion were several-fold higher than those at lower temperatures. Similar oil distribution profiles were observed for the other sampling loca- 3.1. Effects of temperature and wave conditions on the test IFO180 tions and depths (data not shown). distribution in the wave tank Fig. 3 shows the dispersed oil concentrations in the effluent from the wave tank were similar to those observed within the The effectiveness of dispersants under different wave condi- water column (Fig. 2). Under regular waves, the physically dis- tions with currents was evaluated by monitoring spatial and tem- persed oil concentrations in the current flow were consistently poral oil distribution in the flow-through wave tank. Fig. 2 presents low regardless of temperature (Fig. 3A), whereas chemically

4 4 (A) (B) B, 9.9 Water Water B, 12.1 R, 11.6 3 3 B, 13.8 R, 13.5 B, 17.6 R, 16.1 2 2

1 1 Oil concentration (mg/l) Oil Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

1 10 Oil concentration (mg/l) Oil Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

(E) 4 (F) 40 SPC 1000 SPC 1000 R, 12.1 B, 9.9 3 30 R, 12.7 B, 13.1 R, 16.8 B, 15.6 2 20 B, 16.9

1 10 Oil concentration (mg/l) Oil Oil concentration (mg/l)

0 0 0 102030405060 0 102030405060 Time (min) Time (min)

Fig. 3. Effects wave conditions [regular (A, C, and E), breaking (B, D, and F)] and water temperature (numbers in the legend) on dispersed IFO180 oil concentration in the effluent of the wave tank. 160 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 1555 dispersed oil concentrations were increased slightly at the highest regular wave condition increase the oil concentrations in the water temperature (Fig. 3C and E). In the absence of dispersants, breaking column (Fig. 4B and C). These data also clearly show that breaking waves increased the oil concentration in the effluent to some ex- waves reduced the surface oil concentrations (Fig. 4A) and intent, and the values appeared to correlate incrementally with increased the water column oil concentrations (Fig. 4B and C). The creases in temperature (Fig. 3B). In the presence of dispersants, effectiveness of dispersants on the test IFO180 dispersion in the breaking wave conditions dramatically increased oil concentra- water column under breaking waves was dependent on the test tions in the effluent, particularly at higher temperature (Fig. 3D temperature. At lower temperatures, the dispersants did not ap- and F). pear to increase the oil concentrations in the middle (Fig. 4B) or To compare the effects of dispersant type and wave conditions near the bottom (Fig. 4C) of the wave tank. At higher temperatures, on oil dispersion in this dynamic environment, the time-series oil both Corexit 9500 and SPC 1000 under breaking waves produced concentrations at each sampling position (horizontal location and markedly higher oil concentrations in the middle (Fig. 4B) and near depth) were converted to an equivalent oil concentration (Ceq, the bottom (Fig. 4C) of the wave tank than in the absence of disper- mg l1) with the same flux of oil as the time-dependent oil concen- sant. Correspondingly, use of dispersants reduced the surface oil trations over the experimental period (1 h): concentrations compared to those without dispersant under break- P ing waves (Fig. 4A). ðDti ðPCi þ Ci1Þ=2Þ The high sensitivity of the test IFO180 dispersion to seawater Ceq ¼ ð2Þ Dti temperature was ascertained by the time-averaged oil concentrations of the underwater current flow (Fig. 4D). Under regular wave where Dti (i = 5) is the time interval between sampling events, and conditions, the time-averaged dispersed oil concentrations in the Ci is the measured oil concentration at the specific time point. effluent were consistently less than 1.5 mg/L. Applying dispersants Fig. 4 summarizes the time-averaged equivalent oil concentra- under regular waves had no distinct effect on the oil concentration tions at three depths (located at 10 m downstream) in the water in the effluent. Under breaking waves, at lower temperature, oil column (Fig. 4A–C) and the effluent flow (Fig. 4D) of the wave tank. concentrations were also not affected by dispersants. At higher Under regular waves, high oil concentrations were measured at the temperatures, however, the dispersed oil concentration in the surface (Fig. 4A), while a very limited amount of oil was detected in effluent was increased by several-fold and was most noticeable un- the middle (Fig. 4B) and near the bottom (Fig. 4C) of the wave tank. der breaking wave conditions. Applying dispersants did not appear to have reduced the oil con- The effects of dispersants and wave conditions on DDE over the centration at the surface (Fig. 4A), nor did use of dispersants under duration of the entire experiment can be evaluated by computing

500 14 Water, Breaker Corexit, Breaker (B) SPC, Breaker 12 Water, Breaker 400 Water, Regular Corexit, Breaker Corexit, Regular SPC, Regular 10 SPC, Breaker Water, Regular 300 (A) Corexit, Regular 8 SPC, Regular

200 6

4 Oil concentration (mg/l) Oil concentration (mg/l) 100 2

0 0 8 1012141618 8 1012141618 Temperature (ºC) Temperature (°C)

7 20 (C) (D) 6

Water, Breaker 15 5 Corexit, Breaker Water, Breaker Corexit, Breaker SPC, Breaker SPC, Breaker Water, Regular 4 Water, Regular Corexit, Regular 10 Corexit, Regular SPC, Regular SPC, Regular 3

2 Oil concentration (mg/l)

Oil concentration (mg/l) 5

0 0 8 1012141618 8 1012141618 Temperature (ºC) Temperature (ºC)

Fig. 4. Effects of temperature and wave conditions on dispersed IFO180 oil concentrations: (A) at the surface, (B) in the middle, (C) at the bottom, and (D) in the effluent, of the wave tank. 161 1556 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 the fraction of dispersed oil flowing out of the wave tank with the and F, at lower temperature (10 °C) the DDE for Corexit 9500 and effluent current plume and the residual dispersed oil in the water SPC 1000 were estimated to be 4% and 6%, respectively, whereas column of the wave tank at the end of each experiment: at higher temperature (16 °C), applying dispersants under breaking wave conditions increased the DDE of Corexit 9500 and SPC 1000 C Q T þ C V DDEð%Þ¼ effluent effluent sample wt 100 ð3Þ to nearly 90% and 50%, respectively. qOilV Oil Fig. 5 also illustrates the analysis of covariance linear regression model predicting the DDE as a function of temperature for different where Ceffluent is the time-averaged oil concentration in the effluent 1 current carried out of the wave tank (g L ); Q effluent is the flow rate 1 of the current (L min ); Vwt is total effective water volume of the wave tank (27,000 L); T is the duration of each wave tank experi- Table 3 Analysis of variance (ANOVA) of the effects of temperature, dispersant, wave, and ment (60 min); C represents the average concentration of oil sample their interactions. remaining in the water column after T; qoil is the density of the test 1 Source D.f. S.S M.S. F Pr (F)a oil (g mL ), and Voil is the volume of oil used in the experiment (300 mL). Wave (W) 1 0.281 0.281 57.405 0.00003*** Fig. 5 summarizes the estimated DDE of the test IFO180 for the Dispersant (D) 2 0.067 0.034 6.880 0.01537* three different dispersant types under the two wave conditions. Temperature (T) 1 0.269 0.269 54.946 0.00004*** W:D 2 0.067 0.033 6.826 0.01571* For physical dispersion, the average DDE in the water column W:T 1 0.094 0.094 19.205 0.00176*** was 7% under regular wave conditions (Fig. 5A), increasing to D:T 2 0.182 0.091 18.530 0.00064*** 14% under breaking waves (Fig. 5B). The application of dispersants W:D:T 2 0.070 0.035 7.111 0.01404* under regular wave conditions resulted in DDE of 3% and 6%, Residual 9 0.044 0.005 respectively, for Corexit 9500 (Fig. 5C) and SPC 1000 (Fig. 5E). Un- aSingle asterisk (*) denotes effect was significant at p < 0.05 level; Triple asterisks der breaking waves, however, it is clear that dispersant effective- (***) indicates effect was significant at more strict p < 0.007 level (after Bonferroni ness was highly dependent on temperature. As shown in Fig. 5D adjustment).

100 100 (A) Water / Regular (B) Water / Breaker 80 80

60 60

40 40 DDE (%) DDE (%)

20 20

0 0 10 12 14 16 18 10 12 14 16 18 Temperature (ºC) Temperature (ºC)

100 100 (C) Corexit / Regular (D) Corexit / Breaker 80 80

60 60

DDE (%) 40 40 DDE (%)

20 20

0 0 10 12 14 16 18 10 12 14 16 18 Temperature (ºC) Temperature (ºC)

100 100 (E) SPC / Regular (F) SPC / Breaker 80 80

60 60

40 40 DDE (%) DDE DDE (%) DDE

20 20

0 0 10 12 14 16 18 10 12 14 16 18 Temperature (ºC) Temperature (ºC)

Fig. 5. Dynamic dispersant effectiveness (DDE) of IFO180 as a function of temperature and wave conditions. Dots are experimental data, and lines are the best-fits of the analysis of covariance model. 162 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 1557 treatment (dispersant and wave) combinations. An analysis of var- (A) 500 iance (ANOVA) (Table 3) indicated that all the main factors and the multi-factor interactions had significant effects on DDE (p < 0.05). Under more strict criteria with the Bonferroni adjustment 400 (p < 0.007), two-way interactions of ‘‘wave by temperature” and ‘‘dispersant by temperature” and main factors of ‘‘temperature” 300 and ‘‘wave” were found to affect DDE significantly. Multivariate parameter estimation through linear regression in analysis of covariance (ANCOVA, Table 4) revealed that the treatments of VMD ( µ m) 200 breaking wave conditions (p = 0.0047) and Corexit 9500 (p = 0.0003) significantly increased the value of the intercepts 100 (i.e., the DDE at the average tested temperature, 12.5 °C), whereas the treatments with Corexit 9500 (p = 0.0001) and SPC 1000 (p = 0.0101) under breaking wave conditions significantly in- 0 creased the slope, meaning that increased seawater temperature (B) under these treatment conditions can significantly improve DDE. The best-fit linear prediction models of the DDE for different treat- 400 ment combinations are summarized in Table 5 and illustrated in Fig. 5. 300 µ m) 3.2. Effects of dispersant and wave conditions on droplet size distribution VMD ( 200

Dispersant effectiveness of oil is intrinsically determined by the 100 dispersed oil droplet size distribution (Daling et al., 1990; Lewis et al., 1985), and measurement of the in situ droplet size distribu- 0 tion of the testing system may facilitate comparison of various (C) scales and the droplet sizes observed from the ﬁeld (NRC, 2005). In this work, the droplet size distribution of the dispersed oil was 400 determined by a laser particle counter (LISST-100X) that was suspended in the water column at the end of the ﬂow-through wave tank, following the same procedure as has been used for crude 300 oil dispersion experiments (Li et al., 2009). µ m) Fig. 6 summarizes the effect of dispersant type and wave energy on the volume mean diameter of the dispersed oil as a function of VMD ( 200

Table 4 100 Parameter estimates from the analysis of covariance (ANCOVA) model.

Coefﬁcients Parameter Std. error T value Pr (>|t|) 0 0102030405060 Intercept Water, Breaker 0.1358 0.0364 3.7250 0.0047 Time (min) Corexit, Breaker +0.2836 0.0505 5.6129 0.0003 SPC, Breaker +0.0796 0.0536 1.4836 0.1721 Fig. 6. Effect of dispersant type [(A) water, (B) Corexit 9500, and (C) SPC 1000] and wave conditions [regular (open circles) and breaking (solid dots)] on the volume Water, Regular 0.0586 0.0608 0.9648 0.3599 mean diameter of the dispersed IFO180 oil droplet sizes. Corexit, Regular 0.1060 0.0550 1.9288 0.0858 SPC, Regular 0.0950 0.0605 1.5705 0.1507 Slope (Temperature) Water, Breaker 0.0051 0.0123 0.4189 0.6851 time in the water column as recorded by the LISST-100X. Under Corexit, Breaker +0.1328 0.0186 7.1353 0.0001 regular waves, regardless of the presence or absence of dispersants, SPC, Breaker +0.0582 0.0180 3.2407 0.0101 the volume mean diameters of the dispersed oil droplets were con- Water, Regular 0.0109 0.0251 0.4350 0.6738 sistently high, ranging from 400 to 450 lm throughout the entire Corexit, Regular +0.0014 0.0205 0.0701 0.9456 SPC, Regular +0.0065 0.0229 0.2842 0.7827 experimental period. Under breaking waves, however, even without dispersant (Fig 6A), the volume mean diameters were rapidly reduced from 450 lm to less than 200 lm within 10 min, and remained low thereafter. Application of Corexit 9500 under breaking

Table 5 waves lowered the average dispersed oil droplet sizes from 450 to ANCOVA model predicted estimation of the dynamic dispersant effectiveness (DDE) 100 lm within 5 min, where they remained for the duration of of different dispersant types under two different wave conditions as a function of each experiment. Use of SPC 1000 resulted in similar distributions temperature (T, °C). as the application of Corexit 9500, with the initial oil droplet size Treatment DDE equation reduced from 450 lm to less than 200 lm within 5 min, remaining at around 150 lm until the end of the experiments. Water, Breaker 0.1358 + 0.0051 * (T 12.5) Corexit, Breaker 0.4194 + 0.1379 * (T 12.5) The effects of breaking waves and the presence of dispersants SPC, Breaker 0.2154 + 0.0633 * (T 12.5) on the dispersed oil droplet size distribution exhibited the same Water, Regular 0.0772–0.0058 * (T 12.5) pattern for the test IFO180 as those for unweathered MESA crude Corexit, Regular 0.0298 + 0.0065 * (T 12.5) and unweathered ANS crude oil observed previously (Li et al., SPC, Regular 0.0408 + 0.0116 * (T 12.5) 2009). However, the average dispersed droplet sizes of the test 163 1558 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559

IFO180 under regular waves observed in this study (VMD 400– IFO180 testing oil was kindly provided by Mr. Joseph V. Mullin 450 lm) were significantly larger than those of the dispersed crude from the Minerals Management Service of the US Department of oil (Li et al., 2009) under the same regular wave conditions the interior. The authors wish to thank D. Bellibeau, J. Niven, B. (VMD < 300 lm). This can be attributed to the higher viscosity Robinson and X. Ma for skilful technical assistance. Dr. P.E. Kepkay and interfacial tension of the test fuel oil IFO180 than MESA and has provided useful comments on an earlier version to improve the ANS crude oils, which would also explain the generally poor DDE manuscript. The findings, opinions, and recommendations ex- performance under regular wave conditions (Fig. 5A, C, and E). pressed in this report are those of the authors and do not necessar- Physical dispersion of the test IFO180 under breaking waves pro- ily reflect those of the funding agencies. duced droplet sizes in the same range as those generated by physical dispersion of crude oil, and the DDE’s were in good agreement between the test IFO180 (Fig. 5B) and the two crude oils (19% for MESA and 12% for ANS) that were estimated previously (Li et al., References 2009). Chemical dispersion of the test IFO180 under breaking Canevari, G.P., Calcavecchio, P., Becker, K.W., Lessard, R.R., Fiocco, R.J., 2001. Key waves generated average droplet sizes higher than the crude oil parameters affecting the dispersion of viscous oil. In: 2001 International Oil droplet sizes measured in the wave tank under the same breaking Spill Conference, pp. 4217–4221. wave conditions for the same dispersant (VMD < 50 lm). These Chapman, H., Purnell, K., Law, R.J., Kirby, M.F., 2007. The use of chemical dispersants to combat oil spills at sea: a review of practice and research needs in Europe. sizes were also significantly larger than those suggested as the per- Marine Pollution Bulletin 54 (7), 827–838. manently dispersed droplet sizes, 50–70 lm, in the field (Lunel, Clark, J., Becker, K., Venosa, A., Lewis, A., 2005. Assessing dispersant effectiveness for 1993; Lunel et al., 1995). This can be due to the relatively higher heavy fuel oils using small-scale laboratory tests. In: 2005 International Oil Spill 3 Conference, IOSC 2005, pp. 10593–10597. density of heavy fuel oil (test IFO density is 0.96 g cm compared Colcomb, K., Peddar, M., Salt, D., Lewis, A., 2005. Determination of the limiting oil 3 3 to the weathered MESA, 0.88 g cm and fresh ANS, 0.88 g cm ), viscosity for chemical dispersion at sea. In: 2005 International Oil Spill hence the higher threshold dispersed oil droplet sizes for the hea- Conference, IOSC 2005, pp. 11506–11511. Daling, P.S., Mackay, D., Mackay, N., Brandvik, P.J., 1990. Droplet size distributions in vier fuel oil. chemical dispersion of oil spills – towards a mathematical model. Oil and Chemical Pollution 7 (3), 173–198. Delvigne, G.A.L., Sweeney, C.E., 1988. Natural dispersion of oil. Oil and Chemical 4. Conclusion Pollution 4 (4), 281–310. Drennan, W.M., Donelan, M.A., Terray, E.A., Katsaros, K.B., 1996. Oceanic turbulence dissipation measurements in SWADE. Journal of Physical Oceanography 26 (5), The effects of temperature and wave conditions on the effec- 808–815. tiveness of two commercial dispersants in dispersing a heavy fuel Fiocco, R.J., DeMarco, G., Lessard, R.R., Daling, P.S., Canevari, G.P., 1999. Chemical oil have been quantitatively studied using a flow-through experi- dispersibility study of heavy bunker fuel oil. In: Environment Canada Arctic and mental wave tank. Although heavy fuel oil has been reported to Marine Oil Spill Program Technical Seminar (AMOP) Proceedings, pp. 173–186. Funke, E.R., Mansard, E.P., 1979. SPLSH A program for the synthesis of episodic be more difficult to disperse than light and medium crude oils waves. Ottawa, Canada, Hydraulics laboratory technical report LTR-HY-65. (Chapman et al., 2007; NRC, 2005), we achieved high dispersion National Research Council, Ottawa, Canada. efficiencies of heavy fuel oil under breaking wave hydrodynamic Guyomarch, J., Merlin, F., Colin, S., 1999. Study of the feasibility of chemical dispersion of viscous oils and water-in-oil emulsions. In: Environment Canada’s conditions at higher temperatures (>14 °C). Our observations for 22nd Arctic and Marine Oilspill (AMOP) Technical Seminar, Calgary, Alberta, Corexit 9500 and SPC 1000 demonstrated that the strong effect Canada, pp. 219–230. of seawater temperature dictated the DDE of the test IFO180 with- IMO/UNEP, 1995. Guidelines on Oil Spill Dispersant Application, International Maritime Organization, London, UK, 55p. in a narrow window ranging from 10 to 17 °C. The effect of mixing Lewis, A., 2004. Determination of the limiting oil viscosity for chemical dispersion of energy from breaking wave conditions agrees with results from oil at sea. Final report for DEFRA, ITOPF, MCA, and OSPL. MCA project 10/9/180. field trials (Colcomb et al., 2005) and large-scale tank testing (Tru- CP & R Branck, Maritime and Coastguard Agency, Southampton, SO 15 1EG, UK. Lewis, A., Byford, D.C., Laskey, P.R., 1985. The significance of dispersed oil droplet del et al., 2005) on dispersant effectiveness of heavy fuel oil. The size in determining dispersant effectiveness under various conditions. In: observed temperature and wave effects were also consistent with International Oil Spill Conference. the laboratory BFT experimental results on heavy fuel oil disper- Li, Z., Lee, K., King, T., Boufadel, M.C., Venosa, A.D., 2008. Assessment of chemical dispersant effectiveness in a wave tank under regular non-breaking and sion (Srinivasan et al., 2007). The results presented here have dem- breaking wave conditions. Marine Pollution Bulletin 56 (5), 903–912. onstrated that using chemical dispersants in fuel oil dispersion is Li, Z., Lee, K., King, T., Boufadel, M.C., Venosa, A.D., 2009. Evaluating crude oil feasible at appropriate sea state and environmental conditions chemical dispersion efficacy in a flow-through wave tank under regular non- and provide key data to update standard operational guidelines breaking wave and breaking wave conditions. Marine Pollution Bulletin 58, 735–744. (e.g. IMO, USCG) for dispersant use as an oil spill countermeasure Lunel, T., 1993. Dispersion: oil droplet size measurement at sea. In: Proceedings of at sea. the Sixteenth Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, It is worth noting that the test results reported here are based Calgary, Alberta, Canada. Environment Canada, Ottawa, Ontario, Canada, pp. 1023–1056. on one specific IFO180 fuel oil type that was selected as the testing Lunel, T., Davies, L., Brandvik, P.J., 1995. Field trials to determine dispersant model for heavy fuel oil. These test data should not be generalized effectiveness at sea. In: Proceedings of the 18th Arctic and Marine Oilspill to all heavy fuel oils or IFO180 oil types, some of which may have Program (AMOP) Technical Seminar, Edmonton, Alberta, Canada. Environment Canada. Ottawa, Ontario, Canada, pp. 629–651. been more or less dispersible under the same testing conditions NRC, 1989. National Research Council: Using Oil Spill Dispersant on the Sea. due to the effects of different physical properties of IFO180 fuel oils National Academy Press, Washington, DC. (Canevari et al., 2001). Specific chemical compositions (such as re- NRC, 2005. National Research Council: Understanding Oil Spill Dispersants: Efficacy and Effects. National Academies Press, Washington, DC. sin and asphaltene content) and physical properties (such as pour Sorial, G.A., Venosa, A.D., Koran, K.M., Holder, E., King, D.W., 2004. Oil spill point and dynamic viscosity) of heavy fuel oils, especially IFO180 dispersant effectiveness protocol. I: Impact of operational variables. Journal of oils, could vary substantially from batch to batch. Environmental Engineering – ASCE 130 (10), 1073–1084. Srinivasan, R., Lu, Q., Sorial, G.A., Venosa, A.D., Mullin, J., 2007. Dispersant effectiveness of heavy fuel oils using baffled flask test. Environmental Acknowledgements Engineering Science 24 (9), 1307–1320. Terray, E.A., Donelan, M.A., Agrawal, Y.C., Drennan, W.M., Kahma, K.K., Williams, A.J., Hwang, P.A., Kitaigorodskii, S.A., 1996. Estimates of kinetic energy dissipation This research was funded by the Program of Energy Research under breaking waves. Journal of Physical Oceanography 26 (5), 792–807. and Development (PERD), US EPA (contract No. 68-C-00-159), Trudel, B.K., Belore, R.C., Guarino, A., Lewis, A., Mullin, J., 2005. Determining the viscosity limits for effective chemical dispersion: relating ohmsett results to and NOAA/UNH Coastal Response Research Center (NOAA Grant those from tests at-sea. In: 2005 International Oil Spill Conference, IOSC 2005, No.: NA04NOS4190063 UNH Agreement No.: 06-085). The pp. 10687–10692. 164 Z. Li et al. / Marine Pollution Bulletin 60 (2010) 1550–1559 1559

Venosa, A.D., Kaku, V.J., Boufadel, M.C., Lee, K., 2005. Measuring energy dissipation Venosa, A.D., Lee, K., Boufadel, M.C., Li, Z., King, T., Wickley-Olsen, E., 2008. Dispersant rates in a wave tank. In: Proceedings of 2005 International Oil Spill Conference, effectiveness as a function of energy dissipation rate in an experimental wave Miami, FL. American Petroleum Institute, Washington, DC. tank. In: 2008 International Oil Spill Conference, Savannah, GA, pp. 777–784. Venosa, A.D., King, D.W., Sorial, G.A., 2002. The bafﬂed ﬂask test for dispersant Wickley-Olsen, E., Boufadel, M.C., King, T., Li, Z., Lee, K., Venosa, A.D., 2008. Regular effectiveness: a round robin evaluation of reproducibility and repeatability. and breaking waves in wave tank for dispersion effectiveness testing. In: 2008 Spill Science and Technology Bulletin 7 (5–6), 299–308. International Oil Spill Conference Savannah, GA, pp. 499–506.

165 Estuarine and Coastal Marine ,Science (x98o) xo, 609-633

Sedimentation in the Fraser River and its Estuary, Southwestern British Columbia (Canada)

John D. Milliman b Geological ,Survey of Canada, Vancouver, British Columbia, Canada Received z6 December z978 and in revised form x5 June I979

Keywords: estuaries; rivers; suspended sediments; sand transport; seasonal variations; British Columbia coast

The Fraser River, the largest river (in terms of both water and sediment discharge) reaching the west coast of Canada, is a sand-dominated river in which most sediment transport occurs during freshet in late spring and early summer. More than half the sediment discharged during this 2-3 month period is sand. Throughout the rest of the year, the river is characterized by lower flow and low suspended sediment concentrations (primarily silt and clay); net offshore transport during these months is slight, and near- bottom transport appears to be landward. The dominance of sand transport in the Fraser results in an estuarine depositional regime quite different from most mud-dominated rivers and estuaries. Although most sediment in the river is carried in suspension, about 40% of the sand (zo% of the total load) settles from suspension in the upper estuary and most of the rest settles prior to reaching the lower estuary. In a natural situation, much of the river sand probably would continue moving seaward as bed load, as suggested by the prevalence of migrating sand waves in the middle estuary during freshet. Longshore drift of this sand has built tidal fiats that now dominate the nearshore environment. Dredging of river channels removes an appreciable part of the total annual sand load. Jetties across intertidal fiats and at the river mouth have inter- rupted longshore transport and increased resuspension of sand in the outer estuary by channelizing flow. All of these factors should combine in shifting tidal fiats and adjacent shoreline from their natural state.

Introduction The Fraser River, the largest river reaching the west coast of Canada and one of the largest undammed rivers on the North American continent, is more than x2oo km long and drains an area in southern British Columbia in excess of z3o ooo km 2 (Figure x). In many respects the Fraser is a mountain-dominated river: during the first xoo km of its length, the Fraser drops nearly r km in elevation, then levels out to moderate gradients throughout the next 900 km (upstream of Hope, B.C.), passing through a series of canyons and gorges. The bulk of the runoff within the basin occms in the upper and middle Fraser (east of Hope, B.C.) "Contribution No. 4268 from the Woods Hole Oceanographic Institution bPresent address: Woods Hole Oceanographic Institution, "~VoodsHole, MA 02543, U.S.A. 6o9 o3oz-3524/8o/o6o6o9+25 $o2.oo[o © 198o Academic Press Inc. (London) Ltd.

166 6xo .~. D. 13:lillit~mn

LuLU tsLANI \

~km

t i

t ~b

Figure x. The Fraser River, showing the location and drainage basin (upper right), the lower Fraser (center), and the adjoining estuary (upper left). Stations are numbered as follows: x =Port Mann; 2 =Dens Island; 3=Steveston; 4----Elbow; 5=Sand Heads; 6=2"7 krn offshore.

(Table x). The lower Fraser, extending from Hope to the sea receives additional flow from small rivers draining the Coastal Range (Table I). This section of the river is more mature, being bordered by extensive alluvial flood plains (7oo km 2) that comprise some of the richest farming land in North America. To prevent flooding of this land, a system of dikes has been constructed along approximately 9o% of the river bank downstream from Hope. ~,lost of the sediment deposited by the Eraser since Holocene deglaciation (assumed to be 9ooo years B.P.; J. E. Armstrong, oral communication) has accumulated as a subaerial and subaqueous delta, 975 km2 in area, with an average thickness of xxo m (Mathews & Shepard,

167 Sedimentation in the Fraser River, Canada 61 x

TABLE x. Flow and suspended sediment concentrations in middle and lower Fraser and major tributary rivers. Inability to quantify flow at Port Mann stems from the uncertain tidal influence upon river currents. Data from various x,Vater Survey of Canada publications Maximum Minimum Mean Mean mean mean suspended Annual flow monthly flow monthly flow matter 003 m ~ s -1) (to' m ~ s -a) (Ioa m a s -a) (part/to') Fraser at Hope 2"7 7"0 0"8 21o (-4-42) Harrison River 0"4 x.o 0.2 7 (4-7) Chilliwack River o'x 0"2 0"04 31 (4-17) Fraser at Mission 3"5 8"9 x'3 186 (4-43) Stave River o'x o'2 o't 2 Pitt River o't o'z o'oz 32 (4-3) Fraser at Port Mann ? ? ? x35 (4-39)

1962 ) . Extensive sandy tidal flats characterize the narrow shelf between the shore face and the edge of the delta front (Luternauer & Murray, 1973 ; J. L. Luternauer, in prep.) (Figure 2). Some Fraser sediment has escaped the delta proper and spread throughout the southern Strait of Georgia: several basins in the southern Strait contain Holocene sediments more than 5 ° m thick, most of which presumably were derived from the Fraser (Tiffin, 1969; Pharo, I972). In recent years, the Fraser River has become increasingly important for shipping as well as the site of rapid urbanization and industrialization in southwestern British Columbia. Because of shoaling within the river, more than 2 million tons of material must be removed annually from the navigation channels in order to allow the passage of large ships. Moreover, jetties have been constructed from the shoreline (at Steveston) to minimize shoaling at the river mouth (Figure 2). By channelizing river flow, these jetties have effectively extended the Fraser River 7 km across the narrow shelf to Sand Heads. Although understanding the flow and sedimentology of the Fraser River and its estuary is critical both for scientific reasons and for sound economic development of this river, the Fraser is practically unstudied by geologists, the only published works being on deltaic and offshore deposits (Johnson, 1921 ; Mathews & Shepard, 1962; Luternauer & Murray, I973; Pharo, 1972 ). Most of the detailed information concerning the flow and sedimentary patterns comes from unpublished engineering studies, many of them documenting the feasibility of various river structures, and subsequently tucked away in office files and libraries throughout southwestern British Columbia. Some of these references have been listed in an annotated bibliography by Church & Wahlgren (1974) and in a status report by Hoos & Packman (1974). The first half of this paper describes river flow and sedimentation within the Fraser, summarized largely from suspended and bed load data gathered by the Sediment Survey Section of the Water Survey of Canada. The second half of the report deals with estuarine conditions, based on data from a series of cruises made by the Geological Survey of Canada, and whose methods of sampling and analysis are described in a following section.

River flow in the Fraser During 60 years of measurement, the mean discharge of the Fraser (as measured at Hope) has been 27oo m 3 s -x (Table 1). Most of the disc.harge comes from melting snow and as a result, discharge from late fall through early spring is generally less than 15oo m 3 s -1, while during spring freshet (May through mid-July) flowaveragesmorethan4ooom3s-l(Figure3).

168 6xz J. D. Milliman

Upon closer inspection, river flows vary considerably in timing, magnitude and duration of the spring freshet (Figure 3). To a large extent, these events depend upon the level of snow fall during the previous winter, as well as the speed and timing of spring melt. Peak freshet flows, for example, have ranged between 5000 and z5 ooo m a s -1 (Water Survey of Canada, i974). Freshet also tends to peak slightly earlier at Hope than at either Mission or

49020 '

STRAI OF 49o45 ' GEORG

49040 '

49005 '

49=00 ' 425=25 ' 423o20 ` 425 = 45 ~ 125"tO ~ 425=05 ' Figure z. Bathymetric chart of the Fraser River delta and delta front, depths in meters. Stippled areas indicat~ tidal flats. Note North Jetty extending across the entire narrow shelf to Sands Heads. Chart modified from one developed by J. L. Luternauer (GSC).

169 170

Suspended 10od (mgll)

~o o~ ~ o~

8 8 Dischorge (ms/s]

,.q R'o ,-, I "*o-~

S:=~" ~"'~::7 ...~. .~...... %*°'° ~ ,°-°"~° :o- ~-" °°01 .," • ~.'~ ~ • m~,.~ ¢o /. .~" " _~ '1

~-~ ~ . :*z #'== ----..._._.__

..:J ~L ..... ~....~. f <..._ ..*J ....

o'~ ~0 lb., °" ~ ;:t. ,,,

_._.___-----~ N* ~" 0 ~.. "'~-~ =: x::::==~.-, ...... ~ ~ .~ ~':

0 ¢~ t.xa~_,,

~ x 9 t~trouo,D 'a~a.t~ a~sva,_4 ~tl~ u= uo!2v~u~mjp~S 6z4 .7. D. Milllman

Port Mann (Figure 3), indicative of the later melt of snow fields feeding streams draining into the lower Fraser. Flow in the lower Fraser increases by approximately 2o% relative to Hope, the result of Harrison, Sumas-Chilliwaek, Stave, and Pitt river influx. These rivers drain lakes, have low suspended sediment concentrations (2o-3o mg l-X; Table x), and subsequently contribute relatively little to the Fraser sediment load. From Mission to the river mouth, tidal influence becomes marked, particularly seaward of Port Mann-New Westminster. However, saline intrusion occurs considerably seaward of Port ~'Iann. Tides at Port Mann greatly affect river discharge even during spring freshet. At New Westminster (location shown in Figure 6) the river bifurcates into the North Arm (carrying approximately 12% of the total flow) and the Main Arm, carrying the remaining 88%.

600 • , I I I

"g 400 S

200 miner

Winte •

0 2000 4000 6000 8000 Rivet discl'~rge (m~/s) Figure 4- Comparison of average monthly discharge and suspended loads at Mission, 1966-t97x. Fall and winter values generally show low discharge and suspended load; spring has high discharge and load, and summer has high discharge, but lower levels of suspended load.

Fraser River sediment A major source of Fraser sediment is Pleistocene glacial deposits from interior B.C. (e.g., Pharo, I972 ), as well as the erosion of more indurated rocks. Bank erosion also accounts for some sediment contribution to the lower reaches of the river (Morton, I949; Simmons & Buchanan, z955) (see below). The Fraser transports between xz and 3 ° million tons of sediment annually (Figure 4)- Approximately 80% of the sediment discharge occurs during freshet months (Figure 3; Table 2). Early freshet flows carry greater concentrations than comparable flows in subsequent months (Figure 4), presumably reflecting the greater availability of erodable sediment during early spring (e.g. Nordin & Beverage, x965). Closer inspection of this simple picture, however, shows one far more complex: in any one year, as many as six pulses of sediment- laden water may pass Hope prior to the main freshet (Figure 3)" Some of these influxes

171 Sedimentation in the Fraser River, Canada 6z5

TABLE 2. Freshet and annual suspended discharges at Hope, Mission and Port Mann, together with the suspended sand carried during freshets; data supplied by the Water Survey of Canada. Note that freshet totals and freshet sand remain relatively constant relative to annual and freshet totals (columns 3, 5, and 6) (z) (z) (3) (4) (5) (6) Freshet Freshet Freshet total/ sand/ sand/ Freshet Annual Annual Freshet Annual Observed suspended suspended total Freshet total total freshet discharge discharge ( x zoo) sand ( x too) ( x zoo) Year interval ( x Io ° tons)( x xo' tons) (%) ( X zo' tons) (%) (%)

Hope x967 z May- t5 July x7"5 z3"z 75 9"4 54 4 x t968 z 5 Apr- 3 r July 2z'5 z5"3 89 z t'o 49 43 z969 x5 Apr- x5 July zo'5 z3"z 79 4"8 46 36 z97o x May- 3 ° June 9"5 Iz'o 79 4"7 49 39 I97z x5 Apr- 30 June zx'8 x6"z 73 5"8 49 36 z972 z May- 15 July z3"9 3z'z 75 x3"3 56 4z 78-4-6 504-4 4o4-3 Mission z967 z May- z5 July 26"3 30"5 86 x3"4 5x 44 z968 x5 Apr- 3x July x9"4 z3"5 83 xo'6 55 45 z969 x May- z5 July zz'4 z5"8 72 7"3 65 46 x97 o x May- x5 July zo'o zI'5 86 5"z 5x 44 z97x x5 Apr- 30 June I3"6 t7"6 77 8-z 59 46 x97z x l~,iay- z 5 July 26.I 34"z 76 z5"3 59 45 804-6 574-5 454- z Port Mann z967 z May- x5 July 20"3 24"z 84 8.6 47 36 z968 z May- z5 July z4"6 20"6 7z 5"7 42 36 x969 z May- z5 July 8'4 z z'8 7x 3"4 4 ° 29 z97o z May- x5 July 8"4 zo'6 79 z'9 35 z7 I97x x May- z5 July zI'6 z6-2 7z x972 x May- z5 July 20"3 27"9 73 9"5 47 34 754-5 4x 4-,~ 3z4-4 reach Mission, while others do not, presumably depositing along the way and being resuspended during subsequent high flow (e.g. Figure 5). °

*Although these pre-freshet influxes of sediment appear impressive in terms of concentration, the generally low level of river flow during this period results in relatively low sediment loads. The cause of these influxes is not clear, but may be related to the discharge of sediment-laden tributaries into Fraser River Canyon during early spring (lan Stewart, oral communication).

172 616 07. D. Milliman

Sediment transport involves both suspension and bed load movement, the latter involving esentially particle-to-bed interactions. In the Fraser River two types of suspension occur, one being continuous (wash load) in which the material remains in suspension throughout all phases of the tidal cycle, and the other being discontinuous, in which the material is suspended only during higher flows (bed-material load). Discontinuous suspensions consist mostly of very coarse silt and sand (Figure 5), and during peak spring flows, can account for an appreciable amount of the total suspended load: an average of 5o% of the freshet suspended sediment transported past Hope is sand, 57% at Mission and 4x% at Port Mann (Table 2, column 5)- lXiore striking, however, are the consistent percentages of total annual discharge represented by the freshet sand. At ~lission, for instance, freshet sand accounted for 44, 45 4 6, 44, 46 and 45% of the annual suspended discharge from 1967 through x97z (Table z, column 6). The consistency of these figures is particularly startling in view of the threefold variation in annual suspended load during this interval. Peak sand concentrations at Hope

i i

.! 3oo He~, 1969 ae e :, ', _~_ 200 I \

l \ ~-',

\. I I

i i Mission, 1969 i i

400'

~I Fine silt and clay Sand §8

~+"- ",.+.',,.-~S,,,~,/ ,., _.+:,/ t ~,; • .,:,. "-,,, -...

l ! l April I May I June ! July 1 Figure 5. Daily variation of suspended sand and fine silt+clay (finer than x6 lain) at Hope (upper) and Mission (lower) during spring freshet, I969. Note that sand and mud pulses at Hope in mid-April and early May are represented only by mud at Mission, indicating that the sand had settled out. The high sand levels during late May and early June at Mission, however, suggest that this material was subsequently resuspended and transported downstream.

173 Sedimentation in the Fraser River, Canada 6x 7

predate those at giission and Port Mann by a month (e.g. Figure 5). Also, the amount of sand in suspension at Port Mann is always less than that at Mission (presumably a function of downstream deposition and perhaps of increased bed load). Only a few rivers (such as the Columbia; Haushild et al., 1973) have been shown to carry such high percentages of sand. The dominance of sand in the Fraser probably is related to a number of factors, such as the close proximity of the mountains (essentially only IOO km from the river mouth), the undammed nature of the entire river and the marked seasonality in river flow. The coarse nature of the Pleistocene glacial deposits that form a major sediment source presumably is also an important factor (e.g. van Andel, 1955). Apparently most of the Fraser sediment leaving the mountains and passing Hope is in suspension; bed load is assumed to be less than 5% of the total load (D. A. Dobson, oral communication). Annual mean concentrations of suspended load at Hope average 21o mg 1-1 (Table I). Although influx of sediment-poor waters farther downstream (from the Sumas- Chilliwack and Harrison Rivers) causes lower average concentrations at ~Iission (Table i), increased water flow results in higher suspended sediment discharge, about 5% higher than at Hope (Figure 4). Presumably this downstream increase in sediment discharge is related to channel and bank erosion (Simmons & Buchanan, 1955). In contrast, the suspended sediment discharge at Port Mann averages 17% less than the levels calculated for Mission (Table 2), presumably reflecting deposition and increased bed load in the lower Fraser.

The estuary Tidal effects in the Fraser River estuary are felt east of Port Mann throughout the entire year, even during spring freshet. This strong tidal influence reflects the channelized nature of the estuary, as well as the great tidal range within the adjacent Strait of Georgia. The tides, which are mixed with a strong diurnal component, can exceed 4 m at Sand Heads and Steveston. Tidal range decreases landward and also with increasing river flow. Thus, during winter months, tides at Port Mann can change river level by more than I m, while during spring freshet, ranges may only reach xo to 2o cm. However, even a height difference of IO cm can severely impede or accelerate river flow at Port Mann (see below). This impedance and acceleration results from the storage and discharge of tidally-exchanged water in Pitt lake, upstream of Port Mann. Moreover, because of the time required to fill or empty tidal waters from Pitt lake, high water at Port 1Vlann generally occurs about I h after that at Sand Heads, while low tide shows a delay of 2 to 3 h (Ages & Woollard, 1976). Saline intrusion, which apparently has been enhanced by the deepening of ship channels (Hodgins, 1974), occurs as far landward as Annacis Island (location shown in Figure 6) during periods of low discharge, but does not extend past the lower part of the estuary at freshet. The configuration and extent of the saline layer also depend upon tidal conditions, as will be shown in subsequent discussions. Few data on either ~edimentary or hydrographic characteristics of the Fraser River estuary were available before this study. Perhaps the most comprehensive data were taken during the spring freshet of I975 by the Water Survey of Canada, which show a general seaward decrease in suspended load (primarily from decrease in sand transport) but an increase at Steveston Elbow (Figure 6). The 1975 data also show that sand concentrations at low tide are substantially higher than those at high tide while wash load (silt and clay) concentrations remain remarkably consistent (Figure 6). The downstream decrease in suspended sand indicates deposition and infers increased importance of bed load transport in the lower river. Deposition is indicated by shoaling in the river, requiring dredging of 2-3

174 +e.7 Ox b-* Oo

+E MANN moll

SUSPENDED MATTER

,oqo, 175 SUSPENDED~" SAND ~ 5

• 'tO

i I I 'e" "u4 I SUSPENDEDSI LT t i i & t: CLAY ¢ ;l i ~° i "I "i Fi~,ure 6. Down-river variation in total suspended matter, sand, and silt+clay (expressed in mg 1-1) during xt-x3 June x975, as measured by the Water Survey of Canada. Solid lines (and dots in the bottom diagram) represent measurements taken during lower tidal levels (i.e. higher current velocities), and the dashed lines during higher tidal levels (lower current velocities). Numbers on the individual lines refer to vertical station number across the river at each sample area. Note the general increase of suspended sediment concentration at the Elbow. Sedimentation in the Fraser River, Canada 619

TABLE 3. Dredging efforts on the Fraser River (see footnote). Data courtesy of the Department of Public Works, Vancouver, B.C. Per cent dredged from

Average New annual Steveston Steveston- %Vestminster dredging and New and (x xoe m ~) Seaward Westminster Upstream

Before trifurcation x959--7o 2"3 38 35 27 During construction ,97o-73 *'8 38 37 24 Trifurcation completed x973-75 2"4 55 *9 25

× xo6 m 3 (2 million tons) annually (Table 3)-* Perhaps the best indication of bed load is the presence of large migrating sand waves, up to 4"5 m high and x5o xn long, that form seaward of Port ~Vlann and move downstream at a rate of up to 75 m day-* during spring freshet (Pretious & Blench, I95I ; Allen, x973). This equates to a daily bed load during peak runoff of approximately I-2 × xo5 tons. As the freshet wanes, the sand waves become inactive and by autumn generally disappear, suggesting a marked decrease in bed load transport.

Estuarine observations Tidal and seasonal variations in the hydrographic and suspended matter regimes within the estuary were defined by four cruises during x975-76. The first cruise in late August, x975, occupied stations in the middle estuary (at Dens Island), in the channel between the two jetties (Steveston Elbow and Sand Heads), and ~'7 km directly seaward of Sand Heads (Figure ,). Water samples were collected in 6-1 van Dorn bottles, and filtered through an in-line system onto preweighed 47 mm Millipore ® filters, having nominal openings of o.45 pm. Temperature and salinity of the water samples were measured using a bucket thermometer and optical refractometer. Observations were made every other hour for 25 h, with the vessel steaming continuously between two stations at 2-h intervals, Total water depth in the estuary ranged from ,o to I2 m; sample depths were at o, 3 and 6 m. The seaward station, more than xoo m deep, was sampled at o, 3, 6 and xo m. These same stations, plus additional stations in the upper estuary (Port Mann) and lower estuary (Steveston), were occupied during subsequent cruises in February, April, and ~iay, I976. At the landward three stations, the ship was anchored for 25-h periods, while at

aApproximately 25% of the dredging occurs upstream of Port Mann-New West° minster. Historically, the amounts dredged between New Westminster and Steveston.equalled the amount taken from seaward of Steveston. However, construction of river training works at New Westminster in x97o-73 reduced dredging in the middle estuary, but transferred the problem downstream (i.e. increased dredging at Steveston) and did not reduce the total amount of dredging (Table 3). Most of the Fraser dredge spoils have been used for construction in the munici- palities bordering the river. In recent years, hok-ever, demand for spoils has surpassed public dredging capacities, thus necessitating private dredging. This increased need may well double the amount of annual dredging by x98o (R. Pierce, x976, oral communication). Such high levels of dredging may therefore cause river channel deepening; more importantly, the lack of sand escaping offshore may affect offshore topography and development.

176 6zo ft. D. Millirnan

Elbow, Sand Heads and 2. 7 km offshore, sequential observations were made bi-hourly. Samples also were taken r- 3 m above the bottom; the seaward station was sampled at o, 5, io and ao m depths. Observations made during the February cruise showed that coarser suspended material escaped capture by the van Dorn bottle, the result of decreased turbulence in the bottle when it was lowered through the water column. This problem is particularly severe during high flow when sand represents a major portion of the suspended load. Thtis, during the April and May cruises, a suspended sediment river sampler (Model P-6x) was used to obtain water samples at the three anchor stations. At the three drifting stations in the lower estuary and adjacent Strait, van Dorn bottles were used (a river sampler proving ineffective on a moving ship). Sand, when prominent, was separated from the finer material by a 6z ttm sieve, and both size fractions were collected on Millipore filters. Current velocities were measured at the three anchor stations on April and May cruises. In April, a Hydro-Products Savonius current meter measured velocities r m above the bottom. During late iVIay, a Price current meter, attached to the P-6x sampler, was used, with measurements at the surface, 3 m, 6 m, 9 m, and at several intervals above the bottom. Based on river data discussed above, hydrographic and sedimentologie conditions in the Fraser River estuary can be separated into three general seasons: autumn-early spring (low water/sediment discharge), freshet (high water/sediment discharge), and summer (relatively high water di~charge[low sediment discharge). Observations taken during the r975-76 cruises, therefore, will be discussed within this general framework:

Autumn-early spring River flow during both the February and April cruises averaged about iooo m 3 s-X; as a result of this low discharge, even surface waters at Deas Island were brackish and salinities near the bottom often exceeded 2O~oo (Figures 7 and 8). At Sand Heads, surface salinities seldom fell below 5~oo, and offshore only a thin layer of brackish water overlay the more saline Strait of Georgia waters. While maximum salinity at Sand Heads coincided with high tide, it was 2 h later at Steveston and 3 to 4 h later at Deas Island. This lag in saline intrusion up the estuary, as noted by Hodgins (1974), depends upon advection by near-bottom currents, which can continue flowing landward long after a reversal of surface current flow (see below). During April, near-bottom currents at the three anchored stations clos61y followed salinity trends. At Port Mann, surface flow was continually seaward (even during HHT), but near- bottom flow reversed for several hours (Figure 9). In contrast, at both Dens Island and Steveston, surface currents reversed landward briefly during flood tides, while near-bottom flow was primarily landward throughout the entire tidal cycle, thus explaining the pre- dominance of saline waters in bottom waters of the lower estuary. Sediment load during February was extremely low: concentrations were generally 5 to 7 mg I-1 at Port Mann and decreased downstream with increasing salinity (Figure 7). During LLT, however, concentrations at Port Mann increased to 95 mg 1-1 (x m above the bottom), m but decreased again at slack water. Similar flow-related fluctuations in suspended matter concentration were noted at Deas Island and Steveston, but were not seen at Elbow or Sand Heads. Because of the low terrigenous content, percentages of organic material were relatively high (an average of I8~o combustible matter). Judging from visual inspection, a large percentage of this material was wood fibre, derived from logging activities on the river. Although river flow in April was similar to that in February, suspended concentrations were significantly higher, reflecting the higher suspended loads carried by pre-freshet flows (Figure 3)" Average suspended concentrations at Port Mann averaged about 3 ° mg 1-1, and

177 SALINITY (°/m,ol $USPIrNOEO MATTER (rag/I)

Time (h) Time (h)

O0 06 12 18 O0 06 12

OSO$ lslond • .,,o,~ • • "- .f~,. ..

Steveston S

rlbow 5 5

lO SO

15 • 15

S~d Heeds

• • 15 0

5 2,7 km 5 offlhore '°I IO

15 15! Figure 7. Temporal variation of salinity (~; left) and suspended matter (mg l-X; right) in the Fraser River estuary, "3-27 February I976. River discharge was about zoos m 3 s-L Port Mann values were generally low due to sampling errors (see text), and thus are not shown. Vertical axis is depth in meters, horizontal axis is time in hours (dots representing sample polnts). In this figure and following ones salin- Sties within shaded areas are less than t ~o; cross-hatch greater than 3o~o. In suspended matter, hatched areas representvalues less than x mg 1- z cross hatch greater than 6o mg I-L The river surface is shown relative to the tidal curve predicted for each locality on the river, thus showing periods of high and low tide. Bottom lies about t-3 m below the deepest samples.

178 622 .7. D. MilEman SALINITY(%0) SUSPENOEDMATTER (nag/I) Time (h)

18 0 06 12 18 Port '.~~ .-g Fresh Mann • • o • ° ° ~ °

• ° ° , ° g '~' Io

0 9 Time(h} oo 06 IS I1~ O0 o

Deos lstond

Q Io

Steveston 5

v 15-- !

it ~ - Elbow

,o . .'.J...kk•~fi - ,o • •

"0 0 Heads

0 2.7 km offshore

IO • ° • •

•

Figure 8. Temporal variations of salinity (7oo; left) and suspended matter (rag I-*; right) in the Fraser River estuary, 5-9 April x976. River discharge was approximately xooo m ~ s-l• Refer to Figure 7 for further explanation of symbols and configurations used.

179 Sedimentation in the Fraser River, Canada 6z3

more than 60 mg 1-1 during low tides; values were nearly 4 ° mg 1-1 at Steveston (Figure 8). High concentrations during low tide reflected bottom and resuspension by increased flow (Figure 9). Interestingly, silt and clay concentrations increased slightly just prior to peak flow (best seen at Port IyIann; Figure 9), but then fell with increased discharge (during which sand transport increased). Presumably this phenomenon reflects the initial scour of finer

18 (30 06 12 ' DO 06 12 18 O0 06 12 11: (30 I0(~ ~ i T I .~ I i , i i i / i l

Port Mann .J [~os Idond w.,..... Steveslon t. O

Figure 9. Relation of river stage (as indicated by current velocity and salinity) and suspended matter (sand and silt+day) in the Fraser, x m above the bottom, during the April Cruise.

SALINITY (%e3 TOTAL SUSPENDED SUSPENDED MATTER SAND (mg/I) (mg/I}

Time (h) Time (h) Time (hi

06 12 06 12 12 16

1O $0

•. A, • 0 :i, f 2.7 km \ offshore 5

/ L0 I0 IE 15 i Figure xo. Temporal variation of salinity (~oo; left), total suspended matter (rag l-t; center) and suspended sand (mg I-I; right) at Sand Heads and z'7 km offshore during a fresh flood on zz May z976. River flow was 8ooo m ~ s-'. See Figure 7 for further reference to symbols and configurations used.

180 624 3 e. D. Milliman

bottom material during increased river flow; once stripped of this finer sediment, sand is resuspended during higher flow. While maximum sediment movement at Port ~Iann occurred during low tide, highest concentrations at both Deas Island and Steveston occurred during peak landward flow, during flood tide.

VELOCITY (era/s) Time ( h } 12 ~ O0 06 I !

5 ~'" "" ~~ • °

PoEt Moan o /

• .

'" ( "

Oeos ~$1and

tO • ° ° , •

Sleveston I0

D Q 0 15

Figure zz. Temporal variation of current speed (cm s -x) at Port Mann (upper) Deas Island (center) and Steveston (lower) during 25-28 May x976. Vertical scale is depth in meters, horizontal scale is time in hours. Dots represent data points•

181 SALINITY (%o) SUSPENDEDMATTER (n'R/I) Time (h) R 11 O0 O~

Port 5 Fresh Monn "E ...... :

Fresh IMond -. " I¢-

• • • • °/. • II

Fresh Stevnton s " "

Io Time (h) ~°~"I~ 11 O0 m 12 o

Elbow s . " ~

o 0

HeodsSond = ~t

2.7 km s - i offshore

10 •

10)t ...... /

2C ~ Figure z2. Temporal variation in salinity (~; left) and suspended matter (rag I-Z; right) in the Fraser ttiver estuary during =5-a9 May z976. River discharge was ?ooo m s see. Hatched areas are suspensions less than 8o mg 1-1; crossed-hatched are greater than zooo mg ] -1.

182 6z6 if. D. Milliman

Spring freshet Because of warm weather during early spring in 1976, Fraser discharge peaked in early May and then fell later in the month. Peak flow conditions, unfortunately, were measured only during 8 h of sampling on xz May when river flow at Mission was 8000 m 3 s -x (D. A. Dobson, oral communication). Spring tides were nearly 4 m in range, thus accentuating flow conditions in the estuary. At LLT, the entire water column at Sand Heads was fresh, and offshore surface values were as low as 3~oo (Figure 1o). Tidal influence was even more obvious in sediment discharge. At LLT, total suspended matter at Sand Heads exceeded 60o mg 1-: throughout most of the water column and was greater than 18oo mg 1-1 near the bottom, the highest concentration measured during the 1975-76 cruises. Most of the LLT load was sand, with near-bottom concentrations approaching 1ooo mg 1-1 (Figure io). During LHT, however, sand concentrations fell to less than Img 1-t throughout most of the water column, the suspended material being predominantly wash load (silt and clay). Offshore suspensions were very high, more than 400 mg I -t at times. These high values within the fresh water plume clearly reflect sufficient vertical turbulence to maintain high levels of suspension. The estuary was sampled more completely in late May when river flow was slightly lower, about 7ooo m s s -x. As expected, velocities far exceeded those measured in pre-freshet conditions: at Port Mann surface currents fell below zoo cm s -1 only at high tide (Figure 1:). Velocities decreased with increasing depth, but still exceeded too cms -1 at most times. Similar tidal and vertical variations in currents were noted at Dens Island and Steveston, but velocities were markedly reduced compared to Port Mann, partly because of bifurcation of river flow below Port Mann, and partly because of the downstream widening of the river At Steveston, surface and near-bottom currents were nearly identical, maximum velocities occurring at intermediate depths (3 to 9 m). This uniform current velocity throughout the water column suggests effective vertical mixing. Because of the high discharge, surface waters as far downstream as the Elbow remained fresh throughout the tidal cycle (although near bottom salinities ranged from 5 to ao~oo) (Figure lZ). The upper 5 m of water z. 7 km seaward of Sand Heads were generally brackish, and surface values occasionally less than 1~oo (Figure xz). Suspended concentrations were lower than those measured in mid-May, ranging from 8o to xao mg 1-1 except at ebb flow, when surface concentrations increased to more than 13o mg 1-1 and bottom values (at Port Mann) to I4oo mg 1-a. Concentrations at Dens Island and Steveston decreased (relative to Port Mann), particularly in near-bottom waters at LLT (Figure In). However, during low tide at Elbow and Sand Heads, near-bottom concentrations increased to more than 800 mg 1-1, indicating that bottom sediment was resuspended and transported seaward. Although concentrations offshore were not as high as those earlier measured in the month, values in excess of 1oo mg 1-1 were measured at the surface just after low tide. As seen during the April cruise, the high suspended loads during maximum discharge reflect high levels of ~and in suspension. At Port Mann, for instance, near bottom sand concentrations increased from less than ioo mg 1-1 (at HHT) to as high as 14oo mg 1-1 at LLT, while silt and clay concentrations remained relatively constant (Figure 13). Similar tidal variations in sand concentrations were notcd downstream.

Summer (post-freshet) Post-freshet flow in summer has relatively high discharge, but low suspended loads (Figure 4). Unfortunately, only the cruise documenting this period did not occupy stations at Port Mann or Steveston. Moreover, neither near-bottom water samples nor current measure-

183 Sedimentation in the Fraser River, Canada 627

12 O0 12 O0 OI 12 I I I ' • t l I I t I ! l I I I I I I I ` I ' t ' ' I H

• 1400

• 1200

5o. • 1000

0 -S00 Deos Islana Ste~e.~ -6oo ~

A -40(1

- 200 . -0

O0 12 O0 12

.... , , , , , 7

Figure x3. Variation of river stage (expressed by current velocity and salinity) and suspended sand and silt+day, r m above the bottom, Fraser River, 25-29 May x976• ments were taken. Nevertheless, the data do provide some indication of summertime conditions in the Fraser River estuary. River flow during late August (I975) was just under 3o00 m a s-l; although no near- bottom observations were taken, river water at Dens Island was completely fresh, while downstream it was brackish (Figure x4). Salinities at Steveston Elbow reached a maximum just after high tide, while minimum surface salinities at Sand Heads and at the offshore station both occurred several hours after low tide. Suspended matter concentrations averaged between zo and 3 ° mg 1-1 in the river water, except during low tide, when concentrations exceeded 60 mg 1-1. Undoubtedly these high concentrations represented ambient bottom material stirred up by increased flow during low-tide discharge. In contrast, during high tide salt-water intrusion, concentrations at Sand Heads and Elbow often were less than 5 and xo mg 1-x respectively. Only during low tide did the suspended matter offshore exceed zo mg 1-1, and at to m water depth concentrations during high tide were less than x mg 1-1 (Figure z4).

Discussion and summary In terms of both river flow.and sediment transport, the Fraser River can be viewed as having two distinct regimes when conditions are radically different. x. Throughout most of the year, from mid-summer through early spring, river flow is relatively low (generally less than 3ooo m s s -1) and the small suspended sediment load is dominated by silt and clay-size materials. Although reliable measurements are lacking, bed load during these 9-xo months appears to be insignificant. Field observations during the summer, winter, and early spring show that the estuary is partially mixed, with saline bottom waters extending at least to Deas Island (in the middle estuary) at high tide. Throughout much of this period, even the surface waters in the lower

184 6z8 ft. D. Milliman

SALINITY (%0) SUSPENDED MATTER (rag/I] Time Ih)

12 18 O0 06

OtOll lllond b---. ~. o ,,".-,,- • • • i. I ! ...i_ Ih} Time io~¢i/..~---..,,....,,~. 12 18 O0 06

Elbow ~..l', .\.- • .~/..

l0 ~- 0

2O 10 Send Heodl

5 2.7 km offshore

10 10

IS- Figure z4. Temporal variation of salinity (~oo;left) and suspended matter (mg l-a; right) in the Fraser River estuary, z9-zx August x975. River discharge was 3000 m s s -t. Refer to Figure 9 for explanation of symbols and configurations used.

estuary are slightly brackish (Figure z5). Suspended matter concentrations within the estuary during all the~e months are generally less than 5° mg 1-1, and during high tide often less than z 0 mg 1-1 (Figure zS). Current and suspended matter measurements (April x976) allow a rough calculation of the magnitude and direction of near-bottom (z m above the bottom) suspended sediment transport within the estuary during periods of low river discharge. Near-bottom transport at Port Mann is primarily seaward (landward transport being limited to several hours at high tide); suspended sand transport occurs only during peak flows at LLT (Table 4; Figure z6). At both Dens Island and Steveston, net near-bottom transport is landward, with seaward

185 HHT LLT HHT LLT

0 SH/I EL STl DIJ\ PM[ SH El. ST DI PM

• %o

~- February 24-28, 1976 April 5-9, 1976 E E • • ° • • ! •

2" 186 WINTER EARLY SPRING

, s. Et ST ol 3

• August 19-21,1975 May 25 -29,1976

FRESHET SUMMER

Figure z5. High and low tide variations in salinity and suspended matter on the Fraser River throughout the ,,QIO year. In each set of diagrams, HHT salinities (~oo; upper) and suspended matter concentrations (mg l-Z; lower) are shown in the left, LLT in the right, for Port Mann (PM), Deas Island (DI), Steveston (ST), Elbow (EL), Sand Heads (SH) and 2"7 km offshore. Dots represent sample stations. February conditions represent winter conditions; April, early spring; May, spring freshet; and August, summer. 630 J. D. Milllman

transport only during LLT. A somewhat similar situation has been noted in the Columbia River estuary during periods of reduced discharge (Hubbell et al., x975). Because flow in the surface waters of the lower estuary is primarily seaward, the net transit of suspended sediment integrated throughout the water column of the Fraser may be seaward. Nevertheless, sand transported past Port Mann must accumulate in the river channel (and/or continue downstream movement as bed load) since it does not reach Dens Island in suspension. 2. During freshet flow in spring and early summer, the Fraser River estuary is essentially fresh, except at Sand Heads and Elbow during high tide, when a prominent salt wedge

FRESHET MAY 25-28, 1976 (RIVER DISCHARGE~7000m3/sec)

HOURS (RELATIVE,] 0 4 8 {2 16 20 24 u 12.0 EARLY SPRING i PORT MANN APRIL 5-8, {976 IO.O (RIVER DISCHARGE~IO00m3/sec) 8-0 HOURS (RELATIVE) 0 4 8 ~2 t6 20 24 6.0 i 40 ZC / s.T s ctAv

% G.O ~ DEAS ISLAND 4.0 ~ °~ 2.O ~ "~ O!f DEnS |S~ND ~ ~ O. 0 -!.0

STEVESTON g~

2.0 ! .-of O. 0 r STEVESTON,f~ ~ - - : : : 0 -O.It ~ -I.O -0.;3

Figure z6. Calculated sediment transport (kg per cm ~ of cross-sectional area per minute) ! m above the bottom, during 5-8 April (leh) and 25-28 May (right) x976. Transport was calculated by multiplying the suspended sediment load (sand and silt+clay) by the current velocity z m above the bottom. See Figures I x and z 5 for graphical representations of the data. Note that the vertical scale for May is an order of magnitude greater than for April.

187 Sedimentation in the Fraser River, Canada 63x

develops (Figure z5). Eighty percent of the annual suspended sediment is transported during freshet, about half of which is sand (in contrast to non-freshet conditions when silt and clay predominate). Bed load transport also can be significant, as suggested in the following calculations: Current and suspended sediment measurements made during the late May (I976) freshet show a near-bottom sediment flux about 3 ° times higher than that calculated for pre-freshet flow (Figure x6, Table 4). Furthermore, no landward transport was noted at any station. The daily flux of suspended sand, however, decreased seaward from 393 ° kg cm- z cross-sectional area at Port Mann to 680 kg cm-2 at Steveston (Table 4). Presumably, this loss of suspended sand reflects deposition and]or transfer to bed load. Not surprisingly, the river channel between Port Mann and Dens Island has maximum sand wave development during spring freshet (Pretious & Blench, i951 ). Although no current measurements were taken either at Elbow or Sand Heads, near-bottom suspended sand concentrations during ebb flows in May were significantly higher than at Steveston (Figure x7), indicating that sediment is resuspended and subsequently transported out of the estuary. This condition contrasts strongly with April, when little indication of a turbidity maximum was noted (Figure x7). Resuspension would explain the significant amount of suspended sand noted in the surface offshore waters (2. 7 km off Sand Heads) during both May cruises (e.g. Figure x2).

TABLE4- Calculated sediment fluxes (kg cm -2 day-t), x-meter above the bottom at Port Mann, Dens Island and Steveston, early April and late May, x976 Pre-Freshet Freshet (5-8 April t976) (25-28 May I976) Downstream Upstream Net Downstream Upstream Net Port iMann Silt and clay 98 700 -- Sand 54 xt +x4x 3930 __ +4630 Deas Island Silt and clay x3 29 66o Sand -- 5 --2 x x3to -- + x970 Steveston Silt and clay 4 7xo -- Sand __ z7 --23 680 -- + x59°

A final word should be mentioned about the impact of modern civilization upon Fraser River sedimentation. Prior to the 2oth century, an appreciable amount of the freshet sediment presumably was deposited on the subaerial portion of the delta by flooding river waters. Of the material escaping the estuary, much was transported by longshore currents and accumulated on tidal fiats that dominate the narrow shelf between the shoreline and the delta front (J. L. Luternauer, personal; communication). The remainder of the riverine material presumably escaped offshore. Construction of dikes along the river banks has effectively ended flood-related deposition on the low-lying Fraser River valleyland. Active dredging of the river channels (2 million torts per year and probably more than 3 million tons by 198o) has removed much of the sand that could have reached the sea. The jetties between Steveston and Sand Heads undoubtedly have helped channelize river flow, thus increasing resuspension in the outer estuary and facilitating offshore transport. Finally, building breakwaters and jetties across the tidal fiats (Figure 2) has decreased longshore transport of sediment that does escape onto the narrow

188 63z ft. D. 2~Iilliman

40-- i i' i I i I

30 X ! 0 20

APRIL D

0 l l f . I I ., I

! i | E t I = 0 O-I%o at surf d 600 ® X O-t%. at 3m § • O-t%,= at 6m III 2-3%o at surf • 2-3%o at 3m 500 n 2-3%+ at 6m A 4-5%o at surf Im above bottom (aLl sat[~lt[es) P" 400 MAY

300 ® ® 200 x x •

100 o o o o II

0 ! I 1 , ! ! ,- ! PORT DEAS' STEVESTON ELBOW SAND 2.7 km MANN ISLAND HEADS OFFSHORE Figure z7. Average suspended sediment concentrations at various depths and salinities along the Fraser River estuary, 5-8 April (upper) and 25-29 May (lower) z976. In April, the values decrease from Port Mann to Steveston and then remain constant to seaward; in contrast, during May, values increased downstream of Steveston, suggesting resuspension within an active turbidity maximum. Note that vertical scale for May is zo times that for April. shelf. Combined, these actions should decrease the influx of sand to nearshore areas, and result in non-accumulation (perhaps erosion) on some tidal flats (and perhaps shorelines) (Luternauer & Murray, I973). Projected increases in dredging of river chazmels and further jetty/breakwater construction can only hasten this shift from natural state.

Acknowledgements This study was undertaken during z975-76 , when I was employed by the Geological Survey of Canada. Many people helped with various parts of the program. Much of my under-

189 Sedimentation in the Fraser River, Canada 633

standing of the Fraser River and its estuary come from discussions with various colleagues in Vancouver, particularly D. A. Dobson (Water Survey of Canada), Gall Ashley and W. H. Mathews (University of British Columbia), and J. L. Luternauer, J. E. Armstrong and D. L. Tiffin (Geological Survey of Canada). In addition, I am particularly grateful to D. A. Dobson for supplying unpublished data, without which the conclusions in this paper would have been difficult, if not impossible, to reach. Shipboard and analytical help of Lesley Simpson and Davis Swan are particularly appreciated, along with the assistance of J. L. Luternauer, G. Ashley, R. Gardner, G. Seid, M. Waskett-Meyers, R. Linden, J. B. Southard, and R. G. Jackson, III. Luternauer, Mathews, I. N. McCave, M. G. Fitzgerald, R. H. Meade, T. J. Conomos and D. G. Aubrey offered valuable suggestions on the manuscript, for which I thank them.

References Ages, A. & Woollard, A. x976 The tides in the Fraser estuary. Pacific Marine Science (Canada) Rept. 75-5, ioo pp. Allen, J. R. L. x973 Phase differences between bad configuration and flow in natural environments, and their geological relevance. Sedimentology ~o~ 3a3-329. Church, M. & ~,Vahlgran, R. I974 Reference materials on sedimentation and geomorphology of the lower Fraser River. Dept. Geogr., Univ. British Columbia, xz5 pp. (mimeogr.). Haushild, ~,V. L., Stevens, H. H., Jr., Nelson, J. L. & Dempster, G. R., Jr. x973 Radionuclides in transport in the CoIumbia River from Pasco to Vancouver, x~Vashington, U.S. GeoI. Survey. Prof. Paper 433-N. Hodgins, D. O. x974 Salinity intrusion in the Fraser River, British Columbia. Dept. Civil Eng., Univ. British Columbia, PhD Thesis, I47 pp. Hoos, L. M. & Packman, G. A. x974 The Fraser River estuary--status of environmental knowledge to x974. Environ. Canada, Special Estuary Series No. x, 5x8 pp. Hubbell, D. ~V., Glenn, J. L. & Stevens, H. H., Jr. x97x Studies of sediment transport in the Columbia River estuary. Proe. Tech. Conf. on Estuaries of the Pacific Northwest. Eng. Exp. Station (Oregon State Univ.) Circular No. 42, x9o--226. Johnson, W. A. t9at Sedimentation of the Fraser River delta. Geol. Survey Canada, Memoir x25, 46 pp. Luternauer, J. L. & Murray, J. W. x973 Sedimentation on the western delta-front of the Fraser River, British Columbia. Canadian ffournal of Earth Sciences Io, 164z-x663. Mathews, ~V. H. & Shepard, F. P. x96z Sedimentation of Fraser River delta, British Columbia. Bulletin of the American Association of Petroleum Geology 46, x4x6-t438. Morton, K. W. t949 Fraser River system, Province of British Columbia: History o£1mprovements, x87 x to x948. Can. Dept. Publ. Works, Vancouver, 66 pp. Nordin, C. F., Jr. & Beverage, J. P. x965 Sediment transport in the Rio Grande, New Mexico. U.S. Geological Survey Prof. Paper 46a-F. Pharo, C. H. I972 Sediments of the central and southern Strait of Georgia, British Columbia. Dept. Geol., Univ. British Columbia, Ph.D. Thesis, 29o pp. Pretious, E. S. & Blench, T. x95x Final report on special observations of bed movement in lower Fraser River and Ladner Reach during x95o freshet. Natural Resources Council Canada Rept. xa pp. Simmons, G. E. & Buchanan, J. x955 A preliminary report on bank erosion on the lower Fraser River, British Columbia. Dept. Lands and Forests, Water Rights Branch, Water Res. Invest. Rept. ~78, 53 pp. Tiffin, D. L. x969 Continuous seismlc reflection profiling in the Strait of Georgia, British Columbia. Inst. Oceanogr. and Dept. Geophys., Univ. British Columbia, PhD Thesis, x66 pp. van Andel, Tj. H. x955 Sediments of the Rhone Delta. II. Sources and deposition of heavy minerals. Koninklijk Nederland. Geologic en Mijnbouw. Geol. Serie x5, 5x5-543. Water Su~'ey of Canada t974 Historical Streamflow Summary, British Columbia to z973. Inland "~Vaters Directorate, Ottawa, 694 Pp.

190 Mammal Rev. 2009, Volume 39, No. 3, 193–227. Printed in Singapore.

Distribution and movements of ﬁn whales in the

North Paciﬁc Oceanmam_147 193..227

SALLY A. MIZROCH*, DALE W. RICE*, DENNY ZWIEFELHOFER†, JANICE WAITE* and WAYNE L. PERRYMAN‡ *National Marine Mammal Laboratory, Alaska Fisheries Science Center, NMFS, NOAA, 7600 Sand Point Way, Bldg. 4, Seattle WA 98115, USA, †US Fish and Wildlife Service, Kodiak National Wildlife Refuge, 1390 Buskin River Road, Kodiak AK 99615, USA, ‡Marine Mammal Division, Southwest Fisheries Science Center, NMFS, NOAA, 8604 La Jolla Shores Drive, La Jolla CA 92037, USA

ABSTRACT 1. We summarize fin whale Balaenoptera physalus catch statistics, sighting data, mark recoveries and acoustics data. The annual cycle of most populations of fin whales had been thought to entail regular migrations between high-latitude summer feeding grounds and lower-latitude winter grounds. Here we present evidence of more complex and varied movement patterns. 2. During summer, fin whales range from the Chukchi Sea south to 35 °N on the Sanriku coast of Honshu, to the Subarctic Boundary (ca. 42 °N) in the western and central Pacific, and to 32 °N off the coast of California. Catches show concentrations in seven areas which we refer to as ‘grounds’, representing productive feeding areas. 3. During winter months, whales have been documented over a wide area from 60 °N south to 23 °N. Coastal whalers took them regularly in all winter months around Korea and Japan and they have been seen regularly in winter off southern California and northern Baja California. There are also numerous fin whale sightings and acoustic detections north of 40 °N during winter months. Calves are born during the winter, but there is little evidence for distinct calving areas. 4. Whales implanted with Discovery-type marks were killed in whaling operations, and location data from 198 marked whales demonstrate local site fidelity, consistent movements within and between the main summer grounds and long migrations from low-latitude winter grounds to high-latitude summer grounds. 5. The distributional data agree with immunogenetic and marking findings which suggest that the migratory population segregates into at least two demes with separate winter mating grounds: a western ground off the coast of Asia and an eastern one off the American coast. Members of the two demes probably mingle in the Bering Sea/Aleutian Islands area. 6. Prior research had suggested that there were at least two non-migratory stocks of fin whale: one in the East China Sea and another in the Gulf of California. There is equivocal evidence for the existence of additional non-migratory groups in the Sanriku-Hokkaido area off Japan and possibly the northern Sea of Japan, but this is based on small sample sizes.

Keywords: Balaenoptera physalus, discovery mark, migratory, resident, whaling

Mammal Review (2009), 39, 193–227 doi: 10.1111/j.1365-2907.2009.00147.x

Correspondence: S. A. Mizroch. E-mail: [email protected]

INTRODUCTION The fin whale Balaenoptera physalus is distributed nearly worldwide, and has discrete meta- populations in the North Atlantic, the North Pacific and the Southern Hemisphere (Mizroch, Rice & Breiwick, 1984). The distribution and population structure of fin whales in the North Pacific are rather poorly known. Fin whales were the targets of extensive commercial hunting in the 20th century, and they have been seen during many vessel and aerial surveys throughout the North Pacific; however, little information on their migratory patterns and possible stock separations has been summarized. In US marine mammal stock assessment documents produced by the National Marine Fisheries Service, Carretta et al. (2005) state that ‘there is insufficient information to determine a population structure’ for fin whales. Angliss & Lodge (2004) stated that fin whale stock structure is ‘equivocal’ and noted that, for conservative management purposes, three fin whale stocks are recognized: in the Northeast Pacific (Alaska), California-Oregon-Washington and Hawaii. The stock boundaries recognized by the International Whaling Commission (IWC) are very broad and were set in the absence of directed studies of fin whale stock identity in the North Pacific. For the purpose of setting catch limits only, the IWC recognized two stocks: in East China Sea and the rest of the North Pacific (Donovan, 1991). In the North Pacific, fin whales range throughout the temperate and subarctic waters, including the Bering and southern Chukchi seas (Mizroch et al., 1984). The distribution and movements of fin whales can be understood only with knowledge of their feeding ecology and reproductive patterns. Fin whales, like all baleen whales, are filter-feeders which depend upon food organisms such as euphausiids (krill) and small schooling fish that regularly aggregate into large swarms, schools or patches. Such patches are characterized by a high density of individuals, and in the North Pacific and other high-latitude regions occur mainly in the upper water layers less than 280 m from the surface (Brinton, 1962), although fin whales in the Mediterranean have been documented to dive to depths of at least 470 m (Panigada et al., 1999). It has been assumed by many observers that most populations of fin whales – like many species of baleen whales – adhere to a general seasonal pattern of migrating between high- latitude summer grounds and lower-latitude winter grounds (Mackintosh & Wheeler, 1929; Mackintosh, 1965). Specifically, Kellogg (1929) concluded that fin whales in the North Pacific migrated from Baja California to the Bering Sea. Annual migration may be related to the energetic advantages of calving in warmer and calmer seas (Kawamura, 1975), but this relationship is speculative and the reason for long-distance seasonal migrations in mysticetes remains unclear (see Corkeron & Connor, 1999; Clapham, 2001). Full-grown whales have sufficient insulation to maintain normal body temperature even in the coldest seas (Kanwisher & Sundnes, 1966), so there would seem to be little advantage in them migrating to lower latitudes. However, all whales must swim almost constantly in order to surface and breathe, so migrating to warmer waters costs little in terms of energy. There is evidence of more complex and varied movement patterns of fin whales in the North Pacific. Kawakami & Ichihara (1958) hypothesized the existence of an ‘American’ stock and an ‘Asian’ stock based on Discovery-mark recovery data. Fujino (1960) suggested that there were multiple migratory stocks of fin whales based on immunogenetic data, i.e. the analysis of blood group antigens. He found that whales sampled in an area north of the Aleutians had consistent antigen groupings year to year, but whales sampled near Kamchatka and whales sampled south of the Aleutians showed inter-annual fluctuations in such groupings. He concluded that this fluctuation indicated that different stocks may have passed through the area (and been hunted there) at different times in the summer. Nishiwaki

(1966) also described two migratory stocks of fin whales: an eastern group and a western group, recognized by Discovery mark returns and ‘serological means’. Fujino (1960) suggested that whales caught in the East China Sea were part of a local population that did not migrate to northern waters. In addition to his immunogenetic findings, he analysed unpublished data which indicated that fin whales from the East China Sea were different from other North Pacific fin whales in terms of growth rate, length at sexual maturity, external body proportions, shape of skull and shape and growth rate of baleen. Seasonality of sightings and genetic studies confirm that there is also a year-round resident population isolated in a peripheral sea in the Gulf of California (Tershy et al., 1993; Bérubé et al., 1998, 2002). Bérubé et al. (2002) analysed fin whale nuclear and mitochondrial DNA and demonstrated that these whales form an isolated population with a small effective population size. This is consistent with data from other ocean basins where some individuals and even some local populations, especially those in peripheral seas, do not follow a conventional migration schedule. Fin whales in the Mediterranean constitute a separate year-round non-migratory stock (Bérubé et al., 1998; Canese et al., 2006). In addition to populations that may follow conventional north-south migratory patterns in the North Atlantic, Jonsgård (1966) suggested that two separate populations of fin whales may feed in the waters near northern Norway: one during the early spring and the other during the summer months. Here we add to these initial insights by reviewing a large body of available data on the distribution and movements of fin whales in the North Pacific Ocean. We first provide some background information on the species’ prey, reproduction and life history. We then summarize data from catch records, opportunistic and directed sighting surveys, Discovery marking operations, and acoustics studies. By synthesizing these data, we summarize existing knowledge regarding North Pacific fin whale stock structure, as well as seasonal and migratory movements.

BACKGROUND TO THE SPECIES Size and morphology The fin whale is the second largest of all the whales, second in length only to the closely related blue whale Balaenoptera musculus. It is one of the Balaenopteridae, a family of mysticetes (baleen whales) whose members have fringed baleen plates in place of teeth. Balaenopterid whales, also known as rorquals, generally feed on swarms of small crustaceans or fish which they engulf by expanding their pleated throat grooves. They expel the water and the prey is caught in the fringes of the baleen plates. Northern Hemisphere fin whales are somewhat smaller than their southern counterparts, and in all populations females are slightly larger than males. In the North Pacific, the largest female reported was 82 ft (25.0 m) and the average length (both sexes combined) was about 60 ft (18.3 m) (Mizroch & Rice, 2006). Among 357 females measured by Rice at whaling stations in California, the largest was 75 ft (22.9 m), and only three others exceeded 70 ft (21.3 m). In the Southern Hemisphere, Mackintosh (1942) noted that the largest fin whale reported in the International Whaling Statistics was an 88-ft (26.8 m) female. However, those length data are not entirely trustworthy and he noted that the largest reliably documented female was 85 ft (25.9 m), and that among the 996 females measured by the Discovery Committee staff, none exceeded 80 ft (24.4 m). Morphologically, fin whales have a long, slender body form, and are sometimes said to be among the fastest of all the large balaenopterids. Because balaenopterid whales are fast swimmers and tend to sink when killed, most rorquals were unavailable to whalers until the

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 193 196 S. A. Mizroch et al. introduction of steam-powered catcher boats and the explosive harpoon in the late 1860s. As modern whaling spread from the North Atlantic in the late 19th century to the Antarctic and North Pacific in the early 20th century, whalers in some areas preferentially killed humpback whales Megaptera novaeangliae until they were commercially extinct, then turned to blue whales and fin whales (Tønnessen & Johnsen, 1982; Mizroch et al., 1984). In the 20th century, fin whales were hunted in larger numbers than any other species (Mizroch, 1983); 720 000 were killed in the Southern Hemisphere alone (Clapham & Baker, 2002).

Prey and feeding Fin whale prey species have been determined by analysis of stomach contents of whales captured during whaling. Therefore, knowledge of the prey species is limited by the seasonality and location of catches. Throughout the world, fin whales feed on euphausiids or ‘krill’ but they also consume substantial quantities of small nektonic fishes, at least locally and periodically (Nemoto & Kawamura, 1977). Nemoto (1959) analysed stomach contents data from 7505 fin whales killed between 1952 and 1958 in the northern part of the North Pacific and Bering Sea. Over that time span, he noted major variations in prey types and prey distributions, but in most cases, euphausiids were the most common foods in the Aleutians and the Gulf of Alaska, and schooling fishes predominated in the northern Bering Sea and off Kamchatka. Fish was the main food reported for fin whales north of 58 °N in the Bering Sea, mainly capelin Mallotus villosus, Alaska pollock Theragra chalcogramma and Pacific herring Clupea pallasi. Alaska pollock was also most common along the northern Bering Sea shelf edge. In the northern Bering Sea, Thysanoessa raschii was the only species of euphausiid taken by fin whales (Nemoto, 1959). Herring appeared to be the main food for fin whales along the Kamchatka coast of Russia (Zenkovich, 1934) and fin whales were called ‘herring whales’ by the early Russian whalers (Zenkovich, 1954). Sometimes the euphausiid T. raschii was taken together with capelin in the Gulf of Anadyr and Cape Navarin areas on the edge of the northwestern Bering Sea. In areas where the copepod Neocalanus cristatus was abundant, such as around the Aleutians and in Olyutorsky Bay off northeast Kamchatka, fin whales were found to take large quantities of this species (Nemoto, 1959). The fish consumed by fin whales in Arctic and subarctic waters were mainly capelin, walleye pollock, Pacific herring and (in the northern Bering Sea only) saffron cod Eleginus gracilis [Andriyashev, 1954 (1964 translation), 1955; Miller & Schmidtke, 1956; Hourston & Haegele, 1980; Frost & Lowry, 1986; Krieger, 1990]. All of these fish are small, mainly between 10 and 30 cm in length. In the Kuril Islands region (southwest of Kamchatka and northeast of Japan), the prey of 234 fin whales was examined from 1951 and in the period 1953–56 (Betesheva, 1961). Prey species varied greatly year to year and there was some evidence of regional differences in prey between the northern Kuril Islands and the southern areas. In the northern area, prey was mostly zooplankton such as krill (T. longipes, T. inermis and T. raschii) and Neocalanus copepods (N. plumchrus and N. cristatus). In the southern areas, squid Todarodes pacificus pacificus and small schooling fish such as Pacific saury Cololabis saira and Japanese anchovy Engraulis japonicus dominated. In the Subarctic waters around the Aleutian Islands and in the Gulf of Alaska, five species of krill were the predominant prey in the stomachs of animals killed during Japanese whaling: Euphausia pacifica, T. spinifera, T. inermis, T. raschii and T. longipes (Nemoto & Kasuya, 1965).

In the waters off the Canadian west coast, the prey of fin whales processed at the Coal Harbor whaling station from 1963 to 1967 was predominantly ‘euphausiids’ (not identified to species level), although in 1964 and 1965, copepods made up 13 and 37% of the diet, respectively (Flinn et al., 2002). To the south in the cool California Current, only two euphausiid species – E. pacifica and T. spinifera – were recorded as food for fin whales caught between 1959 and 1970 (Rice, 1977). Anchovies Engraulis mordax were the only species of fish regularly eaten by fin whales during the final years of whaling off California (1959–70; Rice, unpublished data), but sardines Sardinops sagax were probably taken before their population collapsed in the 1940s and 1950s. Both species congregate in large, dense schools in the epipelagic zone, and they are also among the most dominant species, at least locally, in terms of biomass. In the Gulf of California (Mexico), Nyctiphanes simplex was the dominant species in the diet of fin whales (Tershy, 1992). All of the species of krill eaten by fin whales are abundant, gather in dense swarms and live near the water surface, which makes them both vulnerable to predation by baleen whales and an energetically profitable food source (Brinton, 1962; Mauchline & Fisher, 1969; Brinton, 1976; Mauchline, 1980). Almost all of the main prey species of fin whales reach their greatest abundance in near-surface waters during the summer months, and in the highly productive areas of cooler waters found in higher latitudes and in eastern boundary currents. Therefore, the whales aggregate during summer in these areas, where they feed heavily. During the winter when food is sparse, the whales generally disperse to mostly unknown destinations where it has been assumed they fast or feed very little (Brodie, 1975).

Reproduction and life history Based on data from Antarctic populations, the 2-year breeding cycle of the fin whale appears to be synchronized with its annual migrations and feeding (Mackintosh & Wheeler, 1929; Mackintosh, 1942, 1965, 1966; Laws, 1961; Brown & Lockyer, 1984). Fin whales have traditionally been assumed to mate in the winter in lower latitudes, and then migrate back to their high-latitude summer feeding grounds. At the end of a gestation period of about 1 year, it has been assumed that the female gives birth to a single calf on or near the winter grounds, but since sightings on the winter grounds are rare this has never been confirmed. In the spring, mothers take their calves to the feeding grounds, fidelity to which appears to be maternally directed (Clapham & Seipt, 1991). Calves are probably weaned sometime before the end of summer or autumn. The complete breeding cycle thus occupies between 1.5 and 2 years, and the majority of females calve at 2-year intervals (Agler et al., 1993). Data from fin whales landed at California whaling stations from 1958 to 1970 indicated that the mean annual incidence of pregnancy among sexually mature females was 36% (Rice, unpublished data). As noted previously, there is some evidence to question the traditional assumption of a strict seasonal migration in fin whales, for at least some populations.

DATA SOURCES During the 1960s and early 1970s, national whale research projects throughout the North Pacific were coordinated by the North Pacific Working Group (NPWG), which was appointed by the Scientific Committee of the IWC. The NPWG consisted of four members, one each from the Fisheries Research Board of Canada (FRBC), the Whales Research Institute (WRI) in Japan, the All-Union Research Institute for Marine Fisheries and Ocean- ography (VNIRO) in the USSR, and the United States Fish and Wildlife Service’s Marine Mammal Biological Laboratory [precursor of the National Oceanic and Atmospheric

Administration’s (NOAA) National Marine Fisheries Service’s National Marine Mammal Laboratory (NMML)]. The NPWG developed standard formats for recording data and exchanged among themselves information on catches, effort, sightings and basic biological parameters such as age, sex and reproductive status. Although not part of the NPWG, the Instituto Nacional de Investigaciones Biologico Pesqueras (INIBP) in Mexico collaborated in some of the marking and sighting cruises. Data from the NPWG were used in a number of the papers cited here and form the basis for much of our review. We used four sources of data in our analysis of ﬁn whale distribution and movements: (i) catch statistics; (ii) sighting data; (iii) recoveries of whales marked for individual identiﬁca- tion; and (iv) acoustics data from offshore hydrophone arrays developed by the US Navy. Details of each of these sources are provided in the subsequent sections.

Catch statistics Fin whales were killed in the North Pacific by various nations until 1985; catches by nation and time period are summarized in Table 1. Such catch statistics are an extensive source of data on whale distribution because of their great volume, broad geographical scope and continuous coverage over many decades. These data were archived by the Bureau of Inter- national Whaling Statistics (BIWS) in Sandefjord, Norway. By 1981, these duties had been transferred to the IWC in Cambridge, UK. Additional catch records were available for the Japanese floating factory Tonan Maru, obtained by the Supreme Commander for the Allied Powers for the 1940 operations. These Tonan Maru statistics are important because they cover the only periods when modern-style whaling was conducted north of the Bering Strait. There were 49 936 reported fin whales kills in the North Pacific during the years 1911 to 1985 (C. Allison, pers. comm. BIWS/IWC Summary Catch Database Version 3.5, distributed on 11 February 2008). Reported catches of fin whales by each whaling nation are shown in Appendix S1. Of these, 48 040 records included information on date, location (latitude and longitude), body length and sex. There were 38 623 records with latitude and longitude reported to the nearest degree and minute, 1072 records with latitude and longitude reported to the nearest degree or half degree and 8574 records with latitude and longitude estimates. There were 1896 records with no latitude or longitude recorded at all, mainly from catches reported by the USSR in the 1960s. After the end of the Cold War in the early 1990s, Russian and Ukrainian biologists were able to reveal the original catch records for Soviet whaling fleets, which conducted a massive campaign of illegal whaling worldwide (including the North Pacific) from 1947 to 1972 (Yablokov, 1994; Yablokov & Zemsky, 2000). These records revealed wholesale falsification of catch data submitted by the USSR to the IWC, including over-reporting of some species to cover up illegal catches of other, protected species (see also Ivashchenko, Clapham & Brownell, 2006). The Soviet catch of fin whales in the North Pacific was under-reported in 1963 and 1964, and over-reported from 1965 to 1967. By 1968, the catch of fin whales, though still over-reported, was at a much lower level (Doroshenko, 2000), but no locality information was reported for the 1064 kills reported as fin whales in 1968 (see Appendix S1). Because the correct Soviet catch data are not in the IWC database, we have omitted all Soviet catches reported as fin whales during the years 1963–68 from the figures plotted here. Omitting these data does not change the overall pattern of seasonal concentrations and zones of high density, and ensures that whales other than fin whales are not shown on the distribution maps. For example, during the 1965–67 period when fin whale catches were over- reported, some whales were caught (and reported as fin whales) much further south than most

Table 1. Annual catches of ﬁn whales in the different whaling grounds, excluding the questionable Soviet data from the years 1963–1967

East Kamchatka/w. China Sanriku Bering Sea/w. E. Bering Sea/e. Gulf of California Year sea Hokkaido Aleutians Aleutians Alaska Vancouver and Mexico Total

1911 208 208 1912 187 187 1913 73 73 1914 88 88 1915 64 64 1916 82 82 1917 39 39 1918 50 69 119 1919 42 64 106 1924 148 148 1925 153 235 1 389 1926 214 197 5 416 1927 85 21 1 107 1928 52 47 1 100 1929 152 216 79 26 1 474 1930 28 21 49 1932 60 60 1933 61 61 1934 156 78 234 1935 150 117 32 20 3 322 1936 107 50 157 1937 170 14 184 1938 66 1 49 116 1939 91 91 1946 7 226 233 1947 15 242 17 274 1948 18 161 56 235 1949 1 269 56 115 441 1950 248 73 18 150 489 1951 381 75 225 681 1952 22 450 447 240 1 159 1953 12 416 607 181 1 216 1954 463 802 752 150 2 167 1955 226 362 212 1 212 120 2 132 1956 277 319 652 820 167 3 2 238 1957 186 251 381 1 101 284 22 2 225 1958 172 378 474 896 574 108 2 602 1959 80 270 795 787 372 108 2 412 1960 20 376 332 1 166 23 138 2 055 1961 9 218 201 905 425 118 1 876 1962 116 6 573 587 158 123 1 563 1963 67 8 1037 225 16 1 353 1964 64 56 55 223 701 169 147 1 415 1965 71 30 634 692 134 113 1 674 1966 19 85 165 336 765 134 42 1 546 1967 18 77 727 83 34 102 44 1 085 1968 53 484 184 61 38 820 1969 83 399 468 205 90 31 1 276 1970 104 269 258 219 146 16 1 012 1971 77 322 239 138 19 4 799 1972 34 55 419 25 70 155 758 1973 14 56 113 26 127 107 17 460 1974 2 36 76 1 129 147 22 413 1975 11 48 2 52 43 6 162 1980 4 4 1985 1 1 Total 1353 6193 8737 12 078 6631 4521 1137 40 650

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 197 200 S. A. Mizroch et al. of the other fin whales. These purported fin whales were caught in areas of high sperm whale Physeter macrocephalus catches, and were probably the latter species. In summary, the catch data analysed in this paper include 40 650 catch records with length and locality information reported for fin whales caught between 1911 and 1985 (Table 1). This total excludes 7390 questionable Soviet records for the years 1963 to 1967. Commercial whaling operations fell into two categories, coastal and pelagic. These are described further in the subsequent sections.

Coastal whaling Coastal whaling includes whaling operations carried out from shore-based stations, as well as operations from old-style factory ships. Reeves & Smith (2006) classified the latter operations in their category Factory Ship Whaling, together with the modern factory ships which had stern slipways for hauling whales on deck and which operated on the high seas. However, we consider this classification inappropriate: the old-style factory ships lacked slipways, so the whales were flensed in the water alongside the vessel. Of necessity they operated from protected anchorages along the coast, in essentially the same fashion as shore stations. From 1905 to 1971 there were many such whaling operations along the western coasts of North America, from Akutan in the eastern Aleutian Islands south to Bahia Banderas in the state of Jalisco, Mexico. North of 35 °N, this coastal whaling was almost entirely a summer fishery, operating between 1 May and 31 October (or from 16 April to 15 October at the central California stations from 1960 to 1968). The fishery off Mexico was prosecuted mainly in the winter, from December to May, but was occasionally extended from October to July. Rice (unpublished data) has compiled catch records and, where available, effort data for almost all of these coastal operations (Appendix S2). Reeves et al. (1985) summarized whaling results for pre-World War 2 catches at Akutan (1912–39) and Port Hobron (1926– 37) in Alaska. Gregr et al. (2000) summarized pre- and post-war results from the Canadian whaling stations. Clapham et al. (1997) published detailed catch records from the California shore stations from 1919 to 1926. From Alaska south to California, the catch was predominantly fin and humpback whales. In Mexico, however, only 12 fin whales were taken, all from December to June; the major targets there were humpback and blue whales. The effect of these pre-war (1905–40) catches on the whale populations has not been quantified, but it was probably great enough at least to reduce somewhat the numbers of fin whales in the eastern North Pacific prior to the advent of large-scale pelagic whaling. On the western side of the North Pacific, Japan had a centuries-old history of old-style whaling (Kasuya, 2002). Modern-style shore station whaling using catcher-boats with harpoon cannons was introduced to Japan in 1899, and it soon spread widely. Whaling by Japanese companies from 1911 to 1949 was summarized by Kasahara (1950) (see Appendix S3). A total of 83 different shore stations operated at various times in this period. They were spread from Paramushir (50 °N), the northernmost of the Kuril Islands, south to Taiwan (23 °N) and the Ogasawara (Bonin) Islands (27 °N). Kasahara (1950) classified Japanese whaling grounds into 16 areas, and presented total annual catches for each of these areas. Mizue (1950) also presented total annual catches for all of these areas combined (Appendix S4). The numbers published by Kasahara and Mizue are in fairly close agreement for 1922 to 1937, but for the earlier years (1911 to 1921), many of Mizue’s figures are much higher than Kasahara’s. This can be explained by the fact that there were many missing data (indicated by ‘?’) in Kasahara’s tabulation, so we have accepted Mizue’s figures as the more accurate ones. Mizue (1950) plotted the catch position of each whale on a series of 12 monthly maps.

Most of these early Japanese catches have not been incorporated in the BIWS database, and we have not located any further data, such as sex and body length, on them. Kondo & Kasuya (2002) noted that some Japanese shore station catches of fin whales had been under-reported to the BIWS and they provided corrected statistics for shore stations operated by the Nihon Kinkai Hogei Company from 1965 to 1978. They noted that fin whale catches were generally under-reported for shore stations off the northern coast of Hokkaido (Mombetsu) and the eastern coast of Hokkaido in the northern Sea of Japan (Wakkanai). Also in the western North Pacific, a Russian company established a shore station at Hajdamak, 180 km east of Vladivostok, in 1889, but it closed in 1890 after its catcher-boat was lost at sea. The station was reopened by another company in 1895 and operated until 1901, when it was destroyed by fire. In 1899, the same company converted a vessel into a floating factory, the Michail; whaling commenced in 1903 but was brought to a halt in 1904 by the Russo-Japanese war (Berzin, 2008). Details of Russian catches in these early years are unavailable. After the Second World War, jurisdiction over the Kuril Islands was transferred from Japan to the Soviet Union. In 1948, the government reopened five of the former Japanese whaling stations. These were Podgorniy on Paramushir I., Skalistiy on Simushir I., Yasniy and Kasatka on Iturup I., and Ostrovhoy on Shikotan I. The stations on Simushir and Shikotan closed after the 1961 season, the other three after 1964. Catch reports from all five were submitted to the BIWS.

Pelagic whaling Modern-style harpoon-cannon whaling on the high seas was made possible when whalers working off South Orkney in the Antarctic invented the stern slipway during the 1912/13 whaling season, although purpose-built stern-slip factory ships did not appear until 1925 (Hart, 2006). Prior to the end of World War II, two pelagic whaling fleets operated in the North Pacific (Tønnessen & Johnsen, 1982). The Soviet floating factory Aleut was the first factory ship with a stern slipway to operate in the North Pacific. In 1933, the Aleut (then on her maiden voyage) and her fleet of three catcher-boats hunted whales while crossing the Pacific from Mexico to Vladivostok (Zenkovich, 1954). In subsequent years, until 1972, this fleet operated every summer in the western North Pacific from the Kuril Islands to the Chukchi Sea. Catch data for the years 1933 to 1935 were published by Tomilin (1937a); data for other years for the Aleut fleet are available from the IWC, although their reliability remains to be determined. Pelagic whaling was suspended during World War II, except, as noted previously, by the Aleut, which operated every year, and the Japanese floating factory Tonan Maru, which operated in 1940 and 1941 (Kasuya, 2002). Mizroch & Rice (2006) detail the development of post-World War II pelagic whaling, and describe the expansion of whaling starting in 1946. From 1946 to 1951, whaling occurred mainly around Japan, the Kuril Islands and in the Bering Sea near Kamchatka. Starting in 1947, pelagic whaling for baleen whales was restricted by the IWC to the 6-month season from May to October, and was limited to waters north of 20 °N in the western and central Pacific (west of 150 °W) and north of 35 °N in the eastern Pacific. North of the Bering Strait, whaling was allowed north to 72 °N in the sector between 150 °E and 140 °W. In this paper, we therefore broadly describe the observational effort involved in the pelagic catch records for the post-war period. Beginning in 1952, the number of Japanese and Soviet factory ships working in the North Pacific began to increase and whaling expanded geographically to include the central Bering

Sea, Aleutian Islands, and a small amount of whaling in the Gulf of Alaska. During the peak years from 1963 to 1967, six pelagic fleets (three Japanese, three Soviet) were hunting baleen whales throughout the North Pacific. After the 1975 season, the IWC banned the killing of fin whales in all oceans.

Sighting data Sighting data are much less extensive than catch data. Most sighting or census cruises are geographically limited, and many are one-time events. Unless a cruise is speciﬁcally dedicated to cetacean research, any sighting data obtained are difﬁcult to quantify. Nevertheless, for many times and places, sighting records are the only data available.

Japanese sighting data Japanese sighting data, which represent the most extensive body of sighting information available for the North Pacific, come from two sources. The first is data from the dedicated scouting vessels that operated throughout the North Pacific pelagic whaling grounds from 1964 to 1988. Because these vessels were attached to the whaling fleets, their temporal and geographical coverage largely paralleled that of those fleets. Methods were described by Ohsumi & Yamamura (1982). Because the vessel tracklines were not developed using line- transect methods and because perpendicular sightings distances cannot be computed from the data collected, the data do not allow the estimation of abundance, but they are useful to show seasonal distribution in the areas and seasons in which whalers operated. Other sightings were made from research vessels chartered by the Japanese Fisheries Agency from 1972 to 1990. From the scouting vessel data, Wada (1981) calculated an ‘Index of Abundance’ by 10-degree squares of latitude and longitude by year, all months combined. Miyashita, Kato & Kasuya (1995) combined the data from the scouting vessels and the research vessel data, and summarized them as sightings per mile by 5-degree squares by month, all years combined. It is not possible to estimate actual numbers from this type of dataset, but the summary is very useful to show overall seasonal distribution of whales throughout the North Pacific (and other areas).

Platforms of Opportunity Program (POP) The NOAA POP was initiated in 1975 as part of an environmental assessment programme conducted in Alaskan waters (Mercer, Krogman & Sonntag, 1978); the resulting data are managed by staff at NMML.1 The programme has received both opportunistic and directed marine mammal sighting data, primarily reported by personnel on vessels operating in the North Pacific. The database also contains records dating back to 1958 that were collected as part of pelagic studies conducted by NMML personnel. POP sighting data are opportunistically collected on NOAA, Navy, US Coast Guard and some fishing and tourist boats. Observer effort is dependent upon the interest of the observers and on their workload. POP data cannot be used to infer abundance, or even indices of abundance, and apparent gaps in distributions could just be an absence of sighting effort in certain areas. Sighting data are evaluated by trained data editors and species identifications are considered to be ‘confirmed’ only if the observer has described a plausible suite of sighting cues or provided other associated data (sketches, behavioral notes, etc.). Unconfirmed sightings were not used in this analysis.

1Data available upon request: National Marine Mammal Laboratory, 7600 Sand Point Way NE, Seattle, WA 98115, USA.

The POP database contains records of 1488 conﬁrmed ﬁn whale sightings during the period 1958–2000 (Appendix S5). Fin whales were seen in all months of the year.

Dale W. Rice, unpublished data Sightings of 92 ﬁn whales were recorded during 30 whale research cruises from 1958 to 1980 (Appendix S5). Data from these cruises covered the Eastern Paciﬁc from the northern Gulf of Alaska (60 °N) south to 12 °S, and west to the Leeward (northwestern) Hawaiian Islands.

Additional sighting data Additional sighting data include sightings of fin whales collected opportunistically by various observers (eight sightings), by personnel at the Kodiak National Wildlife Refuge, Kodiak Island, Alaska, USA from February to April during the years 1980–2001 (200 sightings), directed cetacean aerial surveys conducted in the Bering Sea in July each year from 1998 to 2000 by personnel from the Southwest Fisheries Science Center, La Jolla, California, USA (228 sightings), and directed cetacean aerial surveys conducted in the Gulf of Alaska in June–July 1998 and 2000 by personnel from NMML (95 sightings; Appendix S5). In addition, fin whale sightings were taken from reports of research cruises conducted off the US West coast (Forney & Barlow, 1993, 1998; Barlow, 1994, 1995; Forney, Barlow & Carretta, 1995; Barlow & Gerrodette, 1996). Fin whale sightings in the Bering Sea were summarized by Moore et al. (2002). Fin whale sightings off western Alaska and the central Aleutians were summarized by Moore et al. (2002). Zerbini et al. (2006) presented an abundance estimate of 1652 (95% CI: 1142–2389) and an estimated rate of increase of 4.8% (95% CI: 4.1–5.4%) for a small portion of the fin whale’s range in the North Pacific.

Recoveries of marked whales Extensive whale-marking programmes were undertaken by research groups from Japan (Omura & Ohsumi, 1964; Ohsumi & Masaki, 1975) and the Soviet Union (Ivashin & Rovnin, 1967). These research groups all used Discovery-type marks for tagging whales. The Discovery-type mark is a uniquely numbered steel tube 24-cm long tipped with a conical lead point 38-mm long. It was fired from a 12-gauge shotgun, propelled by a shotgun shell (lacking lead pellets). See Rayner (1940) for a description of the evolution of the Discovery mark and Brown (1977) for a review of the marking programmes. When properly implanted, it is completely buried in the blubber or muscle, and can be recovered only when the whale carcass is flensed; the unique number of each mark provided information on the locations of marking and recovery to be recorded. The number of whales marked with Discovery marks in each whaling area by national whaling fleet is shown in Table 2. Japanese researchers marked a total of 866 fin whales from 1949 to 1972. Japanese whale marking was conducted in the northern Sea of Japan, the Sanriku-Hokkaido area, the Kamchatka-western Aleutians grounds, the Bering Sea-eastern Aleutians grounds, the Gulf of Alaska, and the Vancouver grounds, with the majority of the marks placed in the Bering Sea-eastern Aleutians and Gulf of Alaska. According to Ivashin & Rovnin (1967), Soviet researchers marked a total of 51 fin whales from 1954 to 1966, although they account for only 43 fin whale marking events in their detailed tables of marks placed in each area by Soviet researchers, and they provide no explanation for the discrepancy; consequently, we used the number 43 when summarizing marks by area placed. Soviet whale marking was conducted near Kamchatka, in the Bering Sea, in the Gulf of Alaska, off Vancouver Island, off California, and in some other areas south of the major whaling grounds.

Table 2. Number of ﬁn whale marks and recoveries in each whaling area by each nation

Japan USSR USA Total Total Percent Area marked (1949–72) (1954–66) (1962–69) marked recovered recovered

US and Canadian West Coast 31 3 34 6 18% (II B) Gulf of Alaska (Yakutat to Alaska 182 6 188 45 24% Peninsula) (III B) Eastern Bering Sea 436 10 446 114 26% (IV A) South of Aleutians (160 °W-180) 104 3 107 16 15% (IV B) Western Bering Sea 34 5 39 7 18% (V A) South of Aleutians (180-160 °E) 53 3 56 9 16% (V B) Sea of Okhotsk 41 5360% (VI A) Sanriku-Hokkaido 19 2 21 2 10% (VI B) Sea of Japan 20 2150% (VII A) South of main whaling areas 10 1 0 (VI C) California and Baja California 56 56 11 20% Other areas 10 10 0 0 Total 866 43* 56 899 214 24%

*Soviet researchers marked a total of 51 ﬁn whales from 1954 to 1966, but they account for only 43 ﬁn whale marking events in their detailed tables of marks placed by area.

The US Marine Mammal Biological Laboratory conducted a programme of tagging whales with Discovery marks from 1962 to 1969. Because most whaling operations in the eastern North Paciﬁc were conducted between April and October, marking efforts were concentrated during the winter in low-latitude waters, mainly between 20 °N and 37 °N, in hope of providing evidence that would connect rorqual populations on the winter grounds with the stocks that were being exploited on the summer feeding grounds. In total, 56 ﬁn whales were marked by members of US expeditions (Table 2).

Acoustics data Thompson & Friedl (1982) monitored whale vocalizations received by two fixed underwater hydrophones moored near the northernmost point of Oahu in the Hawai’ian archipelago. Moore et al. (1998) and Watkins et al. (2000) analysed data on fin whale vocalizations received by sea bed-mounted and sound channel-mounted offshore hydrophone arrays. These arrays were deployed in much of the North Pacific between 25 °N and 55 °N by the US Navy Sound Surveillance System (SOSUS) (Wit, 1981; Richelson, 1998). Although the SOSUS system was created for military purposes, it regularly recorded fin and other whales, and in recent years these data have been made available to selected researchers. Stafford et al. (2007) analysed data from six Sound Fixing and Ranging (SOFAR) channel moored hydrophones deployed in the Gulf of Alaska.

RESULTS AND DISCUSSION General distribution and movements The geographical distribution of the total catches for all years combined (Fig. 1) show seven areas of great numbers of fin whale catches, which we will refer to as ‘historical whaling grounds’. For further analysis, we delimit these grounds as follows, from west to east: East China Sea Ground (120 °E to <135 °E). Sanriku Hokkaido Ground (135 °E to <155 °E). Kamchatka/western Bering Sea/western Aleutian Ground (155 °E to <180°). Eastern Bering Sea-eastern Aleutian Ground (180° to 160 °W). Gulf of Alaska Ground (<160 °W to 135 °W); this was called the ‘Northwest Coast Ground,’ which includes the ‘Kodiak Ground’. Vancouver Ground (<135 °W to 125 °W). California and Mexico Ground (<125 °W to 100 °W); this includes the Farallon Ground off central California and a few catches in the Pacific off Baja California. In addition, there is a Gulf of California Ground (20 °N to 30 °N inside the gulf) which is inhabited by a fin whale population that, based on seasonal observations and genetics studies, appears to be isolated (Tershy et al., 1993; Bérubé et al., 1998, 2002), but no whaling for fin whales was ever conducted in the Gulf. Combining data from all years reveals general geographical patterns, but conceals any year-to-year variations in whale or whaling distribution. The annual catch totals on each ground for each year from 1911 to 1985 (Table 1) show that there is considerable variation

Fin Whale Catches, all months combined 1911-1985 (Total catch = 40,650) 100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W

70°N 70°N

Gulf of Alaska Bering Sea

Ohhotsk Sea Kamchatka

Vancouver Ground Sanriku-Hokkaido 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N 10°N

Catch, in numbers of whales

8 11 1 4 -1 200 1 -41 - 844 2,68 0 1,000 2,000 4,000 6,000 19 - 54 55-1 2 - - 11 201 412 km 845 10°S 10°S

140°E 160°E 180° 160°W 140°W 120°W Source:BIWS/IWC catch database Version 3.5, 11 February 2008 release excluding USSR catches from 1963-1967

Fig. 1. Fin whale catch distribution by historical whaling ground, summed by 1° ¥ 1° square.

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 203 206 S. A. Mizroch et al. from year to year, which could be due in part to the distribution of the whaling fleets (the catch effort) as well as the distribution of whales. The pattern of fluctuations of whale densities may differ between grounds. To simplify our analysis of the distributional data, we divide the year into two seasons. ‘Summer’ (May to October inclusive) encompasses the months during which the extensive postwar pelagic whale fishery was conducted, as well as the numerous coastal whaling operations from 1905 to 1971 (Fig. 2a–f). ‘Winter’ (November to April inclusive) includes the months during which almost no pelagic whaling occurred; in the western North Pacific, there was a substantial long-term winter fishery for fin whales in coastal waters of Japan and Korea. The only winter fishery in the eastern North Pacific was that along the coast of Mexico, but very few fin whales were caught.

Summer range The most extensive dataset showing the distribution and numbers of whales on the summer grounds are the catch statistics (Fig. 2a–f). We produced catch maps that show monthly concentrations of whale catches using shaded density grids categorized using the ‘natural breaks’ algorithm (Jenks, 1963) within the program ArcMap 9.2. The Jenks algorithm groups similar values and selects break points based on natural groupings within each dataset. Because each catch map represents a different subset of the catch database, each density grouping gives the optimal class breaks for total catches within that month’s dataset. The pelagic whaling ﬂeets naturally centred the bulk of their hunting effort in those areas and months where experience had shown that higher densities of whales might be expected, although whaling would have been difﬁcult or impossible in the winter months in northern Arctic waters due to the inclement weather and ice. Because of these operational guidelines, raw catch numbers tend to exaggerate the temporal and spatial patterns of whale distribution, but remain the most comprehensive source of data on historical whale densities and distribution.

Northern limits During the summer, fin whales are found throughout the entire subarctic North Pacific (Fig. 2a–f). To the north, they range into the southern Sea of Okhotsk, the Bering Sea and the northern Gulf of Alaska. They pass through the Bering Strait into the southwestern Chukchi Sea during August and September. Many were taken as far west as Mys (Cape) Shmidta (68°55′N, 179°24′E), and as far north as 69°04′N, 171°06′W, by the Aleut fleet from 1933 to 1935, and by the Tonan Maru fleet in 1940. Zenkovich (1934, 1938a), a biologist who worked aboard the Aleut from 1933 to 1936, reported that ‘The Polar Sea, in areas near Cape Dezhnev which we visited, is frequented by large schools (literally hundreds of animals) of fin whales, humpbacks, and grays’. He further described (Zenkovich, 1938b) how the fin whales were ‘. . . encountered from early spring to the beginning of winter in groups of from 2 to 3 up to 10 individuals, mostly at a distance of 10 to 30 miles (16–50 km) from shore, but sometimes very near the shore (Seniavin Strait)’. Similar accounts were published by Tomilin (1937a,b), another biologist assigned to the Aleut fleet, who noted that ‘Cape Schmidt was the extreme western point where fin whales were observed in the Chukchi Sea in 1934’ (Tomilin, 1957). Nikulin (1946), who conducted observations from points along the coasts around the Chukotski Peninsula between 1937 and 1941, counted 336 fin whales from July to October; he said that the species ‘...isfound in groups in the Chukchi area from the second half of June to the end of October’. Sleptsov (1961), in a comprehensive review of the status of cetaceans in the Chukchi Sea, described the

(a)

100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W

70°N 70°N

Gulf of Alaska Bering Sea

Okhotsk Sea Kamchatka Vancouver 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N May catch, in numbers of whales 10°N

4 6 27 91 1-5 -1 - -4 - 131 6 - - 515 15 28 47 92 132

140°E 160°E 180° 160°W 140°W 120°W

(b)

100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W

70°N 70°N

Bering Sea Gulf of Alaska

Kamchatka Okhotsk Sea

Vancouver 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N June catch, in numbers of whales 10°N

10 84 17 -48 - 155 553 1 - 1-26 -3 1 27 49 8 - 85 - 156 31

140°E 160°E 180° 160°W 140°W 120°W

Fig. 2. Fin whale catches from 1911 to 1985, summed by 1° ¥ 1° square (panel). (a) May (n = 3490). (b) June (n = 10 204). (c) July (n = 12 854). (d) August (n = 8640). (e) September (n = 3529). (f) October (n = 1210). Source: BIWS/IWC catch database, version 3.5, 11 February 2008 release, excluding USSR catches from 1963–1967.

70°N 70°N

Gulf of Alaska Bering Sea

Okhotsk Sea Kamchatka

Vancouver 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N July catch, in numbers of whales 10°N

8 9 6 10 19 41 810 1 - 1-2 - 1 29 - 57 3- 58 - 102 10 200 - 417

140°E 160°E 180° 160°W 140°W 120°W

(d) 100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W

70°N 70°N

Gulf of Alaska Bering Sea

Okhotsk Sea Kamchatka

Vancouver 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N August catch, in numbers of whales 10°N

4 38 66 11 1 - 8 - 9 - 20 1- 2 39 67 - 113 114 - 21 215 - 5

140°E 160°E 180° 160°W 140°W 120°W

Fig. 2. (Continued)

(e)

100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W

70°N 70°N

Gulf of Alaska Bering Sea

Okhotsk Sea Kamchatka

Vancouver 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N September catch, in numbers of whales 10°N

3 9 18 1-5 -1 -24 -3 - 73 253 6 -1 14 25 40 74 119 -

140°E 160°E 180° 160°W 140°W 120°W

(f) 100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W

70°N 70°N

Gulf of Alaska Bering Sea

Okhotsk Sea Kamchatka

Vancouver 50°N 50°N

Sea of Japan California

30°N 30°N East China Sea Mexico

10°N 10°N October catch, in numbers of whales

20 36 72 38 1-4 - - 108 5-11 - 2 12 21 - 37 73 - 09 1

140°E 160°E 180° 160°W 140°W 120°W

Fig. 2. (Continued)

fin whale as ‘. . . one of the numerous baleen whales that inhabit the Chukchi Sea’ (Tomilin, 1937a,b; Zenkovich, 1937). He went on to state: ‘This species occurs from the Bering Strait to the arctic ice edge, in the coastal zone as well as in the open sea. It migrates as far west as Proliv Longa (175 °W to 180°), Wrangel (71 °N, 180°) and Herald (71 °N, 179°W) Islands, and prefers areas free of ice, but also occurs in pools of open water among ice floes’. Between 1969 and 1978, the crew of the catcher-boat Zvedny, while hunting gray whales, occasionally saw fin whales within a radius of 50 km of 67°20’N, 171°45’W, on the north coast of the Chukotski Peninsula (Votrogov & Ivashin, 1980). Many other Russian and Japanese writers have described the range of the fin whale as extending into the southwestern Chukchi Sea, without documenting any specific new records (Omura, 1955; Zenkovich, 1955; Arsen’ev, 1961; Berzin & Rovnin, 1966). However, on the latest whale sighting cruises from 1979 to 1992, no fin whales were seen in the Chukchi Sea, or anywhere north of the Gulf of Anadyr (Vladimirov, 1994). Fin whales have rarely been documented in the eastern half of the Chukchi Sea. During extensive aerial surveys for whales in the eastern Chukchi and western Beaufort seas, from 1979 to 1987 (Ljungblad et al., 1982; Ljungblad et al., 1988), live fin whales were seen only once: a group of three, including a mother with a calf, directly north of the Bering Strait at 67°10′N, 168°45′W on 24 July 1981. In August 1998, a stranded fin whale was reported 3.4 km northwest of the village of Wales in the Bering Strait (65°36′N, 168°5′W). A fin whale was seen on 2 July 2008 northeast of Cape Lisburne at 69° 13.8′ N, 165° 35.4′ during aerial surveys that were conducted in the northeastern Chukchi and western Beaufort Seas from June to November 2008 (Morse, NMML, unpublished data). No other sightings or strand- ings of fin whales have ever been reported from the coast of Arctic Alaska (Bee & Hall, 1956; Hall, 1981; Wynne, 1992)

Southern limits In the western North Pacific, Japanese shore stations took many fin whales during the summer months (April to November) around the southern Kuril Islands and Sakhalin, and off the north and east coasts of Hokkaido and the east coast of northern Honshu (these are areas 2, 3, 4 and 5 of Kasahara, 1950; Appendices S3 and S4). To the south of these areas, very few fin whales were taken in the summer (Kasahara, 1950; Mizue, 1950). In most of the North Pacific Ocean, the southern boundary closely approximates the Subarctic Boundary, as defined by Favorite, Dodimead & Nasu (1976). The Subarctic Boundary is the convergence zone between the Subarctic and North Pacific currents; in most of the central Pacific, it usually lies at about 42 °N. As the Subarctic Current approaches the North American coast, it splits at about 42 °N, approximately the latitude of the Oregon- California boundary in the USA. The northern branch flows anti-clockwise around the Gulf of Alaska, and constitutes the Alaska Gyre. The larger southern branch of the Subarctic Current continues south-southeasterly as the California Current; as it flows southward, it becomes warmer and more saline because of insolation and evaporation, and the boundary between it and the warmer water to the west rapidly dissipates. The southern boundary of the summer range of fin whales in the eastern North Pacific is deflected much farther south in the cool California Current, to about 32 °N. This is not evident in the pelagic catch data because hunting of baleen whales south of 35 °N was prohibited by IWC regulations in 1947. However, in the 1920s and 1930s, many fin whales were taken during the summer months by the catcher boats attached to the floating factories Lansing, California and Esperanza, when they were operating from anchorages on Catalina Island off southern California.

Farther south, off Mexico, there are almost no summer records of fin whales. On coastal cruises conducted by Rice in 1965, four fin whales were seen: one on 3 June, two on 12 June and one on 13 June, at 31°00′N, 24°35′N and 27°32′N, respectively, but none was encountered during a cruise from Mazatlan (23°11′N), Sinaloa, to the Islas Los Coronados (32°25′N), Baja California Norte, in September of the same year. The Norwegian factory ships that operated off Mexico did little whaling in May, June, July or October, and none in August and September. Their only summer catches of fin whales were single animals taken by the floating factory Esperanza, which took a 30 ft (9.1 m) female on 1 June 1928, a 65 ft (19.8 m) female on 17 May 1929 and a 57 ft (17.4 m) male on 30 May 1935. The male was taken off Bahia San Juanico (ca. 26°N), but available records on the other two indicate only ‘Mexico’s westkyst [west coast]’.

Monthly variation Monthly catch data indicate high numbers of whale catches on both sides of the North Pacific in May and June (Fig. 2a,b; in the Gulf of Alaska and near Kamchatka-western Aleutians), and high numbers of catches in the Bering Sea/eastern Aleutian area in July and August (Fig. 2c,d). This is consistent with an eastern and western stock of whales moving from low-latitude wintering areas in the eastern and western North Pacific, moving through feeding areas as spring and summer progresses, and feeding in a centralized area in July and August. In May, catches were distributed throughout the North Pacific, but catches were mostly sparse except for near the shore stations and in an area southeast of the Aleutian Islands (Fig. 2a). In June, catches were concentrated off Kamchatka, north and south of Unimak Pass in the Aleutians, and in the Gulf of Alaska (Fig. 2b). By July, catches were concentrated in the Unimak Pass area in the eastern Aleutian Islands and southern Bering Sea (Fig. 2c). In August, catches were spread along the Bering Sea shelf edge, all the way to Cape Navarin (Fig. 2d). In September, catches were concentrated at Unimak Pass, the Vancouver Island shore station and the Kuril Islands (Fig. 2e), and by October, catches were sparse except for near the Kuril Island shore station and in coastal Japanese waters (Fig. 2f). Distribution of fin whales has also been plotted using POP sighting data from vessel and aerial surveys and opportunistic sightings collected from 1958–2001. The POP sighting data are more limited in geographic coverage and cover a different range of years than the catch data, and sample sizes were small in most months (less than 150 sightings in total); however, both catch and sighting data sets show similar concentration areas and monthly variation in northerly waters during the summer months (Fig. 3a). In May, most of the sightings were in the Gulf of Alaska, but there were also some sightings off California and Oregon. In June, most of the fin whale sightings were in the Gulf of Alaska and Bering Sea, but there were also some sightings off Baja California from cruises conducted by Rice in 1965. In July, there were fin whale sightings in the Gulf of Alaska and Bering Sea, and what appears to be a large concentration of sightings in Bristol Bay in the Bering Sea. In August, there were fewer sightings in the Gulf of Alaska. Sightings in September and October were in similar locations, with fin whales seen in the Bering Sea along the shelf edge, in the Gulf of Alaska, and along the US coast, but one whale was seen farther south (off Baja California) in October. The fin whale was the most common large cetacean species seen during directed surveys in the central and eastern Bering Sea in 1999 (5 July–5 August) and the southeastern Bering Sea in 2000 (10 June–3 July; Moore et al., 2002). Fin whales were seen consistently and predictably during US Fish and Wildlife Service surveys in the vicinity of Kodiak, in the Gulf of Alaska, every month of the year except for December and January, when the surveys were not conducted.

(a) 100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W 80°W

70°N 70°N

50°N 50°N

30°N 30°N

10°N 10°N

Summer Fin Whale Sightings

4 8 1 1 - 5 - 11 2- 1 19 - 2728 - 36 37 - 103 0 1,000 2,000 4,000 6,000 Km 10°S 10°S

140°E 160°E 180° 160°W 140°W 120°W

(b)

100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W 80°W

70°N 70°N

50°N 50°N

30°N 30°N

10°N 10°N

Winter Fin Whale Sightings

4 6 1 - 2 3- 5 - 7 -5 8 - 14 15 - 4041 0 1,000 2,000 4,000 6,000 Km 10°S 10°S

140°E 160°E 180° 160°W 140°W 120°W

Fig. 3. Fin whale sightings from 1958 to 2001, summed by 1° ¥ 1° square (panel). (a) Summer sightings (May–October) (n = 1091). (b) Winter sightings (November–April) (n = 328). Source: NMML Platforms of Opportunity Data (1958–2000), NMML/NMFS unpublished data, Kodiak NWR unpublished data.

Sighting surveys were conducted in the North Pacific by Japanese scouting vessels from 1964 to 1990. From May to September, extensive surveys conducted in the Kamchatka, Bering Sea and Gulf of Alaska grounds showed patterns similar to the catch data, including high densities in June both off Kamchatka and in the Gulf of Alaska, and an increase in densities in the northern Bering Sea in August. In October, in surveys conducted in the Sea of Japan and off the Japanese islands, no fin whales were found (Miyashita et al., 1995). Sightings surveys conducted by personnel at the Southwest Fisheries Science Center confirmed the presence of fin whales off the California, Oregon and Washington coasts during summer and autumn (Barlow, 1994, 1995; Forney, Barlow & Carretta, 1995; Barlow & Gerrodette, 1996), but there were no confirmed fin whale sightings during a winter aerial survey in 1991 (Forney & Barlow, 1993). Forney & Barlow (1998) confirm that fin whales were significantly more abundant in California waters during summer months than during winter months.

Winter range It is difficult to infer winter (November to April) distribution of fin whales based on whaling records, because there are only 948 records of winter fin whales catches in the IWC catch database. Most of these are records of fin whales caught in April (742), a transitional winter/spring month. There was never any pelagic whaling during the winter months. On the western side of the Pacific Ocean, fin whales were caught near shore stations in winter but data from many of the early Japanese shore stations are not georeferenced and are not in the IWC catch database (Appendices S3 and S4). Kasahara (1950) reported a very small catch of fin whales in winter months only (peaks in March and April) from 1911–15 in the Sea of Japan near Usetsu (Area 10 in Appendix S3). The catch in 1911 totalled 61 whales; an unknown number of fin whales were caught from 1912–14; 11 fin whales were caught in 1915; 2 fin whales were caught in 1919, and no other fin whales catches were reported in this area until 1949, the last year reported in his paper (Appendix S4). Kasahara (1950) and Mizue (1950) reported that from 1916 until the mid-1930s, 100 to 200 fin whales were caught annually in the southern Sea of Japan, off the east coast of Korea, and in the Yellow Sea off the west coast of Korea (Areas 11 to 16 in Appendices S3 and S4). Fin whales were taken during every month of the year in these waters, but mainly during the winter months (Mizue, 1950). Based on the timing and distribution of these catches, it is possible that these whales came from a non-migratory population in the Sea of Japan (see later section on Discovery marks). Off the eastern (Pacific) side of Japan, there were very few winter catches of fin whales; almost no catches took place south of about 33 °N off southern Honshu, Shikoku and Kyushu. Some fin whales were caught during winter months off Kodiak Island (Alaska) and Vancouver Island (Canada), but most of these catches were in April, a transitional winter/ spring month. POP sighting records during the winter months (Fig. 3b) show that fin whales are distributed from the Bering Sea down to California. Data from November show sightings around Kodiak Island and in the Shelikof Strait area between Kodiak and the Alaska mainland. There were also sightings off southern California. In December, there was one sighting in the Bering Sea and the others were along Baja California and the Gulf of California. Fin whales were seen off Baja California in January and February, in the Bering Sea in January and February, and there was one sighting around the Aleutian Islands near Adak in February. In March, fin whales were still seen off Baja California and in the Bering Sea, but were also reported off California and Oregon, and were beginning to concentrate around Kodiak. In

April, fin whales were seen all along the US and Canadian coasts and in the Gulf of Alaska. They were concentrated around Kodiak, and were also seen in the Bering Sea. Moore et al. (1998, 2006), Watkins et al. (2000) and Stafford et al. (2007) examined acoustic data for fin whale calls monitored at various sites throughout the North Pacific. At most stations, fin whale calls were produced most frequently during the winter months (December to February), and almost ceased during mid-summer (June and July). This pattern was similar in the northwestern, north-central and northeastern North Pacific (40 °N-55 °N, 150 °E- 120 °W) as well as the southeastern North Pacific (25 °N-40 °N, 150 °W-120°W). In an area centred just south of the eastern Aleutian Islands, calls were most commonly recorded in July (Moore et al., 1998), and in the Gulf of Alaska, where fin whale calls were detected year-round, call numbers increased from July to a peak from October to December and then tapered off until March (Stafford et al., 2007). The annual pattern may primarily reflect a strong seasonal variation in the numbers of calls produced by fin whales, rather than variations in the numbers of whales occurring within range of the hydrophones. Watkins et al. (2000) indicated that the midwinter call sequences ‘appeared to function as male breeding displays’. In studies conducted in the Gulf of California, Croll et al. (2002) indicated that only male fin whales produce long, patterned low-frequency vocalizations. Because such displays are probably seasonal, being tied into the presumed winter mating period of this species, their absence at other times is probably not informative with regard to the actual distribution of fin whales.

Northern limits The conventional view is that fin whales desert their summer feeding grounds and shift to low latitudes for the winter months, although this was based primarily on equivocal observations made in the Southern Hemisphere (Mackintosh & Wheeler, 1929; Mackintosh, 1965). In the North Pacific, almost no whaling was conducted from November to May on the higher- latitude grounds. While it is likely that a scarcity or lack of whales was at least partly the reason, shorter days plus the much greater frequency of high winds and rough seas in the winter also played a major role in setting the whaling seasons. For the same reasons, whale sighting cruises in northern waters have been uncommon during the winter months. However, contrary to expectations, there were a number of sightings of fin whales in northerly waters during the winter. According to Zenkovich (1937), fin whales are seen year-round off the Commander Islands (north of 54 °N, west of 168 °E). In January 1963, a group of 20 fin whales was seen in the Gulf of Alaska at 58 °N, 148°03′W by the crew of a Soviet research vessel (Berzin & Rovnin, 1966). Fin whales are present during winter months around Kodiak Island in the Gulf of Alaska. Especially large concentrations occur during February, April and November, which in part may be due to the increased survey effort in winter months for ongoing winter seabird surveys. Fin whales are also present in the adjacent Shelikof Strait during winter. The Shelikof Strait is dominated by the Alaska Coastal Current which provides a tremendous vehicle for nutrient mixing and foraging opportunities for many marine species (Incze, Siefert & Napp, 1997). Prey presence and distribution is probably the reason for the presence of fin whales in the Shelikof Strait waters during the winter months. Acoustic detections also confirm the presence of substantial numbers of fin whales north of 40 °N throughout the winter months. Notably, fin whale pulses were detected year-round in the Gulf of Alaska; most calls were detected from August to February (Moore et al., 2006; Stafford et al., 2007). Although it has been shown that substantial numbers of fin whales may be found in higher latitudes all winter, it has not been demonstrated that any individual whales stay there

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 212 Fin whales in the North Paciﬁc Ocean 215 year-round. There are no sightings of ﬁn whales with small calves during the winter months, so there is no direct evidence that high-latitude areas are used for calving.

Southern limits Published sighting records of ‘fin whales’ in warm-temperate and tropical waters are all suspect because Bryde’s whales Balaenoptera edeni or brydei can be mistaken for fin whales. One instance is the report by Berzin (1978) of many ‘fin’ and ‘sei’ whales seen during the American/Soviet marking cruise of the catcher Vnushitelnyi in the tropical eastern Pacific in 1975; Rice (1979) identified all of these animals as Bryde’s whales. He also found that Berzin spent almost no time on watch on the bridge, but simply accepted the observations of the crew members, who, it turned out, were unaware of the existence of Bryde’s whales. From December to April, on the eastern side of the North Pacific, Rice found many fin whales from 35°30′N off the Big Sur coast of California south to 21°25′N off the coast of Nayarit, Mexico. The greatest numbers were encountered west of the Channel Islands off southern California. The factory ships operating off the west coast of Mexico between 1913 and 1935 reported taking only four fin whales between December and April. Rather than indicating a scarcity of fin whales, these low catches could be due to the abundance of blue and humpback whales, which were easier to catch. The POP data contain few if any con- vincing reports of fin whales in lower latitudes during the winter, even though the POP database includes many sightings of other species in lower latitudes during winter months. From the western North Pacific, Japanese shore stations took substantial numbers of fin whales during all months from September to May in the southwestern Sea of Japan off the east coast of Korea, in the Korea Strait, and in the Yellow Sea off the west coast of Korea [Areas 10, 11, 12, 13 and 14 of Kasahara (1950) (Appendix S3). To the north of this region, only a few fin whales were taken from December to March (Kasahara, 1950; Mizue, 1950)]. The pelagic waters across the Pacific between the western and eastern winter grounds have yielded almost no sightings of fin whales. In the Hawaiian Islands, Shallenberger (1981) reported a stranding at Kohakuloa, Maui, in the 1950s, and a sighting north of Oahu in May 1976, but provided no further details. Rice encountered one animal in the Kauai Channel (21°24′N, 158°23′W) on 16 February 1979. Berzin & Rovnin (1966) reported some whales at 37 °N, 138 °W in February 1964. K. C. Balcomb (Center for Whale Research, Friday Harbor, WA, USA) saw 8 to 12 individuals at 17°54′N, 158°48′W on 20 May 1966 (Rice, 1974) which is rather too late for ‘wintering’ animals. Based on data received by two fixed underwater hydrophones from December 1978 to April 1981 near the northernmost point of Oahu, Hawaii, Thompson & Friedl (1982) found that fin whale sounds were recorded most often from December to February, with a somewhat lower peak in August and September; none was recorded from May to July. Mobley et al. (1996) report a fin whale sighting off Kauai, Hawaii, in February 1994, and McDonald & Fox (1999) presented methods developed to estimate fin whale population density near Hawaii using one of the same hydrophones that Thompson and Friedl used; they confirmed that recent seasonality in call detections was similar to that detected in the Thompson and Friedl study, although fin whale calls were detected during only 12.5% of the recordings.

Calving areas The sizes of foetuses carried by fin whales taken during the summer months demonstrate that most calves must be born during the winter months (Rice, unpublished data); similar assump- tions are made about Southern Hemisphere fin whales (Mackintosh & Wheeler, 1929; Laws, 1959). However, we know of only one location in the North Pacific where calving has been

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 213 216 S. A. Mizroch et al. documented in lower latitude waters during the winter months. Rice has observed many ﬁn whales during the winter months in the waters off Mexico and California, but encountered only four adults accompanied by young calves, all in either late February or early March 1965, at locations ranging from 23° to 32 °N and 109° to 117 °W. This contrasts with his experiences with gray Eschrichtius robustus, humpback and Bryde’s whales, which were regularly accompanied by neonates in those same waters.

Movements of individual whales In all, 965 Discovery marks were ﬁred into ﬁn whales, and 214 marks (23.8%) were subsequently recovered (Table 2). In 16 cases, Discovery marks were found long after the whale’s carcass was processed, so reliable locations and dates of recovery were available for 198 individuals for subsequent analyses of seasonality and movement (Appendix S6). The sex of 164 of the 198 whales was reported. Fin whales were marked during 10 different months, and recovered mostly during the whaling season; 183 of 198 marking events, and 197 of 198 recoveries, occurred from May to September. Most of the marking was conducted in the Bering Sea, near the eastern Aleutians and the Gulf of Alaska (Table 2), and the majority of the recoveries were in those areas. The United States was the only whaling nation which had a programme to mark whales during the winter months (November–February) in low latitudes, and all of the recoveries of these whales occurred during the whaling season from May–August.

Movements of whales marked and recovered within the same whaling season Forty-two (21.2%) of the whales marked were killed during the same whaling season in which they were marked (Fig. 4). All but one of these whales were marked and recovered in the high-latitude feeding areas. Whale US-14 was marked in November 1962 off northern Baja California and killed south of the Queen Charlotte Islands in May 1963. The high proportion of recoveries within the first season reflects the fact that the seasonal movements of whalers into the whaling areas over the summer season were coordinated with the occurrence of the whales on these feeding grounds. Although many whales were recovered very near to the marking location, some whales moved substantial distances seasonally from west to east (JPN-10475: 1000 km and JPN-10773: 1891 km), east to west (JPN-9535: 2933 km) and south to north (US-14: 2434 km; Fig. 4). The median distance between the marking and recovery positions during the first season was 97 km (First Quartile: 46 km, Third Quartile: 339 km, minimum distance: 0 km, maximum distance: 2933 km).

Movements of whales recovered after the ﬁrst season There were 156 marks (78.8%) recovered at least one whaling season after marking. For these 156 marks, the median time span between marking and recovery was 3.04 years (First Quartile: 1.88 years, Third Quartile: 5.89 years; minimum time span 0.70 year; maximum 17.56 years). The shortest interval between marking and recovery was of whale JPN-6834, a male marked late in the whaling season in September in the Unimak Pass area and recovered the following May near the western end of the Aleutian chain, over 1700 km away from the marking site. The longest interval between marking and recovery was of whale JPN-151, a female marked in the Sanriku-Hokkaido area in July 1949 and killed in the same general area (less than 650 km away) almost 18 years later in February 1967. The median distance between the marking and recovery positions of whales recovered after the ﬁrst season was 449 km (First Quartile: 225, Third Quartile: 889, minimum distance: 40, maximum distance: 4842 km).

140°E 150°E 160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W 110°W 100°W

70°N 70°N

60°N JPN-6511 JPN-9172JPN-9165 60°N JPN-8618JPN-9269 JPN-9535 JPN-8632JPN-9456 JPN-9503 JPN-7720 JPN-8432JPN-8429 JPN-8649 JPN-8567 JPN-4560 JPN-9736 JPN-3369 JPN-6050 JPN-4266 JPN-6498 JPN-8416 JPN-10773 JPN-9898 JPN-8797JPN-8793 JPN-10475 50°N JPN-9579JPN-9577 50°N

JPN-11570/11575

40°N Month Marked 40°N

4 5 6 7 8 9 11 Month Recovered US-14

30°N 30°N 5 6 7 8 9 Males Females 0 1,000 2,000 4,000 Unknown km

20°N 20°N

160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W

Fig. 4. Whales marked and recovered within the ﬁrst season (n = 42). Source: Omura and Ohsumi, 1964; Ivashin and Rovnin, 1967; Ohsumi and Masaki, 1975; Rice, unpublished data.

Consistent movements Ohsumi & Masaki (1975) noted that on two occasions, two whales that were marked together at the same time were later recovered together at the same time and place some distance from where they were marked. One set was recovered after 23 days (JPN-8793 and JPN-8797, see Appendix S7), and the other set after 5 days (not included in Appendix S6 because precise location data were not available). The data presented here include an additional four sets of two or three whales that were marked together at one location on the summer grounds, and then recovered together at a different location at around the same time. In two cases, the whales were recovered 8–10 days or 23 days after marking (Appendix S7). In the other three cases, the whales were recovered 3–6 years after being marked (Appendix S7, Fig. 5). Also, in one of these cases, a third whale (JPN-8649) made an almost identical movement except that it had been marked 5 years later than the other two whales. There were an additional four sets of whales which showed similar movements, but were recovered in different years. The whales in each of these sets were marked on the summer grounds within a few days of each other, and were recaptured at some distance from the marking location, at nearly the same location and at the same time of year, but 2 to 16 years apart (Appendix S8, Fig. 5). In these cases, the distances between marking and recovery ranged from over 400 km to almost 3000 km (Appendix S8, Fig. 5). Three whales were marked on the winter grounds, two (US-616 and US-618) on the same day, the other one (US-414) nearby 3 days earlier. Two and half years later, two of them (US-414 and US-616) were recovered in summer at the same time and place, the third (US-618) close by 37 days earlier (Appendix S7, Fig. 5).

140°E 150°E 160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W 110°W 100°W

70°N 70°N

60°N 60°N JPN-9535JPN-9406 JPN-7043 JPN-6172JPN-8649 JPN-6977JPN-6976 JPN-7863JPN-7968 JPN-6208 JPN-6052 JPN-6060/6061 JPN-6044 JPN-10463/10481/10482 JPN-8797JPN-8793 50°N JPN-10475 JPN-8822 JPN-8815 50°N

40°N Month Marked 40°N

4 6 7 8 9 12 US-414 Month Recovered US-618US-616

30°N 30°N 6 7 8 Males km Females 0 1,000 2,000 4,000 Unknown

20°N 20°N 160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W

Fig. 5. Consistent movements (n = 22). Source: Omura and Ohsumi, 1964; Ivashin and Rovnin, 1967; Ohsumi and Masaki, 1975; Rice, unpublished data.

Some of these events may reﬂect long-term associations among individual whales, although the data are inadequate to assess whether the individuals concerned were socially associated at the time of marking and recovery. However, these data do suggest a tendency for groups of whales to conduct feeding migrations along similar routes, and that individual whales may exhibit seasonal as well as annual site ﬁdelity within and between seasons, presumably as long as prey distributions remain similar.

Seasonality of mark and recovery locations In order to explore the seasonality of mark and recovery locations independently of the number of years between marking and recovery, the number of ‘season-days’ between mark and recovery dates was calculated as the absolute value of the difference between the day of the year marked and the day of the year recovered. For example, whale JPN-3287/3343 (double-marked) was marked on 6 August 1954 and JPN-4658 was marked on 10 August 1954. Mark JPN-4658 was recovered on 6 July 1955 (330 days later) and mark JPN-3287/3343 was recovered on 2 July 1961 (2522 days later). However, in both cases there were 35 season-days between marking and recovery. During much of the summer whaling season (i.e. within the ﬁrst 80 season-days between marking and recovery), there is no apparent trend in season-days vs. distance between marking and recovery (Fig. 6). With the exception of a female whale which was marked near Kamchatka in May and recovered in the Gulf of Alaska in June (USSR-637: 4842.2 km), two whales marked near each other in the Gulf of Alaska in June and recovered near each other in the Aleutians in August, and two whales that travelled long distances (JPN 9406: 2881 km and JPN-9535: 2932.5 km), most whales killed within 80 season-days of marking were recovered less than 1500 km from their marking location (Fig. 6). The median distance between

Distance (km) between marking and recovery of marks (n=198)

5000 May-June April-August Feb-July Kamchatka-GOA Sea of Japan Sanriku-Hokkaido 4500 Mark No. USSR-637 Mark No. JPN-10498 Mark No. JPN-151

4000 US June-August Winter GOA-Aleutians Marking 3500 Marks No. JPN- Project 9535 and JPN 9406 3000 Marked or recovered June-August in Sept Kamchatka-Aleutians 2500 Mark No. JPN-10773 Distance in km 2000

1500

1000

500

0 0 1 2 3 5 7 8 10 12 14 15 17 23 23 25 28 29 30 32 34 36 37 38 40 42 45 50 51 55 57 60 62 66 69 74 82 91 127 152 175

Season Days between Mark and Recovery

Fig. 6. Distance between mark and recovery locations by ‘season-day’ (n = 198). mark and recovery locations of the 175 whales caught within 80 season-days from marking was 324 km (First Quartile: 123 km, Third Quartile: 732 km, minimum distance: 0 km, maximum distance: 4842.2 km; USSR-637, see previous discussion). In contrast, the median distance travelled by the 23 whales of which the marks were recovered more than 80 season-days after marking was 1600 km [Fig. 6; First Quartile: 689, Third Quartile: 2375 km, minimum distance: 46 km (JPN-10498, marked in April and recovered in August in the Sea of Japan), maximum distance: 3253.3 km; US-78, marked in January off Baja California and recovered in June in the Gulf of Alaska; Fig. 7]. Eleven of those 23 whales were marked during the winter months off Baja California, and nine were either marked or recovered in September near Unimak Pass. These data conﬁrm seasonal migrations north from wintering areas off Baja California to feeding areas along the US west coast and Gulf of Alaska. In addition, Unimak Pass appears to be a concentration area for ﬁn whales in September. Two of the ‘out-of season’ recoveries were marked and recovered in Japanese waters, one in the Sea of Japan and the other off Sanriku-Hokkaido, indicating the possible existence of non-migratory stocks in these areas. One (JPN-10498) was recovered after 127 season-days about 46 km from the marking location; the other (JPN-151) was recovered after 166 season- days 642 km from the marking location (Figs 6 and 7). Numbers of recoveries of whales marked in Japanese waters were low, and none of these recoveries documented long-distance movements to any of the other whaling grounds. This further supports the hypothesis that one or more non-migratory stocks exist that do not migrate to northern waters where the catch effort was much higher, but could also be explained by movements to areas where extensive whaling was not occurring. However, the small sample size constrains a detailed evaluation of these hypotheses.

130°E 140°E 150°E 160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W 110°W 100°W 90°W

60°N JPN-6097 60°N

JPN-4564 JPN-9239US-78 JPN-6808 US-18 JPN-7043 JPN-6042 JPN-6834 JPN-9577 US-14 JPN-6060/6061 US-613 50°N US-290 50°N

JPN-10498 US-26 JPN-151 US-221 US-616US-414

40°N 40°N US-6

30°N 30°N Month Marked

1 2 4 5 6 7 9 11 12 20°N Month Recovered 20°N

2 5 6 7 8 9 Females 10°N Males 10°N km Unknown 021,250 ,5005,000

150°E 160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W

Fig. 7. Recoveries more than 80 season-days (n = 23). Source: Omura and Ohsumi, 1964; Ivashin and Rovnin, 1967; Ohsumi and Masaki, 1975; Rice, unpublished data.

CONCLUSIONS Fin whales may be found throughout the entire North Pacific from the southern Chukchi Sea south to the Tropic of Cancer. Our distributional and mark-recovery data agree with early immunogenetic findings which suggest that the population is segregated into two major demes with separate winter mating grounds: one in the western and one in the eastern North Pacific. During the summer, at least some members of these two demes probably mingle in the Bering Sea-Aleutian Islands area. This corresponds with findings based on Antarctic marking data which suggested that fin whales from different Southern Hemisphere winter grounds intermingle on the higher-latitude feeding grounds (Brown, 1954, 1959, 1962). During summer (May to October), fin whales are found throughout the North Pacific, from the southern Chukchi Sea south to the Subarctic Boundary and the southerly reaches of the California Current. Catch densities and movements based on Discovery-mark recoveries show that individual whales may disperse longitudinally, some moving all the way from Kamchatka to the Gulf of Alaska, or vice versa, as the season progresses. It also appears that whales which moved from both the eastern and western North Pacific tend to concentrate in the Bering Sea-eastern Aleutian Islands area in July and August, and to move north in the Bering Sea along the shelf edge as the ice recedes in late summer. Some mark recoveries suggest that individual whales may follow the same routes and schedules in successive years, which indicates the possibility of predictable migrations and seasonal site fidelity to specific summer feeding grounds. Perhaps because they are fast swimmers, fin whales appear to be capable of rapid long- range movements, possibly to take advantage of widely dispersed prey concentrations. At least in the past when stomach content data were collected by whaling operations, fin whales

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 218 Fin whales in the North Pacific Ocean 221 in the western North Pacific appeared to switch prey from krill to fish as they moved north along the shelf edge in the Bering Sea and through the coastal waters of the Kamchatka Peninsula, presumably as the ice edge receded during the summer months. Based on median distances between marking and recovery locations within the whaling season, it appears that fin whales typically travel 200–800 km while foraging. In winter (November to April), it has been assumed that the bulk of the North Pacific population disperses southward toward the Tropic of Cancer, where mating and calving are presumed to take place, but specific low-latitude wintering areas remain undiscovered for the migratory whales. Coastal whalers took fin whales regularly during all winter months around Korea and southern Japan, and whales have been seen regularly during winter in the waters off southern California and Baja California. North of 40 °N there are numerous fin whale sightings and acoustic detections during winter months. Fin whales are seen predictably and regularly in winter months in the Kodiak region of the Gulf of Alaska, and some of the earliest winter surveys off the Commander Islands documented fin whales year-round. We cannot currently say whether these northern observations indicate: (i) that some individuals remain year-round on feeding grounds; (ii) that some whales undertake very short migrations in winter before returning north; and/or (iii) that the population structure of fin whales is more complicated than a simple migratory-movements model would suggest, so that varying patterns of occurrence and seasonal distribution are exhibited by different sections or classes of the population(s). There is good evidence from genetic and other data for the existence of discrete, non- migratory stocks of fin whales in the Gulf of California and in the East China Sea. Marking and seasonal catch data from the Sanriku-Hokkaido and northern Sea of Japan areas indicate the possible existence of additional non-migratory stocks of fin whales, but sample sizes are very small and further investigations must be conducted. Additional research is needed to answer key questions, including: 1. Where are migratory fin whales concentrated during the winter months? Are they distributed throughout the North Pacific, or are most of them in the waters immediately off the American and Asian coasts? Do they migrate to any particular areas to mate and calve? Extensive sighting, acoustic and satellite tagging studies could address these questions. 2. Do fin whales show site fidelity or philopatry to any particular areas? Although the Discovery mark data presented in this paper suggest habitat usage patterns, there are no other longitudinal data on North Pacific fin whale habitat use within or between seasons. Satellite tracking and photo-identification studies of individual whales could address these questions. 3. Are fin whale populations on the eastern and western winter grounds genetically differ- entiated? Are there two or more migratory stocks of fin whales in the North Pacific? Can the existence of additional non-migratory stocks in the waters near Japan and Korea be confirmed? Biopsy samples should be collected so that molecular genetic analyses can be performed to answer these questions.

ACKNOWLEDGEMENTS G. C. Pike of the FRBC, H. Omura of the WRI, and V. A. Arsen’ev and M. V. Ivashin (VNIRO), who served as members of the NPWG, collaborated in some of the research planning. A. A. Berzin (TINRO, Magadan), T. Kasuya (WRI, Tokyo), D. Lluch B. (INIBP, Mexico City), M. Nishiwaki (WRI, Tokyo), and V. A. Zemsky (VNIRO, Moscow) took part

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 219 222 S. A. Mizroch et al. in some of our marking and sighting cruises. T. Kasuya also provided many helpful suggestions and copies of Japanese publications. W. A. Watkins (Woods Hole Oceanographic Institution) let us see and cite his unpublished manuscript on fin whale phonations in the North Pacific. R. L. Brownell (Southwest Fisheries Science Center, Pacific Grove, California) supplied information on shore whaling by the USSR in the Kuril Islands. P. J. Clapham and J. Durban (NMML) provided many useful suggestions and criticisms of drafts of this manuscript. J. Benson, M. Cameron, J. Davies, J. Forbes and A. Greig (all at Alaska Fisheries Science Center, Seattle) provided useful advice on mapping, and R. Lowell Bush (NMML) assisted with data preparation and error-checking. Two other reviewers provided useful criticisms during the journal review process.

REFERENCES Agler, B.A., Schooley, R.L., Frohock, S.E., Katona, S.K. & Seipt, I.E. (1993) Reproduction of photographi- cally identified fin whales, balaenoptera physalus, from the Gulf of Maine. Journal of Mammalogy, 74, 577–587. Andriyashev, A.P. (1954 [1964 translation]) Fishes of the northern seas of the U.S.S.R. [Ryby severnykh morei SSR]. Opredeliteli po faune SSR [Keys to the Fauna of the USSR], Vol. 53, Izdatel’stvo Akademii Nauk SSR, Moskva-Leningrad, Translated by the Israel Program for Scientific Translations, Jerusalem, M. Artman. Andriyashev, A.P. (1955) A contribution to the knowledge of the fishes from the Bering and Chukchi seas. U. S. Fish and Wildlife Service, Special Scientific Report – Fisheries, 145, 1–81. Angliss, R. & Lodge, K. (2004) Alaska Marine Mammal Stock Assessments, 2003. US Department of Commerce. NOAA Technical Memorandum, NMFS-AFSC-144, Seattle, Washington, USA. Arsen’ev, V.A. (1961) Rasprostranenie kitov v Beringovom More i vozmozhnosti razvitiya kitoboinogo promysla (Distribution of whales in the Bering Sea and the prospects of development of whaling fisheries). Trudy Soveshchanii Ikhtiologicheskaya Komissiya, Akademiya Nauk SSSR, 12, 112–124. Barlow, J. (1994) Abundance of large whales in California coastal waters: a comparison of ship surveys in 1979/80 and in 1991. Reports of the International Whaling Commission, 44, 399–406. Barlow, J. (1995) The abundance of cetaceans in California waters. Part I: ship surveys in summer and fall of 1991. Fishery Bulletin, U.S., 93, 1–14. Barlow, J. & Gerrodette, T. (1996) Abundance of Cetaceans in California Waters based on 1991 and 1993 Ship Surveys. US Department of Commerce, NOAA Technical Memorandum, NMFS-SWFSC-233, La Jolla, California, USA. Bee, J.W. & Hall, E.R. (1956) Mammals of Northern Alaska on the Arctic slope. Miscellaneous Publications, Museum of Natural History University of Kansas, 8, 1–309. Bérubé, M., Aguilar, A., Dendanto, D., Larsen, F., Di Sciara, G.N., Sears, R., Sigurjonsson, J., Urban-R, J. & Palsboll, P.J. (1998) Population genetic structure of North Atlantic, Mediterranean Sea and Sea of Cortez fin whales, Balaenoptera physalus (Linnaeus 1758): analysis of mitochondrial and nuclear loci. Molecular Ecology, 7, 585–599. Bérubé, M., Urban, J., Dizon, A., Brownell, R. & Palsboll, P. (2002) Genetic identification of a small and highly isolated population of fin whales (Balaenoptera physalus) in the Sea of Cortez, Mexico. Conservation Genetics, 3, 183–190. Berzin, A.A. (1978) Whale distribution in tropical eastern Pacific waters. Reports of the International Whaling Commission, 28, 173–178. Berzin, A.A. (2008) The truth about Soviet whaling: a memoir. Marine Fisheries Review, 70, 4–56. Berzin, A.A. & Rovnin, A.A. (1966) Raspredelenie i migratsii kitov v severo-vostochnoi chasti Tikhogo Okeana, v Beringovom i Chukotskom moryakh (Distribution and migrations of whales in the north-east part of the Pacific Ocean, Bering and Chukchi seas). Izvestiya Tikhookeanskogo Nauchno-Issledovatel’skogo Instituta Rybnogo Khozyaistva i Okeanografii (TINRO), 58, 179–207. Betesheva, E.I. (1961) Pitanie promyslovykh kitov Prikuril’skogo raiona (Food of commercial whales in the Kuril region). Trudy Instituta Morfologiya Zhivotnykh Akademiya Nauk SSSR, 34, 7–32. Brinton, E. (1962) The distribution of Pacific euphausiids. Bulletin of the Scripps Institute of Oceanography, 8, 51–270. Brinton, E. (1976) Population of Euphausia pacifica off southern California. Fishery Bulletin, 74, 733–762. Brodie, P.F. (1975) Cetacean energetics, an overview of intraspecific size variation. Ecology, 56, 152–161. Brown, S.G. (1954) Dispersal in blue and fin whales. Discovery Reports, 26, 355–384. Brown, S.G. (1959) Whale marks recovered in the Antarctic seasons 1955/56, 1958/59, and in South Africa 1958 and 1959. Norsk Hvalfangst-Tidende, 12, 609–616.

Brown, S.G. (1962) Movements of blue and fin whales within the Antarctic Zone. Discovery Reports, XXXIII, 1–54. Brown, S.G. (1977) Whale marking: a short review. In: A Voyage of Discovery (Ed. by M. Angel), pp. 569–581. Pergamon Press, London. Brown, S.G. & Lockyer, C.H. (1984) Whales. In: Antarctic Ecology (Ed. by R.M. Laws), pp. 717–781. Academic Press, London. Canese, S., Cardinali, A., Fortuna, C.M., Giusti, M., Lauriano, G., Salvati, E. & Greco, S. (2006) The first identified winter feeding ground of fin whales (Balaenoptera physalus) in the Mediterranean Sea. Journal of Marine Biological Association of the United Kingdom, 86, 903–907. Carretta, J.V., Forney, K.A., Muto, M.M., Barlow, J., Baker, J.D. & Lowry, M.S. (2005) U.S. Pacific Marine Mammal Stock Assessments: 2004. US Department of Commerce. NOAA Technical Memorandum, NMFS-SWFSC-375, La Jolla, California, USA.. Clapham, P. (2001) Why do baleen whales migrate? A response to Corkeron and Connor. Marine Mammal Science, 17, 432–436. Clapham, P. & Seipt, I. (1991) Resightings of independent fin whales, Balaenoptera physalus, on maternal summer ranges. Journal of Mammalogy, 72, 788–790. Clapham, P., Leatherwood, S., Szczepaniak, I. & Brownell, R. (1997) Catches of humpback and other whales from shore stations at Moss Landing and Trinidad, California, 1919–26. Marine Mammal Science, 13, 368–394. Clapham, P.J. & Baker, C.S. (2002) Whaling, modern. In: Encyclopedia of Marine Mammals (Ed. by W.F. Perrin, B. Würsig & J.G.M. Thewissen), pp. 1328–1332. Academic Press, New York. Corkeron, P. & Connor, R. (1999) Why do baleen whales migrate? Marine Mammal Science, 15, 1228–1245. Croll, D.A., Clark, C.W., Acevedo, A., Tershy, B., Flores, S., Gedamke, J. & Urban, J. (2002) Only male fin whales sing loud songs. Nature, 417, 809. Donovan, G.P. (1991) A review of IWC stock boundaries. In: Reports of the International Whaling Commission, Special Issue 13 (Ed. by A.R. Hoelzel), pp. 39–68. International Whaling Commission, Cambridge, UK. Doroshenko, N. (2000) Soviet whaling for blue, gray, bowhead and right whales in the North Pacific Ocean, 1961–79. In: Soviet Whaling Data (1949–79) (Ed. by A. Yablokov & V. Zemsky), pp. 96–103. Center for Russian Environmental Policy, Marine Mammal Council, Moscow, Russia. Favorite, F., Dodimead, A.J. & Nasu, K. (1976) Oceanography of the Subarctic Pacific Region. International North Pacific Fisheries Commission Bulletin, 33, 1–187. Flinn, R.D., Trites, A.W., Gregr, E.J. & Perry, R.I. (2002) Diets of fin, sei, and sperm whales in British Columbia: an analysis of commercial whaling records, 1963–67. Marine Mammal Science, 18, 663–679. Forney, K.A. & Barlow, J. (1993) Preliminary winter abundance estimates for cetaceans along the California coast based on a 1991 aerial survey. Reports of the International Whaling Commission, 43, 407–415. Forney, K.A. & Barlow, J. (1998) Seasonal patterns in the abundance and distribution of California cetaceans, 1991–92. Marine Mammal Science, 14, 460–489. Forney, K.A., Barlow, J. & Carretta, J.V. (1995) The abundance of cetaceans in California waters. Part II: aerial surveys in winter and spring of 1991 and 1992. Fishery Bulletin, 93, 15–26. Frost, K.J. & Lowry, L.F. (1986) Sizes of walleye pollock, Theragra chalcogramma, consumed by marine mammals in the Bering Sea. Fishery Bulletin, 84, 192–197. Fujino, K. (1960) Immunogenetic and marking approaches to identifying subpopulations of the North Pacific whales. Scientific Reports of the Whales Research Institute, 15, 85–142. Gregr, E.J., Nichol, L., Ford, J.K.B., Ellis, G. & Trites, A.W. (2000) Migration and population structure of northeastern Pacific whales off coastal British Columbia: an analysis of commercial whaling records from 1908–67. Marine Mammal Science, 16, 699–727. Hall, E.R. (1981) The Mammals of North America, 2nd edn, Vol. 2. John Wiley & Sons, New York. Hart, I.B. (2006) Whaling in the Falkland Islands Dependencies 1904–1931: A History of Shore and Bay-Based Whaling in the Antarctic. Pequena, Newton St Margarets, Hertfordshire, UK. Hourston, A.S. & Haegele, C.W. (1980) Herring on Canada’s Pacific coast. Canadian Special Publication of Fisheries and Aquatic Science, 48, 1–23. Incze, L.S., Siefert, D.W. & Napp, J.M. (1997) Mesozooplankton of Shelikof Strait, Alaska: abundance and community composition. Continental Shelf Research, 17, 287–305. Ivashchenko, Y.V., Clapham, P.J. & Brownell, R.L., Jr. (2006) Scientific reports of Soviet whaling expeditions, 1955–78. US Department of Commerce, NOAA Technical Memorandum, NMFS-AFSC-175, Seattle, Washington, USA. Ivashin, M.V. & Rovnin, A.A. (1967) Some results of the Soviet whale marking in the waters of the North Pacific. Norsk Hvalfangst-tidende, 56, 123–135.

Jenks, G. (1963) Generalization in statistical mapping. Annals of the Association of American Geographers, 53, 15–26. Jonsgård, Å. (1966) Biology of the North Atlantic fin whale Balaenoptera physalus (L). Taxonomy, distribution, migration and food. Hvalrådets Skrifter, 49, 1–62. Kanwisher, J. & Sundnes, G. (1966) Thermal regulation in cetaceans. In: Whales, Dolphins and Porpoises (Ed. by K.S. Norris), pp. 397–409. University of California Press, Berkeley, California, USA. Kasahara, A. (1950) Nihon Kinkaino Hogei gyoto Sono Shigen (Whaling and whale resources off Japan). Bulletin of the Nihon Suisan Research Institute, 4, 1–103. (in Japanese) Kasuya, T. (2002) Japanese whaling. In: Encyclopedia of Marine Mammals (Ed. by W.F. Perrin, B. Würsig & J.G.M. Thewissen), pp. 655–662. Academic Press, San Diego, CA, USA. Kawakami, T. & Ichihara, T. (1958) Japanese whale marking in the North Pacific in 1956 and 1957. Norsk Hvalfangst-Tidende, 47, 285–291. Kawamura, A. (1975) A consideration on an available source of energy and its cost for locomotion in fin whales with special reference to the seasonal migrations. Scientific Reports of the Whales Research Institute, 27, 61–79. Kellogg, R. (1929) What is known of the migrations of some of the whalebone whales? Smithsonian Institution. Annual Report of the Board of Regents, 1928, 467–494. Kondo, K. & Kasuya, T. (2002) True catch statistics for a Japanese whaling company in 1965–78. Paper SC/54/O13 available from the International Whaling Commission, Cambridge UK. Krieger, K.J. (1990) Relationship between prey abundance and usage of Glacier Bay by humpback whales. In: Proceedings of the Second Glacier Bay Science Symposium (Ed. by A.M. Milner & J.D.Wood, Jr.), pp. 90–95. U. S. Department of the Interior, National Park Service, Alaska Regional Office, Anchorage, AK, USA. Laws, R.M. (1959) The foetal growth rates of whales with special reference to the fin whale, Balaenoptera physalus Linn. Discovery Reports, XXIX, 281–308. Laws, R.M. (1961) Reproduction, growth and age of southern fin whales. Discovery Reports, XXXI,327– 486. Ljungblad, D.K., Moore, S.E., Van Schoik, D.R. & Winchell, C.S. (1982) Aerial surveys of endangered whales in the Beaufort, Chukchi, and northern Bering seas. Naval Ocean Systems Center, San Diego, California, Technical Document, 486, 1–73, appendices A-B. Ljungblad, D.K., Wursig, B., Swartz, S.L. & Keene, J.M. (1988) Observations on the behavioral responses of bowhead whales (Balaena mysticetus) to active geophysical vessels in the Alaskan Beaufort Sea. Arctic, 41, 183–194. McDonald, M.A. & Fox, C.G. (1999) Passive acoustic methods applied to fin whale population density estimation. Journal of the Acoustical Society of America, 105, 2643–2651. Mackintosh, N.A. (1942) The southern stocks of whalebone whales. Discovery Reports, XXII, 197–300. Mackintosh, N.A. (1965) The Stocks of Whales. Fishing News (Books) Ltd, London. Mackintosh, N.A. (1966) The distribution of southern blue and fin whales. In: Whales, Dolphins, and Porpoises (Ed. by K.S. Norris), pp. 125–144. University of California Press, Berkeley and Los Angeles, CA, USA. Mackintosh, N.A. & Wheeler, J.F.G. (1929) Southern blue and fin whales. Discovery Reports, IXX, 285–296. Mauchline, J.S. (1980) The biology of mysids and euphausiids. Advances in Marine Biology, 18, 1–681. Mauchline, J.S. & Fisher, L.R. (1969) The biology of euphausiids. Advances in Marine Biology, 7, 1–454. Mercer, R.W., Krogman, B.D. & Sonntag, R.M. (1978) Marine mammal data documentation for the Plat- forms of Opportunity project and Outer Continental Shelf Environmental Assessment program; Northwest and Alaska Fisheries Center, Processed Report, Seattle, Washington, USA. Miller, D.J. & Schmidtke, J. (1956) Report on the distribution and abundance of Pacific herring (Clupea pallasi) along the coast of central and southern California. California Fish and Game, 42, 163–187. Miyashita, T., Kato, H. & Kasuya, T. (1995) Worldwide Map of Cetacean Distribution Based on Japanese Sighting Data, Vol. 1, National Research Institute of Far Seas Fisheries, Shimizu, Shizuoka, Japan, p. 134. Mizroch, S.A. (1983) Reproductive rates in Southern Hemisphere baleen whales. Master’s Thesis. University of Washington, Seattle, Washington, USA. Mizroch, S.A. & Rice, D.W. (2006) Have North Pacific killer whales switched prey species in response to depletion of the great whale populations? Marine Ecology Progress Series, 310, 235–246, Appendix. Mizroch, S.A., Rice, D.W. & Breiwick, J.M. (1984) The fin whale, Balaenoptera physalus. Marine Fisheries Review, 46, 20–24. Mizue, K. (1950) Surface water temperature and position of catch of each whale species, from 1911 to 1948. Japan Whaling Association, Tokyo.

Mobley, J.R., Smultea, M., Norris, T. & Weller, D. (1996) Fin whale sighting north of Kauai, Hawaii. Pacific Science, 50, 230–233. Moore, S., Stafford, K., Dahlheim, M., Fox, C., Braham, H., Polovina, J. & Bain, D. (1998) Seasonal variation in reception of fin whale calls at five geographic areas in the north Pacific. Marine Mammal Science, 14, 617–627. Moore, S.E., Waite, J.M., Friday, N.A. & Honkalehto, T. (2002) Cetacean distribution and relative abundance on the central-eastern and the southeastern Bering Sea shelf with reference to oceanographic domains. Progress in Oceanography, 55, 249–261. Moore, S.E., Stafford, K.M., Mellinger, D.K. & Hildebrand, J.A. (2006) Listening for large whales in the offshore waters of Alaska. Bioscience, 56, 49–55. Nemoto, T. (1959) Food of baleen whales with reference to whale movements. Scientific Reports of the Whales Research Institute, 14, 149–290. Nemoto, T. & Kasuya, T. (1965) Foods of baleen whales in the Gulf of Alaska of the North Pacific. Scientific Reports of the Whales Research Institute, 19, 45–51. Nemoto, T. & Kawamura, A. (1977) Characteristics of food habits and distribution of baleen whales with special reference to the abundance of North Pacific sei and Bryde’s whales. Reports of the International Whaling Commission, Special Issue, 1, 80–87. Nikulin, P.G. (1946) O raspredelenii kitoobraznykh v moryakh omyvayushchikh Chukotskiy Paluostrov. Izvestiya Tikhookeanskogo Nauchno-Issledovatel’skogo Instituta Rybnogo Khozyaistva i Okeanografii (TINRO), 22, 255–257. (in Russian) Nishiwaki, M. (1966) Distribution and migration of the larger cetaceans in the North Pacific as shown by Japanese whaling results. In: Whales, Dolphins and Porpoises (Ed. by K.S. Norris), pp. 171–191. University of California Press, Berkeley, CA, USA. Ohsumi, S. & Masaki, Y. (1975) Japanese whale marking in the North Pacific, 1963–72. Bulletin of the Far Seas Fisheries Research Laboratory, 12, 171–219. Ohsumi, S. & Yamamura, K. (1982) A review of the Japanese whale sightings system. Reports of the Interna- tional Whaling Commission, 32, 581–586. Omura, H. (1955) Whales in the northern part of the North Pacific. Norsk Hvalfangst-tidende, 44, 195–213. Omura, H. & Ohsumi, S. (1964) A review of Japanese whale marking in the North Pacific to the end of 1962, with some information on marking in the Antarctic. Norsk Hvalfangst-tidende, 53, 90–112. Panigada, S., Zanardelli, M., Canese, S. & Jahoda, M. (1999) How deep can baleen whales dive? Marine Ecology Progress Series, 187, 309–311. Rayner, G.W. (1940) Whale marking progress and results to December 1939. Discovery Reports, XIX, 245– 284. Reeves, R.R., Leatherwood, S., Karl, S.A. & Yohe, E.R. (1985) Whaling results at Akutan (1912–39) and Port Hobron (1927–37), Alaska. Reports of the International Whaling Commission, 35, 441–457. Reeves, R.S. & Smith, T.D. (2006) A taxonomy of world whaling. In: Whales, Whaling, and Ocean Ecosystems (Ed. by J.A. Estes, D.P. Demaster, D.F. Doak, T.M. Williams & R.L. Brownellpp), pp. 82–98. University of California Press, Berkeley, CA, USA. Rice, D.W. (1974) Whales and whale research in the eastern North Pacific. In: The Whale Problem: A Status Report (Ed. by W.E. Schevill), pp. 170–195. Harvard University Press, Cambridge, MA, USA. Rice, D.W. (1977) Synopsis of biological data on the sei whale and Bryde’s whale in the eastern North Pacific. Reports of the International Whaling Commission, Special Issue, 1, 92–97. Rice, D.W. (1979) Bryde’s whales in the equatorial eastern Pacific. Reports of the International Whaling Commission, 29, 321–324. Richelson, J.T. (1998) Scientists in black. Scientific American, 278, 48–55. Shallenberger, E.W. (1981) The status of Hawaiian cetaceans. Final report for MMC contract MM7AC028. Available from: National Technical Information Service, Springfield, Virginia. PB82-109398. Sleptsov, M.M. (1961) O kolebanii chislennosti kitov v Chukotskom More v raznyye gody. Trudy Instituta Morfologiya Zhivotnykh Akademiya Nauk SSSR, 34, 54–64. (in Russian) Stafford, K.M., Mellinger, D.K., Moore, S.E. & Fox, C.G. (2007) Seasonal variability and detection range modeling of baleen whale calls in the Gulf of Alaska, 1999–2002. Journal of the Acoustical Society of America, 122, 3378–3390. Tershy, B.R. (1992) Body size, diet, habitat use, and social-behavior of Balaenoptera whales in the Gulf of California. Journal of Mammalogy, 73, 477–486. Tershy, B.R., Urbán-Ramírez, J., Breese, D., Rojas-Bracho, L. & Findley, L.T. (1993) Are fin whales resident to the Gulf of California? Revista de Investigación Científica, 1, 69–72. Thompson, P.O. & Friedl, W.A. (1982) A long term study of low frequency sounds from several species of whales off Oahu, Hawaii. Cetology, 45, 1–19.

Tomilin, A.G. (1937a) Kity Dal’nego Vostoka [Whales of the Far East]. Uchenye Zapiski Moskovskogo Gosudarstvennogo Universiteta, Seriya Biologicheskikh Nauk, 13, 119–167. Tomilin, A.G. (1937b) Observations on Far-Eastern whales. Doklady Academii nauk SSSR (Comptes rendus de l’Academie des sciences de l’URSS), 14, 399–402. Tomilin, A.G. (1957) Kitoobraznye [Cetacea]. Zveri SSSR i prilezhashchikh stran [Mammals of the U.S.S.R and Adjacent Countries], Vol. IX, Izdatel’stvo Akademi Nauk SSSR, Moskva, English translation, 1967, Israel Program for Scientific Translations. Tønnessen, J.N. & Johnsen, A.O. (1982) The History of Modern Whaling. University of California Press, Berkeley and Los Angeles, CA, USA. Vladimirov, V.L. (1994) Recent distribution and abundane level of whales in Russian far-eastern seas. Russian Journal of Marine Biology, 20, 1–9. Votrogov, L.M. & Ivashin, M.V. (1980) Sightings of fin and humpback whales in the Bering and Chukchi seas. Reports of the International Whaling Commission, 30, 247–248. Wada, S. (1981) Japanese whaling and whale sighting in the North Pacific 1979 season. Reports of the International Whaling Commission, 31, 783–792. Watkins, W.A., Daher, M.A., Reppucci, G.M., George, J.E., Martin, D.L., DiMarzio, N.A. & Gannon, D.P. (2000) Seasonality and distribution of whale calls in the North Pacific. Oceanography, 13,1. Wit, J.S. (1981) Advances in anti-submarine warfare. Scientific American, 244, 31–41. Wynne, K. (1992) Guide to marine mammals of Alaska. Marine Advisory Bulletin 44. Alaska Sea Grant College Program. Yablokov, A. & Zemsky, V. (Eds) (2000) Soviet Whaling Data (1949–1979). Center for Russian Environmen- tal Policy, Moscow, Russia. (In Russian and English). Yablokov, A.V. (1994) Validity of whaling data. Nature, 367, 108. Zenkovich, B.A. (1934) Kitoboinyi promysel v Kamchatskom i Beringovom moryakh, sezon 1933. Rybnoye Khozyaistvo Dal’nego Vostoka, 1934, 113–118. (In Russian). Zenkovich, B.A. (1937) O migratsii kitov v severmoi chasti Tikhogo okeana. Izvestiya Tikhookeanskogo Nauchno-Issledovatel’skogo Instituta Rybnogo Khozyaistva i Okeanografii (TINRO), 10, 3–18. (In Russian). Zenkovich, B.A. (1938a) Kitoboinyi promysel v DVK. Priroda, 1938, 96–101. (In Russian). Zenkovich, B.A. (1938b) O kosatke ili kite-ubiitse. Priroda, 1938, 109–112. (In Russian). Zenkovich, B.A. (1954) Vokrug Sveta za Kitami. Gosudarstvennoe Izdatel’stvo Geograficheskoi Literatury, Moscow, Russia. (In Russian). Zenkovich, B.A. (1955) O migratsiyakh kitov. Promyslovye raiony v Dal’nevostochnykh vodakh. In: Kitooboi- nyi Promysel Sovetskogo Soyuza (Ed. by S.E. Kleinenberg & M.V. Makarova), pp. 51–68. Izdat, Zhurnala ‘Rybnoe Khozyaistvo’, Moscow, Russia. (In Russian). Zerbini, A.N., Waite, J.M., Laake, J.L. & Wade, P.R. (2006) Abundance, trends and distribution of baleen whales off Western Alaska and the central Aleutian Islands. Deep-Sea Research Part I, 53, 1772– 1790.

Submitted 26 February 2007; returned for revision 24 August 2007; revision accepted 30 April 2009 Editor: JD

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1. Fin whales killed by the main whaling nations between 1911 and 1985 and reported to the IWC. Appendix S2. Coastal whaling operations in the eastern North Pacific, and years for which monthly catch-per-unit-of-effort are available. Appendix S3. List of areas and numbers of shore stations in the western North Pacific from 1911 to 1949 (from Kasahara, 1950; Mizue, 1950). Appendix S4. Japanese shore station catches of fin whales for the years 1911–49 (from Kasahara, 1950). Appendix S5. Fin whale sightings by year and month [National Marine Mammal Labo- ratory’s POP database (1958–2000), NMFS unpublished data, Kodiak NWR unpublished data].

Appendix S6. Discovery mark recoveries where mark and recovery locations were reported. Appendix S7. Persistent associations of whales marked at around the same time and caught near each other at around the same time. Appendix S8. Seasonal movements to the same feeding areas in different years. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Journal compilation © 2009 Mammal Society, No claim to original US government works. Mammal Review, 39, 193–227 225 Contributed Paper Acoustically Detected Year-Round Presence of Right Whales in an Urbanized Migration Corridor

JANELLE L. MORANO,∗† AARON N. RICE,∗ JAMEY T. TIELENS,∗ BOBBI J. ESTABROOK,∗ ANITA MURRAY,∗‡ BETHANY L. ROBERTS,∗§ AND CHRISTOPHER W. CLARK∗ ∗Bioacoustics Research Program, Cornell Lab of Ornithology, Cornell University, Ithaca, NY 14850, U.S.A. ‡Current address: Cetacean Ecology and Acoustic Laboratory, School of Veterinary Science, University of Queensland, Gatton, Qld 4343, Australia §Current address: University of St. Andrews, School of Biology, Fife, KY16 8LB, U.K.

Abstract: Species’ conservation relies on understanding their seasonal habitats and migration routes. North Atlantic right whales (Eubalaena glacialis), listed as endangered under the U.S. Endangered Species Act, migrate from the southeastern U.S. coast to Cape Cod Bay, Massachusetts, a federally designated critical habitat, from February through May to feed. The whales then continue north across the Gulf of Maine to northern waters (e.g., Bay of Fundy). To enter Cape Cod Bay, right whales must traverse an area of dense shipping and fishing activity in Massachusetts Bay, where there are no mandatory regulations for the protection of right whales or management of their habitat. We used passive acoustic recordings of right whales collected in Massachusetts Bay from May 2007 through October 2010 to determine the annual spatial and temporal distribution of the whales and their calling activity. We detected right whales in the bay throughout the year, in contrast to results from visual surveys. Right whales were detected on at least 24% of days in each month, with the exception of June 2007, in which there were no detections. Averaged over all years, right whale calls were most abundant from February through May. During this period, calls were most frequent between 17:00 and 20:00 local time; no diel pattern was apparent in other months. The spatial distribution of the approximate locations of calling whales suggests they may use Massachusetts Bay as a conduit to Cape Cod Bay in the spring and as they move between the Gulf of Maine and waters to the south in September through December. Although it is unclear how dependent right whales are on the bay, the discovery of their widespread presence in Massachusetts Bay throughout the year suggests this region may need to be managed to reduce the probability of collisions with ships and entanglement in fishing gear.

Keywords: cetacean, critical habitat, diel, Eubalaena glacialis, passive acoustic monitoring, vocalization

Resumen: La conservacion´ de especies depende de la comprension´ de sus habitats´ estacionales y rutas migratorias. Ballenas francas (Eubalaena glacialis), enlistadas en peligro por el Acta de Especies en Peligro de E. U. A., migran del sureste de la costa de E.U.A a la Bah´ıa Cape Cod, Massachusetts, un habitat´ cr´ıtico designado federalmente, de febrero a mayo para alimentarse. Las ballenas luego continuan´ al norte por el Gofo de Maine hasta aguas nortenas˜ (e.g., la Bah´ıa de Fundy). Para entrar a la Bah´ıa Cape Cod, las ballenas deben atravesar un area´ de navegacion´ y actividad pesquera intensivas en la Bah´ıa Massachusetts, en donde no hay regulaciones obligatorias para la proteccion´ de ballenas o manejo de su habitat.´ Utilizamos registros acusticos´ pasivos de ballenas recolectados en la Bah´ıa Massachusetts entre mayo 2007 y octubre 2010 para determinar la distribucion´ espacial y temporal anual de las ballenas y su actividad sonora. Detectamos ballenas en la bah´ıa durante todo el ano,˜ en contraste con resultados de censos visuales. Las ballenas fuero detectadas en por lo menos 24% de los d´ıas de cada mes, excepto en junio 2007, cuando no hubo detecciones. Promediados en todos los anos,˜ los sonidos de ballenas francas fueron mas´ abundantes de

†email [email protected] Paper submitted March 4, 2011; revised manuscript accepted December 23, 2011. 698 Conservation Biology, Volume 26, No. 4, 698–707 C 2012 Society for Conservation Biology DOI: 10.1111/j.1523-1739.2012.01866.x 226 Morano et al. 699 febrero a mayo. Durante este per´ıodo, los sonidos fuero mas´ frecuentes entre las 17:00 y 20:00 tiempo local; no hubo patron´ nictimeral aparente en otros meses. La distribucion´ espacial de localidades aproximadas de sonidos de ballenas sugiere que pueden utilizar la Bah´ıa de Massachusetts como un conducto hacia la Bah´ıa Cape Cod en la primavera y conforme se mueven entre el Golfo de Maine y aguas del sur de septiembre a diciembre. Aunque no es claro que tanto dependen de la bah´ıa las ballenas, el descubrimiento de su presencia en la Bah´ıa Massachusetts durante todo el ano˜ sugiere que esta region´ puede requerir ser manejada para reducir la probabilidad de colisiones con barcos y de captura incidental por redes de pesca.

Palabras Clave: cet´aceo, Eubalaena glacialis h´abitat cr´ıtico, monitoreo acustico´ pasivo, nictimeral, vocal- izacion´

Introduction Massachusetts Bay does not have mandatory regulations or designated areas for protecting right whales because Understanding a species’ behavior and ecology is critical neither feeding aggregations nor other reproductive be- for developing effective conservation and recovery strate- haviors have been documented in the bay that would gies. Visual-survey data are the basis for current under- indicate the area is of importance for right whales. standing of migration patterns and seasonal feeding aggre- Massachusetts Bay is an urbanized coastal region; that gations of right whales (Eubalaena glacialis), a species is, it has high levels of noise and human activity (follow- listed as endangered under the U.S. Endangered Species ing Kraus & Rolland 2007), such as intensive fishing (e.g., Act. From February through October, right whales mi- Wiley et al. 2003) and commercial shipping (e.g., Hatch grate from waters off the coast of the southern United et al. 2008). To reduce the risk of ships colliding with States to waters off the coast of the northern United States right whales in this region, the Boston Traffic Separation and Canada to feed. From February through June, they are Scheme was established as a voluntary guide for directing sighted in Cape Cod Bay, Massachusetts Bay, the Great ship traffic to and from the Port of Boston through Stell- South Channel, and in the Bay of Fundy (e.g., Winn et al. wagen Bank National Marine Sanctuary (hereafter Stell- 1986; Hamilton & Mayo 1990; Kenney et al. 1995). Most wagen Bank), east of Massachusetts Bay (USCG 2006). right whales depart Cape Cod Bay by mid-May (Hamilton Because right whales must traverse Massachusetts Bay & Mayo 1990). In June and July, they move through the and the Boston Traffic Separation Scheme to move be- Gulf of Maine into the Bay of Fundy and the southeast- tween Cape Cod Bay and other southern or northern ern Scotian Shelf, where they typically remain through habitats, Massachusetts Bay is a migratory corridor (i.e., October until returning south for calving (Winn et al. an area used by an animal to move between habitats) 1986). (Rosenberg et al. 1997; Hess & Fischer 2001). Because migrating right whales move through areas of The periods of highest risk to right whales from hu- high human activity, they are exposed to multiple threats man threats in Massachusetts Bay occur from February to their survival. Collisions with ships and entanglement through May, when right whales move out of Cape Cod in fishing gear cause one-third of all right whale mortal- Bay to areas in the north (Winn et al. 1986; Hamilton ities (Kraus 1990; Knowlton & Kraus 2001; Kraus et al. & Mayo 1990; Nichols et al. 2008), and in October and 2005). To reduce mortality from these sources, areas have November, when right whales migrate south to waters been designated for the protection of right whales on the off Georgia and Florida (Winn et al. 1986; Hamilton & basis of migration patterns and locations of seasonal feed- Mayo 1990). Although most right whales likely use Mas- ing aggregations. These areas may be protected season- sachusetts Bay during these periods, there have been rare ally or year-round, and the legal requirements for right sightings of individuals in Massachusetts Bay, Cape Cod whale protection vary by area. Bay, and the Great South Channel in December, January, Critical habitat designation, determined by U.S. Fish and July (Winn et al. 1986; Kenney et al. 1995; Nichols and Wildlife Service or National Oceanic and Atmo- et al. 2008). These records reflect a change in seasonal spheric Administration (NOAA), is one type of legal pro- right whale occurrence in this region; historical whaling tection for endangered terrestrial and aquatic species. records show peak abundance of right whales in Mas- U.S. federal agencies must assess whether any of their sachusetts Bay was from November through March (Allen proposed actions in critical habitat may negatively affect 1908). Because the time when right whale occurrence is the species in question. Adjacent to Massachusetts Bay, at its peak can vary among years (Winn et al. 1986; Hamil- Cape Cod Bay and the Great South Channel are federally ton & Mayo 1990; Nichols et al. 2008), determining the designated critical habitats for right whales (NOAA 1994). year-round presence or absence of right whales in Mas- Cape Cod Bay and other areas in the Gulf of Maine also sachusetts Bay is crucial for their management in this have seasonal regulations for fishing gear, ship speed lim- region. its, and other regulations that reduce the risks of ships col- The National Oceanic and Atmospheric Administra- liding with right whales (NOAA 2002, 2008). However, tion, the agency responsible for the recovery of right

Conservation Biology Volume 26, No. 4, 2012 227 700 Right Whales in Massachusetts Bay whales, is reviewing a petition to designate the Gulf of All MARUs were programed to record continuously Maine and other regions as critical habitats (NOAA 2010). at a sampling rate of 2 kHz. Each MARU had a 10 Hz In the review of the petition, data supporting the tempo- high-pass filter to reduce electrical interference from the ral and spatial use of the Gulf of Maine by right whales will recording unit and an 800 Hz low-pass filter to prevent be used to determine critical habitat. In 2006, NOAA re- aliasing (i.e., distortion of the sound signal due to un- quired 2 companies operating offshore to monitor right dersampling). The MARUs had a flat frequency response whales and other marine mammals as part of the envi- of –151.2 dB (SD 1.0) (referenced to 1 µPa), between ronmental permitting requirements (NOAA 2007, 2009). 15 and 585 Hz. With these settings, MARUs recorded As a result, we assessed the presence of right whales in for 3–4 months/deployment, at which point they were Massachusetts Bay. refurbished and redeployed in the same location to pro- Data on the locations of right whales have been de- vide near-continuous, year-round recordings for 3 years rived primarily from visual surveys, which are logistically (Supporting Information). For analysis we synchronized constrained (Pittman et al. 2006). Thus, the presence of and concatenated data from all 19 MARUs into a 19- right whales in Massachusetts Bay and other waters may channel chronological sound file. We analyzed 1152 days be underestimated (following Clark et al. 2010). Long- of acoustic data recorded 23 May 2007 through 13 Oc- term passive acoustic monitoring is an effective method tober 2010. We used only days that had 24 continuous for detecting sound-producing animals in any environ- hours of recording (Supporting Information). ment (e.g., Mellinger et al. 2007a; Blumstein et al. 2011; We applied an automated-detection algorithm Farina et al. 2011) and can overcome many of the tech- (Urazghildiiev et al. 2009) to the recorded sound data nical and logistical limitations of visual surveys. Passive to identify E. glacialis contact calls. Contact calls, also acoustic monitoring is the recording of environmental referred to as up-calls, are the most common calls sound by an instrument that does not produce sound. produced by right whales and are used frequently to In contrast, with active acoustic monitoring (e.g., sonar), determine presence of right whales in an area (hereafter an instrument produces a sound and the sound’s echo vocal presence) (e.g., Clark 1983; Parks & Clark 2007; is analyzed (Urick 1983). Analyses of long-term passive Mellinger et al. 2011). The fundamental frequency acoustic recordings provide a cost-effective and labor- range of a contact call is approximately 100–400 Hz saving method to study areas where species have not been regularly sighted but still may occur (e.g., Mellinger et al. 2011) or to determine whether a species occupies a seasonal habitat longer than previously known (e.g., right whales present on the Scotian shelf through December [Mellinger et al. 2007b]). We used passive acoustic-monitoring data to document the spatial and temporal occurrence of calling right whales in Massachusetts Bay in an effort to provide information on whether Massachusetts Bay functions as a migratory corridor and whether increased regulations for this area may be warranted and consistent with management of right whales in surrounding areas.

Methods

Sound Recordings and Call Identiﬁcation Massachusetts Bay is in the Gulf of Maine (west of 70◦00W and between 42◦05Nand42◦40N, north of Cape Cod Bay) and is separated from the Atlantic Ocean on the east by Stellwagen Bank (Pittman et al. 2006). We recorded the environmental and biological sounds of Mas- sachusetts Bay on 19 archival recording devices, called Figure 1. Location of marine autonomous recording Marine Autonomous Recording Units (MARUs) (Calupca units (MARUs, black circles) in Massachusetts Bay et al. 2000), placed on the seafloor in a hexagonal grid ar- (MA Bay) and the approximate acoustic detection ray. The MARUs were approximately 9.3 km apart (Fig. 1). range in the recording area (white, dashed line) The array was approximately 38 km from west to east and (SBNMS, Stellwagen Bank National Marine Sanctuary; 32 km from north to south, covered an area of 893 km2, TSS, Traffic Separation Scheme [i.e., designated and had a perimeter of 111 km. shipping lanes, 4 lines]).

Conservation Biology Volume 26, No. 4, 2012 228 Morano et al. 701

(Urazghildiiev et al. 2009; Dugan et al. 2010), and a monthly vocal presence for each month over the entire contact call has a maximum source level of 168 dB data set by calculating the percentage of days with right (referenced to 1 μPa) (Parks & Tyack 2005). Analysts whale presence normalized by the total number of days trained in the identification of right whale contact calls sampled in the month. To determine the annual trend in reviewed all sound events detected by the detection monthly variation of vocal presence, data from all 3 years algorithm to confirm the event was a contact call (i.e., were merged to calculate the monthly average through- true detection). False detections, detections determined out a year normalized for the number of days sampled. to not be right whale contact calls, were removed from We performed more detailed analyses on 681 days analyses. A performance test of the detection algorithm (279,408 h), from 23 May 2007 through 21 May 2009, showed that it accurately detected 75% of contact calls to examine the potential diel and monthly patterns of (Dugan et al. 2010); thus, our reported detections can be right whale vocal activity. To compare the vocal activity considered a conservative estimate of the total number of right whales over time, we determined the total num- of recorded right whale calls. We could not quantify the ber of detected contact calls in each month. We analyzed number of individual whales or estimate the probability the relation between the number of calls detected per of failing to record a right whale call. It is possible that month and vocal presence in each month with a polyno- a whale was present, but did not vocalize, in which mial regression analysis in JMP 8.0 (SAS Institute, Cary, case it would not have been detected as acoustically North Carolina). We calculated diel vocal activity as the present (MacKenzie et al. 2002). However, a comparison total number of calls detected within each hour over each of data from coincident aerial surveys and passive sampled month. We report times of calls as local time in acoustic monitoring demonstrated that the probability of the Eastern Standard Time zone (–5 hours from Green- detecting a right whale is greater with passive acoustics wich Mean Time) and made no adjustment for daylight than with aerial surveys (Clark et al. 2010). saving time. A contact call produced by a right whale could have We did not locate calling whales with triangulation been recorded on multiple MARUs in the array. To elimi- (i.e., determining direction bearings from multiple sen- nate pseudoreplication of a call in our analyses, analysts sors); instead, we determined the area in which a whale reviewed all true detections and assigned the call to the was calling on the basis of the location of the MARU on first MARU to detect a contact call (i.e., the closest MARU which the call was first detected. To understand where to the whale), thus establishing vocal presence. We re- whales were within Massachusetts Bay throughout the moved calls recorded on other MARUs from the analyses year, we used the number of calls detected on MARUs and (Supporting Information). We used only the first detec- the locations of the MARUs to analyze the approximate tions of calls in our analyses. distribution of whales. The total number of calls detected The first detection of a call also provided the approxi- per month by each MARU, normalized for recording ef- mate location of the calling whale: within a radius of ap- fort (the number of calls multiplied by the percentage of proximately 4.7 km from the MARU, or half the distance the month recorded), for 681 days from 23 May 2007 to between MARUs. The detection range of the MARUs lim- 21 May 2009 was mapped in ArcGIS 10 (ESRI, Redlands, ited the maximum distance the calling right whale could California) to qualitatively compare the spatial distribu- be located outside the array. We did not measure the tion of whales among months. detection range of the MARUs; however, the detection range of an underwater device recording right whale contact calls in the Bay of Fundy was estimated as 20–30 km Results (Laurinolli et al. 2003). Therefore, we estimated that the Vocal Presence and Vocal Activity detection range of the MARU array was 25 km for right whales in Massachusetts Bay under typical conditions of Right whales were acoustically detected on at least 10% present-day background noise. Given this assumption, of days in every month from May 2007 through Octo- the recording area of our MARU array was approximately ber 2010, except in June 2007, when there were no 4000 km2, which covered Massachusetts Bay and over- detections (Fig. 2a). Right whales were acoustically de- lapped approximately 80% of Stellwagen Bank, but did tected on 80–100% of sampled days within each month not overlap Cape Cod Bay (Fig. 1). from March through June 2008, February through April 2009, and February through April 2010 (Fig. 2a). Eight months had vocal presence of 10–20% (August, Novem- Data Analyses ber, December 2007; September 2008; June, July, Au- We determined the daily vocal presence of right whales gust, December 2009) (Fig. 2a). The overall trend of for 1152 days from 23 May 2007 through 13 October monthly vocal presence showed a primary peak in the 2010. Daily vocal presence indicated that one or more spring (March through June 2008, February through May right whales were present within Massachusetts Bay and 2009, and February through April 2010) and a secondary produced a detected contact call. We determined the increase in the fall (September through October 2007,

Conservation Biology Volume 26, No. 4, 2012 229 702 Right Whales in Massachusetts Bay

Figure 2. (a) Percentage of days in each month and (b) mean (SE) daily percentage by month that right whale contact calls were detected in Massachusetts Bay, May 2007 through October 2010. The mean daily percentage by month is normalized for recording effort (the number of days with acoustic presence divided by the number of days sampled).

October through December 2008, September through month (Fig. 2b). Vocal presence was lowest in January, October 2009, and September 2010) (Fig. 2a). August, and December (<30% of sampled days within Averaged over the 42-month study, whales were vo- each month) (Fig. 2b). cally present >24% of days sampled within each month Vocal activity was highest from March through May (Fig. 2b). Peak presence occurred from February through 2008 and February through April 2009, when there were May; whales were vocally present on >60% of days sam- >1300 calls/month. This is a larger number of calls than pled within each month (Fig. 2b). There was a secondary other months by a factor of approximately 10 (Fig. 3a). peak in September and October, when whales were vo- The greatest number of calls (6015) occurred in April cally present on 45–47% of days sampled within each 2008, and the least number of calls occurred in Novem-

Conservation Biology Volume 26, No. 4, 2012 230 Morano et al. 703

Figure 3. (a) Total number of right whale contact calls detected in all sampled months and (b) the mean (SE) number of right whale calls in each month (normalized for recording effort, number of calls divided by the number of days sampled) in Massachusetts Bay, May 2007 through May 2009. ber 2007 (6 calls) (Fig. 3a). The monthly pattern of after this interval. Fewer than 500 calls were detected in right whale calls followed a bimodal trend in which each hour between 04:00 and 14:00. peaks occurred in March through April and November For February through May, months of high vocal ac- through December (Fig. 3b). September had the least tivity, most calls occurred between 17:00 and 20:00 vocal activity (average of 24 calls [SE 0]) (Fig. 3b). (Fig. 4b). The maximum number of calls in an hour (2411) February through May had high vocal activity (>800 was detected between 18:00 and 19:00, and the min- calls/month), and June through January had low vocal imum number of calls in an hour (215) was detected activity (<300 calls/month) (Fig. 3b). There was a signifi- between 10:00 and 11:00. Because hundreds to thou- cant positive relation between vocal activity and monthly sands of calls occurred from February through May and vocal presence (polynomial regression with second order only tens to hundreds of calls occurred in the other 2 fit, R = 0.65, F2,22 = 20.42, p < 0.001). months, the months with high vocal activity skewed the results that were based on analyses of the entire data set. Diel Vocal Activity Months with low vocal activity displayed a weak diel From May 2007 through May 2009 vocal activity in- pattern, which differed from the months with high vo- creased between 17:00 and 20:00 (Fig. 4a). We detected cal activity. During June through January, the maximum 16,789 right whale contact calls in 681 days of analyzed number of calls in any hour was 158, which occurred recordings, and more than 1,000 calls were detected each between 10:00 and 11:00 (Fig. 4c). Between 17:00 and hour between 17:00 and 20:00. The most calls (2,466) 20:00, 174 total calls were detected in January and June were made between 18:00 and 19:00, which is almost through December. The minimum number of calls (18) double the number of calls detected in hours before and occurred between 02:00 and 03:00 (Fig. 4c).

Conservation Biology Volume 26, No. 4, 2012 231 704 Right Whales in Massachusetts Bay

Figure 4. Number of right whale calls (a) per hour, (b) per hour during months (February–May) of high Figure 5. For right whales in Massachusetts (MA) Bay, vocal activity (>800 calls/month), and (c) per hour relative number of total calls detected on each marine during months of low vocal activity (June– autonomous recording unit (MARU) normalized by January) (radial axes, number of calls in increments the number of days of recording during months of (a) as indicated) in Massachusetts Bay, May 2007 high vocal activity (February–May) (small circles, through May 2009. 72–209 calls; medium circles, 302–453 calls; medium- large circles, 888–2840 calls; large circles, 3499–6844 calls) and (b) low vocal activity (June–January) Spatial Pattern of Calls (small circles, 6–25 calls; medium circles, 38–129 calls; Right whale calls were detected throughout Mas- medium-large circles, 227–237 calls; large circles, 587 sachusetts Bay in all months (Fig. 5). In months with high calls) (SBNMS, Stellwagen Bank National Marine and low vocal activity, all MARUs detected at least one Sanctuary; 4 Traffic Separation Scheme [TSS] lines, call, which indicates that whales were using the entire designated shipping lanes).

Conservation Biology Volume 26, No. 4, 2012 232 Morano et al. 705 area of the array and were not confined to Stellwagen Bay not only as a corridor to and from Cape Cod Bay, but Bank or areas closest to Cape Cod Bay (Fig. 5). During also as part of their nonmigratory habitat. February through May (months with high vocal activity), We did not quantify the number of right whales in the detections were predominantly in the eastern and south- area or determine whether the daily presence of right ern portions of the array, including within Stellwagen whales represented a single individual detected acous- Bank (Fig. 5a). The MARU in the southeast corner of the tically once or multiple individuals detected repeatedly array had the greatest number of detections (6844). The over time. However, we were able to determine there 3 MARUs within the Traffic Separation Scheme also had was at least one right whale for every day of vocal pres- a high number of detections (305–3499) (Fig. 5a). The ence. Given that the overall right whale population size lowest number of detections (72–302) occurred on the 7 is small and their probability of extinction is high (Kraus MARUs in the northwest quadrant of the array (Fig. 5a). et al. 2005), improved knowledge of spatial and seasonal The spatial distribution of right whale calls in January distribution and occurrence directly benefits efforts to and June through December (months with low vocal ac- prevent the loss of even one individual. tivity) differed from the distribution of calls in months Our demonstration of year-round presence of right with high vocal activity. During the months with low vo- whales in Massachusetts Bay could be used in evaluating cal activity, detections were predominantly in the north- the bay for designation as a region that warrants manda- ern edge and eastern half of the array (Fig. 5b). The 4 tory regulations to reduce the risk of whales colliding MARUs with the greatest number of detections (224–587) with ships and becoming entangled in fishing gear. Our were in Stellwagen Bank (Fig. 5b). The lowest number of data are shared with the Northeast Fisheries Science Cen- detections (6–38) occurred on 6 MARUs in the western ter and Stellwagen Bank National Marine Sanctuary, and half of the array (Fig. 5b). personnel may use our data to inform policy decisions. We suggest the following actions. First, on the basis of our results, regulators could determine whether the Gulf of Maine, inclusive of Massachusetts Bay, is a critical habitat Discussion for right whales (NOAA 2010). Although critical habitat designation does not, in itself, establish specific protec- Right whales were abundant off the coast of Mas- tions for an endangered species, the results of our study sachusetts from November to March until the middle of provide information on areas of high use by right whales, the 18th century (Allen 1908), but recent visual surveys and the seasonality of their vocal presence may help iden- rarely reported sightings in winter (Winn et al. 1986; tify habitats and lead to consideration of further protec- Pittman et al. 2006). In our acoustic survey, we found tive measures in the area. Second, current mandatory right whale vocal presence throughout the year, on at restrictions on ship speeds enacted in other areas may be least 24% of days in all months, and over the entire spa- appropriate either seasonally or year-round in this region tial extent of the study area in Massachusetts Bay. Histor- (NOAA 2008). Ship traffic is high in this region; thus, ical records and our findings support the idea that Mas- lower ship speeds would decrease the risk of collisions sachusetts Bay is an important habitat for the species. between ships and whales (Vanderlaan & Taggart 2009). Although we could not determine the behaviors of Third, seasonal or year-round restrictions on fishing gear right whales in Massachusetts Bay, our spatial and tem- could be applied to Massachusetts Bay to reduce the risk poral data on the distribution of right whale calls is con- of gear entanglement (e.g., Knowlton & Kraus 2001). sistent with their migratory patterns and suggests Mas- Due to its efficacy in detecting the presence of right sachusetts Bay may be primarily a migration corridor that whales, continuing passive acoustic monitoring in Mas- connects Cape Cod Bay with areas to the north in the Gulf sachusetts Bay and other known or potential habitats of of Maine and areas to the south. The estimated acoustic right whales would allow for a further understanding of detection range of the array, the seasonality of peak vo- seasonal and annual variability of right whale vocal pres- cal activity, and vocal presence along the eastern and ence. Acoustic observations could be conducted in con- southern edges of the array in February through May sug- cert with visual observations to determine the number gest right whales were feeding as they moved in and of individuals that enter Massachusetts Bay and how they out of Cape Cod Bay. Between June and January, vocal use the habitat (e.g., feeding, engaging in social or repro- detections along the eastern half and northern edge of ductive behavior). the array may represent right whales returning from the Our results could be used to develop acoustic monitor- northern Gulf of Maine region, entering or passing just ing strategies that minimize the time it takes to analyze outside Massachusetts Bay, but bypassing Cape Cod Bay, data and the cost of acoustic surveys for detecting the and perhaps beginning their southern migration between presence of right whales in Massachusetts Bay, and poten- September and December. However, given that right tially in other regions. We recommend the use of acoustic whales move into western Massachusetts Bay through- recordings to maximize the probability of detection dur- out the year, right whales may be using Massachusetts ing the times of year when visual surveys are not possible

Conservation Biology Volume 26, No. 4, 2012 233 706 Right Whales in Massachusetts Bay or in areas or periods of high human activity. Between illustrated method of identifying the first detection of a February and May, the period of peak vocal activity, the call on an MARU (Appendix S2) are available online. The probability of detecting right whale presence in the Bay authors are solely responsible for the content and func- was highest between 17:00 and 20:00, when vocal ac- tionality of these materials. Queries (other than absence tivity is highest. From June through January, months of of the material) should be directed to the corresponding low vocal activity, right whales exhibited no diel period- author. icity of calling; thus, to increase the probability of detecting right whales during these months monitoring should occur at all hours. These acoustic monitoring strategies, Literature Cited which capitalize on a species’ behavior, can be applied to Allen, J. A. 1908. The North Atlantic right whale and its near allies. a variety of taxa and areas to efficiently monitor species. Bulletin of the American Museum of Natural History 24:277–329. Passive acoustic monitoring can be used in a variety of Blumstein, D. T., et al. 2011. Acoustic monitoring in terrestrial environ- applications for species conservation. For right whales, ments using microphone arrays: applications, technological considerations and prospectus. Journal of Applied Ecology 48:758–767. acoustic recordings made over long periods could elu- Clark, C. W. 1983. Acoustic communication and behavior of the south- cidate whale occurrence patterns. These data could be ern right whale. Pages 163–198 in R. S. Payne, editor. Behavior and correlated with ecological and oceanographic features Communication of Whales. Westview Press, Boulder, Colorado. to help in the understanding of factors driving whale Calupca, T. A., K. M. Fristrup, and C. W. Clark. 2000. A compact digital habitat use and migration, or to determine the effects recording system for autonomous bioacoustic monitoring. Journal of the Acoustical Society of America 108:2582–2582. of acute and chronic noise exposure (Hatch et al. 2008; Clark, C. W., M. W. Brown, and P. J. Corkeron. 2010. Visual and acous- Clark et al. 2009; Van Parijs et al. 2009). Real-time passive tic surveys for North Atlantic right whales, Eubalaena glacialis, acoustic monitoring systems may reduce the acute threats in Cape Cod Bay, Massachusetts, 2001–2005: management implica- posed by ships to whales. Such systems could be used to tions. Marine Mammal Science 26:837–854. alert ships to the presence of right whales in an area Clark, C. W., W. T. Ellison, B. L. Southall, L. Hatch, S. M. Van Parijs, A. Frankel, and D. Ponirakis. 2009. Acoustic masking in marine and direct ships to reduce speed when year-round regu- ecosystems: intuitions, analysis, and implication. Marine Ecology lations on speed are not possible (Spaulding et al. 2010). Progress Series 395:201–222. The ability of passive acoustic monitoring to record any Dugan,P.J.,A.N.Rice,I.R.Urazghildiiev,andC.W.Clark. sound-producing animal and environmental noise condi- 2010. North Atlantic right whale acoustic signal processing: tions provides research opportunities for understanding Part I. comparison of machine learning recognition algorithms. IEEE Long Island Systems, Applications, and Technology 2010 the behavior and ecology of a variety of species, their in- DOI:10.1109/LISAT.2010.5478268. teractions with each other and their environments, and Farina, A., N. Pieretti, and L. Piccioli. 2011. The soundscape method- the effects of acute and chronic anthropogenic activities ology for long-term bird monitoring: a Mediterranean Europe case- on their survival. study. Ecological Informatics 6:354–363. Hamilton, P. K., and C. A. Mayo. 1990. Population characteristics of Acknowledgments right whales (Eubalaena glacialis) observed in Cape Cod and Mas- sachusetts Bays, 1978–1986. Report of the International Whaling We thank K. Conklin, S. Dedrick, C. Diamond, J. Evans- Commission Special Issue 12:203–208. Hatch,L.,C.Clark,R.Merrick,S.VanParijs,D.Ponirakis,K.Schwehr, Wilent, D. Fitzpatrick, K. Hodge, B. Howard, A. Izzi, M. M. Thompson, and D. Wiley. 2008. Characterizing the relative con- Loman, C. MacCarald, C. Muirhead, D. Nelson, C. Pelkie, tributions of large vessels to total ocean noise fields: a case study M. Pitzrick, A. Rahaman, E. Rowland, D. Salisbury, K. Van- using the Gerry E. Studds Stellwagen Bank National Marine Sanctu- nicola, A. Warde, and K. Wurtzell for their assistance with ary. Environmental Management 42:735–752. data analyses, without them the analyses would not have Hess, G. R., and R. A. Fischer. 2001. Communicating clearly about conservation corridors. Landscape and Urban Planning 55:195–208. been possible. The MARUs were deployed, retrieved, and Kenney, R. D., H. E. Winn, and M. C. Macaulay. 1995. Cetaceans maintained by C. Tremblay, with assistance from F. Chan- in the Great South Channel, 1979–1989: right whale (Eubalaena nell, D. Doxey, E. Moore, III, J. Michalec, and C. Tessaglia- glacialis). Continental Shelf Research 15:385–414. Hymes. Funding for this project was provided by Exceler- Knowlton, A. R., and S. D. Kraus. 2001. Mortality and serious injury ate Energy and Neptune LNG. The scope and direction of of northern right whales (Eubalaena glacialis)inthewestern North Atlantic Ocean. Journal of Cetacean Research and Manage- this project substantially benefited from discussions with ment 2:193–208. M.Brown,P.Corkeron,L.Hatch,S.Kraus,C.Mayo,S. Kraus, S. D. 1990. Rates and potential causes of mortality in North At- Van Parijs, and D. Wiley. We also thank 2 anonymous re- lantic right whales (Eubalaena glacialis). Marine Mammal Science viewers and the editors of this journal whose comments 6:278–291. strengthened the manuscript. Kraus, S. D., and R. M. Rolland, editors. 2007. The Urban Whale: North Atlantic Right Whales at the Crossroads. Harvard University Press, Cambridge, Massachusetts. Supporting Information Kraus, S. D., et al. 2005. North Atlantic right whales in crisis. Science 309:561–562. Laurinolli, M. H., A. E. Hay, F. Desharnais, and C. T. Taggart. 2003. The dates of the MARU configuration array deployments Localization of North Atlantic right whale sounds in the Bay of Fundy and the number of days analyzed (Appendix S1) and the using a sonobuoy array. Marine Mammal Science 19:708–723.

Conservation Biology Volume 26, No. 4, 2012 234 Morano et al. 707

MacKenzie, D., J. Nichols, G. Lachman, S. Droege, J. Royle, and C. Parks, S. E., and C. W. Clark. 2007. Acoustic communication: social Langtimm. 2002. Estimating site occupancy rates when detection sounds and the potential impacts of noise. Pages 310–332 in S. D. probabilities are less than one. Ecology 83:2248–2255. Kraus and R. M. Rolland, editors. The urban whale: North Atlantic Mellinger, D. K., K. M. Stafford, S. E. Moore, R. P. Dziak, and H. Mat- right whales at the crossroads. Harvard University Press, Cambridge, sumoto. 2007a. An overview of fixed passive acoustic observation Massachusetts. methods for cetaceans. Oceanography 20:36–45. Pittman, S., B. Costa, C. Kot, D. Wiley, and R. D. Kenney. 2006. Mellinger, D. K., S. L. Nieukirk, H. Matsumoto, S. L. Heimlich, R. P. Cetacean distribution and diversity. Pages 265–326 in T. Battista, R. Dziak, J. Haxel, M. Fowler, C. Meinig, and H. V. Miller. 2007b. Clark, and S. Pittman, editors. An ecological characterization of the Seasonal occurrence of North Atlantic right whale (Eubalaena Stellwagen Bank National Marine Sanctuary region: oceanographic, glacialis) vocalizations at two sites on the Scotian Shelf. Marine biogeographic, and contaminants assessment. Technical memoran- Mammal Science 23:856–867. dum NCCOS 45. National Oceanic and Atmospheric Administration, Mellinger, D. K., S. L. Nieukirk, K. Klinck, H. Klinck, R. P. Dziak, P. J. Silver Spring, Maryland. Clapham, and B. Brandsdottir.´ 2011. Confirmation of right whales Rosenberg, D. K., B. R. Noon, and E. C. Meslow. 1997. Biological corri- near a nineteenth-century whaling ground east of southern Green- dors: form, function, and efficacy. BioScience 47:677–687. land. Biology Letters 7:411–413. Spaulding, E., M. Robbins, T. Calupca, C. W. Clark, C. Tremblay, A. NOAA (National Oceanic and Atmospheric Administration). 1994. Waack, A. Warde, J. Kemp, and K. Newhall. 2010. An autonomous, Critical habitat for northern right whales. Federal Register near-real-time buoy system for automatic detection of North At- 50:28805–28835. lantic right whale calls. Proceedings of Meetings on Acoustics NOAA (National Oceanic and Atmospheric Administration). 2002. Tak- 6:010001–01000122. ing of marine mammals incidental to commercial fishing operations; USCG (U.S. Coast Guard). 2006. Port access routes study of potential Atlantic Large Whale Take Reduction Plan regulations. Federal Reg- vessel routing measures to reduce vessel strikes of North Atlantic ister 67:1142–1160. right whales. Federal Register 70:8312–8314. NOAA (National Oceanic and Atmospheric Administration). 2007. Urazghildiiev, I. R., C. W. Clark, T. P. Krein, and S. E. Parks. 2009. Small takes of marine mammals incidental to specified activities; Detection and recognition of North Atlantic right whale contact taking marine mammals incidental to construction and operation of calls in the presence of ambient noise. IEEE Journal of Oceanic an LNG facility off Massachusetts. Federal Register 70:11328–11335. Engineering 34:358–368. NOAA (National Oceanic and Atmospheric Administration). 2008. En- Urick, R. J. 1983. Principles of underwater sound. 3rd edition. McGraw- dangered fish and wildlife; final rule to implement speed restrictions Hill, New York. to reduce the threat of ship collisions with North Atlantic right Van Parijs, S. M., C. W. Clark, R. S. Sousa-Lima, S. E. Parks, S. Rankin, whales. Federal Register 73:60173–60191. D. Risch, and I. C. Van Opzeeland. 2009. Management and research NOAA (National Oceanic and Atmospheric Administration). 2009. Tak- applications of real-time and archival passive acoustic sensors over ing and importing marine mammals; operations of a liquefied varying temporal and spatial scales. Marine Ecology Progress Series natural gas port facility in Massachusetts Bay. Federal Register 395:21–36. 74:9801–9810. Vanderlaan, A. S. M., and C. T. Taggart. 2009. Efficacy of a voluntary area NOAA (National Oceanic and Atmospheric Administration). 2010. En- to be avoided to reduce risk of lethal vessel strikes to endangered dangered and threatened wildlife and designating critical habitat whales. Conservation Biology 23:1467–1474. for the endangered North Atlantic Right whale. Federal Register Wiley, D. N., J. C. Moller, K. A. Zilinskas, and D. N. Wiley. 2003. 75:61690–61691. The distribution and density of commercial fisheries and baleen Nichols, O. C., R. D. Kenney, and M. W. Brown. 2008. Spatial and tempo- whales within the Stellwagen Bank National Marine Sanctuary: ral distribution of North Atlantic right whales (Eubalaena glacialis) July 2001-June 2002. Marine Technology Society Journal 37:35– in Cape Cod Bay, and implications for management. Fishery Bulletin 53. 106:270–280. Winn, H. E., C. A. Price, and P. W. Sorensen. 1986. The distributional Parks, S. E., and P. L. Tyack. 2005. Sound production by North Atlantic biology of the right whale (Eubalaena glacialis) in the western right whales (Eubalaena glacialis) in surface active groups. Journal North Atlantic. Report of the International Whaling Commission of the Acoustical Society of America 117:3297–3306. Special Issue 10:129–138.

Conservation Biology Volume 26, No. 4, 2012 235 THE NATIONAL ACADEMIES PRESS

This PDF is available at http://nap.edu/21834 SHARE    

Spills of Diluted Bitumen from Pipelines: A Comparative Study of Environmental Fate, Effects, and Response (2016)

DETAILS

166 pages | 6 x 9 | PAPERBACK ISBN 978-0-309-38010-2 | DOI 10.17226/21834

CONTRIBUTORS

GET THIS BOOK Committee on the Effects of Diluted Bitumen on the Environment; Board on Chemical Sciences and Technology; Division on Earth and Life Studies; National Academies of Sciences, Engineering, and Medicine FIND RELATED TITLES

SUGGESTED CITATION National Academies of Sciences, Engineering, and Medicine 2016. Spills of Diluted Bitumen from Pipelines: A Comparative Study of Environmental Fate, Effects, and Response. Washington, DC: The National Academies Press. https://doi.org/10.17226/21834.

Visit the National Academies Press at NAP.edu and login or register to get: – Access to free PDF downloads of thousands of scientiﬁc reports  – 10% off the price of print titles  – Email or social media notiﬁcations of new titles related to your interests  – Special offers and discounts

Distribution, posting, or copying of this PDF is strictly prohibited without written permission of the National Academies Press. (Request Permission) Unless otherwise indicated, all materials in this PDF are copyrighted by the National Academy of Sciences.

236 Spills of Diluted Bitumen from Pipelines: A Comparative Study of Environmental Fate, Effects, and Response

Spills of Diluted Bitumen from Pipelines A Comparative Study of Environmental Fate, Effects, and Response

Committee on the Effects of Diluted Bitumen on the Environment

Board on Chemical Sciences and Technology

Division on Earth and Life Studies

THE NATIONAL ACADEMIES PRESS 500 Fifth Street, NW Washington, DC 20001

This activity was supported by Contract No. DTPH5614C00001 with the U.S. Department of Transpor- tation. Any opinions, findings, conclusions, or recommendations expressed in this publication do not necessarily reflect the views of any organization or agency that provided support for the project.

International Standard Book Number-13: 978-0-309-38010-2 International Standard Book Number-10: 0-309-38010-3 Library of Congress Control Number: 2015959867

Additional copies of this report are available for sale from the National Academies Press, 500 Fifth Street, NW, Keck 360, Washington, DC 20001; (800) 624-6242 or (202) 334-3313; http://www.nap.edu.

Printed in the United States of America

Suggested citation: National Academies of Sciences, Engineering, and Medicine. 2016. Spills of Diluted Bitumen from Pipelines: A Comparative Study of Environmental Fate, Effects, and Response. Washington, DC: The National Academies Press.

COVER (FRONT):

(top) Credit: John W. Poole/NPR An oil sheen appears along the shore of the Kalamazoo River in August 2012. In July 2010, more than 800,000 gallons of tar sands oil entered Talmadge Creek and flowed into the Kalamazoo River, a Lake Michigan tributary. Heavy rains caused the river to overtop existing dams and carried oil 30 miles downstream.

(bottom left) Credit: Jacqueline Michel Sorbents and booms deployed in Dawson Cove in response to a crude oil spill in Mayflower, Arkansas in April 2013. In March 2013, over 3,000 barrels of crude oil spilled from a rupture in the Pegasus pipeline spilling oil in a residential neighborhood and eventually into a heavily wooded cove.

(bottom center) Credit: Douglas Friedman Oil spill response workers shuttling oiled debris from the beach below. In May 2015, an estimated 100,000 gallons of heavy crude oil discharged from Plains All American pipeline 901 near Refugio State Beach in Santa Barbara County, California.

(bottom right) Credit: NOAA On November 26, 2004, the single-hulled tanker Athos I unknowingly struck a large anchor submerged in the Delaware River while preparing to dock at a refinery just outside Philadelphia, Pennsylvania. The impact punctured the tanker’s hull, and it began leaking more than 263,000 gallons of heavy oil into the tidal waters of this busy East Coast shipping route. A worker in protective gear power-washes the oily rocks while boom in background collects oil five months after the spill occurred.

COVER (BACK):

Credit: Jonathon Gruenke Jeremy Blackford of Clean Harbors uses a suction hose to clean oil from atop the Kalamazoo River in a containment area in Augusta, a village in Kalamazoo County in Michigan.

The National Academy of Sciences was established in 1863 by an Act of Con- gress, signed by President Lincoln, as a private, nongovernmental institution to advise the nation on issues related to science and technology. Members are elected by their peers for outstanding contributions to research. Dr. Ralph J. Cicerone is president.

The National Academy of Engineering was established in 1964 under the charter of the National Academy of Sciences to bring the practices of engineering to advising the nation. Members are elected by their peers for extraordinary contributions to engineering. Dr. C. D. Mote, Jr., is president.

The National Academy of Medicine (formerly the Institute of Medicine) was established in 1970 under the charter of the National Academy of Sciences to advise the nation on medical and health issues. Members are elected by their peers for distinguished contributions to medicine and health. Dr. Victor J. Dzau is president.

The three Academies work together as the National Academies of Sciences, Engineering, and Medicine to provide independent, objective analysis and advice to the nation and conduct other activities to solve complex problems and inform public policy decisions. The Academies also encourage education and research, recognize outstanding contributions to knowledge, and increase public understanding in matters of science, engineering, and medicine.

Learn more about the National Academies of Sciences, Engineering, and Medi- cine at www.national-academies.org.

COMMITTEE ON THE EFFECTS OF DILUTED BITUMEN ON THE ENVIRONMENT: A COMPARATIVE STUDY

Members DIANE MCKNIGHT (Chair), University of Colorado Boulder MICHEL BOUFADEL, New Jersey Institute of Technology MERV FINGAS, Independent Consultant STEPHEN K. HAMILTON, Michigan State University ORVILLE HARRIS, O.B. Harris, LLC JOHN M. HAYES, Woods Hole Oceanographic Institution (Ret.) JACQUELINE MICHEL, Research Planning, Inc. CARYS L. MITCHELMORE, University of Maryland Center for Environmental Science, Chesapeake Biological Laboratory DENISE REED, The Water Institute of the Gulf ROBERT SUSSMAN, Sussman and Associates DAVID VALENTINE, University of California, Santa Barbara

Staff DOUGLAS FRIEDMAN, Study Director CAMLY TRAN, Associate Program Officer CARL-GUSTAV ANDERSON, Research Associate COTILYA BROWN, Senior Program Assistant NAWINA MATSHONA, Senior Program Assistant

BOARD ON CHEMICAL SCIENCES AND TECHNOLOGY

Members DAVID BEM (Co-Chair), PPG Industries DAVID WALT (Co-Chair), Tufts University HÉCTOR D. ABRUÑA, Cornell University JOEL C. BARRISH, Bristol-Myers Squibb MARK A. BARTEAU, University of Michigan JOAN BRENNECKE, University of Notre Dame MICHELLE V. BUCHANAN, Oak Ridge National Laboratory DAVID W. CHRISTIANSON, University of Pennsylvania JENNIFER S. CURTIS, University of California, Davis RICHARD EISENBERG, University of Rochester SAMUEL H. GELLMAN, University of Wisconsin-Madison SHARON C. GLOTZER, University of Michigan MIRIAM E. JOHN, Sandia National Laboratories (Ret.) FRANCES S. LIGLER, University of North Carolina, Chapel Hill and North Carolina State University SANDER G. MILLS, Merck Research Laboratories (Ret.) JOSEPH B. POWELL, Shell, Houston PETER J. ROSSKY, Rice University TIMOTHY SWAGER, Massachusetts Institute of Technology

Staff TERESA FRYBERGER, Director DOUGLAS FRIEDMAN, Senior Program Officer KATHRYN HUGHES, Senior Program Officer CAMLY TRAN, Associate Program Officer CARL-GUSTAV ANDERSON, Research Associate ELIZABETH FINKELMAN, Program Coordinator COTILYA BROWN, Senior Program Assistant NAWINA MATSHONA, Senior Program Assistant

Preface

“I tell this story to illustrate the truth of the statement I heard long ago in the Army: Plans are worthless, but planning is everything. There is a very great distinction because when you are planning for an emergency you must start with this one thing: the very definition of ‘emergency’ is that it is unexpected, therefore it is not going to happen the way you are planning.” President Dwight D. Eisenhower November 14, 1957

The transport of crude oil through transmission pipelines in the U.S. has been essential to move crude oil from production fields to refineries for many decades, and has thus been an integral aspect of the U.S. energy infrastructure. Starting with the impact of the large crude oil spill off the coast of Santa Barbara, California, in 1969, the inherent environmental risks associated with the transport of crude oil became more widely recognized. This spill contributed to the passage of the National Environmental Policy Act of 1969 (NEPA), the creation of the President’s Council on Envi- ronmental Quality, and the U.S. Environmental Protection Agency. Several changes in the governmental approach to environmental policy would follow, eventually leading to the Oil Pollution Act of 1990 (OPA 90) in the wake of the Exxon Valdez spill in 1989. Through the resulting legislation of 1990 and Executive Order 12777, as amended, the U.S. Department of Transportation (USDOT) assumed responsibility to oversee the safe transport of crude oil in transmission pipelines, including thorough reviews of response plans and other actions. Now, 25 years after OPA 90 was passed, a shift in the distribution of vii

viii PREFACE

the types of crude oil carried in transmission pipelines has occurred and is anticipated to continue. Dense and viscous bitumen extracted using new technology from sources primarily in northern Alberta, Canada, is being diluted with less viscous hydrocarbons and transported to refineries throughout North America via transmission pipelines. This shift, along with a major spill of diluted bitumen in Marshall, Michigan, in 2010 and other spills elsewhere, has prompted Congress and USDOT to ask the National Academies of Sciences, Engineering, and Medicine to consider the use of transmission pipelines to transport diluted bitumen. The Acad- emies’ first study, released in 2013, focused on whether diluted bitumen was more likely to cause pipeline spills when compared to commonly transported crude oils. That study found no evidence of any causes of pipeline failure that are unique to the transportation of diluted bitumen. In this follow-on study, our committee was charged with addressing the question of whether the transport of diluted bitumen in pipelines has potential environmental consequences that are sufficiently different from those of commonly transported crude oils to warrant changes in regulations governing spill response planning, preparedness, and cleanup. The committee brought together diverse expertise on the chemistry and environmental impacts of crude oils and broad experience in spill response. Two members, including the study chair, have backgrounds in hydrology and environmental engineering. We had members with expertise in oil chemistry, geochemistry and biogeochemistry, and oil fate, behavior, and toxicity. Several of these scientists have been, and continue to be, actively involved in oil spill response activities. Beyond the scientific and engineering expertise, experts in pipeline operations and environmental regulations ensured that the committee considered the practical and policy aspects of our recommendations. In May 2015, while this study was still in its information-gathering phase, a rupture in Plains All American Line 901 spilled over 100,000 gallons of a heavy crude oil in Santa Barbara County, California and impacted almost 100 miles of shoreline. In addition to the two members of our committee who participated directly in the spill response as experts, we were able to observe the highly organized incident command in action four days after the spill. We discussed active response strategies with the National Oceanic and Atmospheric Administration (NOAA) Scientific Support Coordinator, the Liaison from the California Office of Spill Prevention and Response (CalOSPR), and several key members of the response team. A focus of these discussions was on the practical uses of formal response plans and on the daily decision-making process. As stated in President Eisenhower’s famous quote, it was clear that the preparation of response plans was an invaluable process that has improved the effectiveness of response.

PREFACE ix

The focus of this report and its recommendations is on the current concerns related to the transport of diluted bitumen in pipelines. We are confident that, by updating the planning process and taking greater advantage of available information about diluted bitumen when it is spilled, the effectiveness of spill response can be enhanced. However, given the nature of pipeline operations, response planning, and the oil industry, it is likely that our recommendations will be applicable to spill response, preparedness, and cleanup for many types of crude oil.

Diane McKnight, Chair Douglas Friedman, Study Director

Acknowledgments

The completion of this study would have not been successful without the assistance of many individuals and organizations. The committee would especially like to thank the following individuals and organizations for their contributions during this study:

U.S. Department of Transportation, Pipeline and Hazardous Mate- rials Safety Administration, which sponsored the study and provided valuable information on the agency’s responsibilities and structure. The committee would especially like to thank the Associate Administrator, Jeffrey Wiese, as well as Eddie Murphy, David Lehman, and Robert Smith. Mr. Smith served as the agency’s liaison to the committee and was effective in responding to the committee’s requests for information and site visits. U.S. Coast Guard for providing information on the agency’s regulatory responsibilities and technical information on the topic area. The committee would particularly like to thank Captain Claudia Gelzer, Captain Joseph Loring, Lt. Brandon Aten, and Lt. Sara Thompson. U.S. Environmental Protection Agency, which provided information regarding the agency’s regulatory responsibilities and experiences involved with oil spill response. The committee would like to thank Ralph Dollhopf, who served as an informal liaison to the committee; as well as Mark Howard, Greg Powell, Chris Ruhl, and Brian Schlieger. Speakers and invited participants at the committee’s data-gathering meetings. These individuals are listed here: Andy Black, Association

xii ACKNOWLEDGMENTS

of Oil Pipe Lines; Anthony Swift, Natural Resources Defense Council (NRDC); Bruce Hollebone, Environment Canada; Chris Reddy, Woods Hole Oceanographic Institution; Dan Capone, Mannik & Smith Group; David Westerholm, National Oceanic and Atmospheric Administra- tion (NOAA); Faith Fitzpatrick, U.S. Geological Survey (USGS); Gary Shigenaka, NOAA; Heather Dettman, Natural Resources Canada; John Zhou, Alberta Innovates; Ken Lee, Centre for Offshore Oil, Gas and Energy Research (COOGER); Liam Stone, Government of Canada; Lyman Young; Paul Connors, Government of Canada; Peter Hodson, Queens University; Peter Lidiak, American Petroleum Institute (API); Robin Rorick, API; Steve Larter, University of Calgary; Steve Lehmann, NOAA; Thomas King, COOGER; Tim Nedwed, ExxonMobil; and Tom Miesner, Pipeline Knowledge and Development. Jordan Stout, NOAA Scientific Support Coordinator, and Joy Lavin- Jones, Liaison Officer for California Office of Spill Prevention and Response (CalOSPR), for hosting a subgroup of the committee to observe the spill response operations for the Santa Barbara, California spill on May 19, 2015. And last, but certainly not least, the Academies staff for organizing and facilitating this study. Study Director Douglas Friedman and Associ- ate Program Officer Camly Tran organized the committee meetings and assisted the committee with research, report writing, and review. Senior Program Assistants Nawina Matshona and Cotilya Brown managed logistics of the meetings and publication. Senior Program Assistant Claire Ballweg and Communications Associate Sharon Martin contributed to the design of the figures and tables.

Acknowledgment of Reviewers

This report has been reviewed in draft form by individuals chosen for their diverse perspectives and technical expertise. The purpose of this independent review is to provide candid and critical comments that will assist the institution in making its published report as sound as possible and to ensure that the report meets institutional standards for objectivity, evidence, and responsiveness to the study charge. The review comments and draft manuscript remain confidential to protect the integrity of the deliberative process. We wish to thank the following individuals for their review of this report:

Mark Barteau, University of Michigan Jim Elliott, T&T Marine Salvage, Inc. Abbas Firoozabadi, Reservoir Engineering Research Institute Katherine H. Freeman, The Pennsylvania State University Elliott P. Laws, Crowell & Moring Ken Lee, Commonwealth Scientific and Industrial Research Organization Patricia Maurice, University of Notre Dame Stephen A. Owens, Squire Patton Boggs Chris Reddy, Woods Hole Oceanographic Institution Calvin H. Ward, Rice University

Although the reviewers listed above have provided many construc- tive comments and suggestions, they were not asked to endorse the con-

xiii

xiv ACKNOWLEDGMENT OF REVIEWERS

clusions or recommendations nor did they see the final draft of the report before its release. The review of this report was overseen by Thomas Leschine of the University of Washington and Michael Ladisch of Purdue University, who were responsible for making certain that an independent examination of this report was carried out in accordance with institutional procedures and that all review comments were carefully considered. Responsibility for the final content of this report rests entirely with the authoring committee and the institution.

Acronyms

ACP Area Contingency Plan ADIOS Automated Data Inquiry for Oil Spills AOPL Association of Oil Pipe Lines API American Petroleum Institute AWB Access Western Blend

BSEE Bureau of Safety and Environmental Enforcement BTEX benzene, toluene, ethylbenzene, xylenes

CalOSPR California Office of Spill Prevention and Response CERCLA Comprehensive Environmental Response, Compensation, and Liability Act CLWB Cold Lake Winter Blend COOGER Centre for Offshore Oil, Gas and Energy Research

Da Dalton DSD droplet size distribution

FRP Facility Response Plan

GNOME General NOAA Operational Modeling Environment

HCA high-consequence area

xvi ACRONYMS

ICCOPR Interagency Coordinating Committee on Oil Pollution Research

MCL maximum contaminant level

NAPL non-aqueous-phase liquid oil NAS National Academy of Sciences NCP National Contingency Plan NEB National Energy Board NEPA National Environmental Policy Act NGO nongovernmental organization NOAA National Oceanic and Atmospheric Administration NOSAMS National Ocean Sciences Accelerator Mass Spectrometry NRC National Research Council NRDA Natural Resource Damage Assessment NRDC Natural Resources Defense Council NRT National Response Team

OPA 90 Oil Pollution Act of 1990 OPA oil-particle aggregate OSC On-Scene Coordinator OSHA Occupational Safety and Health Administration OSPR Office of Spill Prevention and Response OSRO Oil Spill Removal Organization

PAH polycyclic aromatic hydrocarbon PHMSA Pipeline and Hazardous Materials Safety Administration

RRT Regional Response Team

SDWA Safe Drinking Water Act SDS Safety Data Sheet

TM Trans Mountain TPH Total Petroleum Hydrocarbons

USCG U.S. Coast Guard USDOT U.S. Department of Transportation USEPA U.S. Environmental Protection Agency USGS U.S. Geological Survey

VOC volatile organic compound

Contents

PREFACE vii

ACKNOWLEDGMENTS xi

ACKNOWLEDGMENT OF REVIEWERS xiii

ACRONYMS xv

SUMMARY 1 Study Approach, 2 Key Findings and Conclusions, 2 recommendations, 6 Oil Spill Response Planning, 7 Oil Spill Response, 8 USCG Classification System, 8 Advanced Predictive Modeling, 8 Improved Coordination, 8 Improved Understanding of Adhesion, 8 research Priorities, 9

1 INTRODUCTION 11 Background, 11 effects of Diluted Bitumen on Crude Oil Transmission Pipelines, 13 charge to the Committee, 14

xvii

xviii CONTENTS

Addressing the Statement of Task, 15 Data Gathering, 15 Defining Commonly Transported Crude Oil in the U.S. Pipeline System, 16 Organization of the Report, 19

2 CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 21 Introduction, 21 chemical Composition of Diluted Bitumen, 22 Saturated Hydrocarbons, 23 Aromatic Hydrocarbons, 24 Resins and Asphaltenes, 25 Chemical Composition of Diluents, 25 Weathering and Its Effects on Physical Properties, 28 Density, 28 Viscosity, 29 Flash Point, 30 Adhesion, 30 Conclusion, 33

3 ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY OF DILUTED BITUMEN 35 Introduction, 35 environmental Processes, 35 Chemical Processes, 36 Physical-Chemical Partitioning Processes, 38 Physical Processes, 41 Environmental Behavior, 50 Spills on Land and into Groundwater, 50 Inland Lakes and Reservoirs, 53 Wetlands, 54 Streams and Rivers, 54 Estuaries and Coastal Zones, 55 Beaches, 58 toxicity, 60 Toxicity of Commonly Transported Crude Oils, 60 Toxicity of Diluted Bitumen, 61 Human Health, 64 Conclusions, 69

CONTENTS xix

4 SPILL RESPONSE PLANNING AND IMPLEMENTATION 73 introduction, 73 implementation of Plans, 75 Predicting the Behavior of Spilled Oil, 75 Health and Safety Concerns, 76 Cleanup Endpoints, 77 tactics for Detection, Containment, and Recovery of Spills of diluted Bitumen, 82 Spills to Land, 82 Spills to Water and Wetlands, 82 Floating Oil Response Tactics, 83 On Water Containment and Recovery, 83 Dispersants, 84 In situ Burning, 84 Surface Washing Agents, 84 Response to Nonfloating Crude Oil and Its Residues, 85 Waste Management and Disposal, 87 conclusions, 87

5 COMPARING PROPERTIES AFFECTING TRANSPORT, FATE, EFFECTS, AND RESPONSE 89 Potential Outcomes and Level of Concern, 89 transport, 90 Fate, 92 effects, 94 response, 96 conclusions, 100

6 REGULATIONS GOVERNING SPILL RESPONSE PLANNING 101 Federal Spill Planning and Response Framework, 101 National Response Team, 103 Regional Response Teams, 103 Area Contingency Planning, 103 Facility Response Plans, 104 Onshore Pipeline Spill Response Plans, 104 Weaknesses of the Current Planning and Response Framework in Addressing Spills of Diluted Bitumen, 106 Adequacy Review versus Checklist Approach, 106 Strengthening the Area Contingency Planning Process, 108 Identifying the Type of Crude Oil Being Transported and Its Properties, 109 Certification of Oil Spill Removal Organizations, 110

xx CONTENTS

Updating Response Plans, 111 Strengthening Drills and Exercising of Plans, 111 conclusions, 112

7 Recommendations 113 Oil Spill Response Planning, 114 Oil Spill Response, 117 USCG Classification System, 118 Advanced Predictive Modeling, 118 Improved Coordination, 119 Improved Understanding of Adhesion, 120 priority Research Areas, 121 Final Thoughts, 122

REFERENCES 125

APPENDIXES

A GLOSSARY 135 B COMMITTEE MEMBER AND STAFF BIOGRAPHIES 139

Summary

In January 2012, Congress tasked the Secretary of Transportation to “determine whether any increase in the risk of release exists for pipelines transporting diluted bitumen.”1 In response to the congressional request, the U.S. Department of Transportation (USDOT) asked the National Acad- emies of Sciences, Engineering, and Medicine (the Academies) to study the likelihood of release of diluted bitumen from crude oil transmission pipelines. The Academies released a report in 2013 concluding that “[t]he committee does not find any causes of pipeline failure unique to the transportation of diluted bitumen.”2 Following the 2013 release of Effects of Diluted Bitumen on Crude Oil Transmission Pipelines, Congress subsequently charged USDOT to “investigate whether the spill properties [of diluted bitumen] differ sufficiently from other liquid petroleum products to warrant modifications to the spill response plans, spill preparedness, or cleanup regulations and report on those findings to the House and Senate Committees on Appropriations within 180 days of enactment.”3 USDOT returned to the Academies in 2014 with a request to form an ad hoc committee to help address this concern. Specifically, this committee was taskedi to review the available literature and data to examine the current state of knowledge, and to identify the relevant properties and characteristics of the transport, fate, and effects of diluted bitumen and commonly transported crude oils when spilled in the environment from U.S. transmission pipelines. Based on a comparison of the relevant

i The committee’s full statement of task can be found in Box 1-1. 1

2 SPILLS OF DILUTED BITUMEN FROM PIPELINES

properties of diluted bitumen and of a representative set of crude oils that are commonly transported via pipeline, the committee was asked to determine whether the differences between properties of diluted bitumen and those of other commonly transported crude oils warrant modifications to the regulations governing spill response plans, preparedness, and cleanup.

STUDY APPROACH In order to answer the questions outlined in the statement of task, the committee analyzed information in a variety of forms. Part of the committee’s data gathering included hearing presentations, meeting with stakeholders, and reviewing the literature. A detailed list of the individuals the committee met with can be found in the Acknowledgments section of this report. In the early phases of the study, an opportunity for public comment was provided. After considering all of the available data and information, the history of the study, and the sponsor’s request, the committee focused on environments that would most likely be affected by an oil spill from a transmission pipeline–that is, the contiguous U.S., including the near-shore coastline with far offshore not being considered. The report also focuses on spills from transmission pipelines and does not explicitly address other modes of transportation (e.g., rail, barge, truck, and tanker). It is likely that many of the topics covered in this report, and many of the conclusions and recommendations, will be applicable to these other transportation modes because many aspects of environmental impact are independent of mode of transportation. The committee’s task requires a comparison between diluted bitumen and “crude oils commonly transported in U.S. transmission pipelines.” After an analysis of the total volumes of crude oil transported by U.S. pipelines (see Chapter 1), a set of light and medium crudes was chosen as representative of those “commonly transported” and likely to be encountered in a response scenario. The committee’s approach is described in greater detail in Chapter 1.

KEY FINDINGS AND CONCLUSIONS The starting point for assessing the Chemical and Physical Proper- ties of Crude Oils (Chapter 2) was the intrinsic complexity of crude oils as mixtures of hydrocarbons with diverse structures and widely varying molecular weights. Mixtures of these compounds combine to make up the bulk properties of any particular crude oil. The bitumen fraction, in particular, is associated with reservoirs of recalcitrant and immobile crude oils. Unconventional extraction methods are required to access bitumen

SUMMARY 3

reservoirs and addition of a diluent is needed to transport the bitumen product through unheated transmission oil pipelines. In comparison to other commonly transported crude oils, many of the chemical and physical properties of diluted bitumen, especially those relevant to environmental impacts, are found to differ substantially from those of the other crude oils. The key differences are in the exceptionally high density, viscosity, and adhesion properties of the bitumen component of the diluted bitumen that dictate environmental behavior as the crude oil is subjected to weathering (a term that refers to physical and chemical changes of spilled oil). Immediately following a spill, the Environmental Processes, Behav- ior, and Toxicity of Diluted Bitumen (Chapter 3) are similar to those of other commonly transported crudes. Beginning immediately after a spill, however, exposure to the environment begins to change spilled diluted bitumen through various weathering processes. The net effect is a rever- sion toward properties of the initial bitumen. An important factor is the amount of time necessary for the oil to weather into an adhesive, dense, viscous material. For any crude oil spill, lighter, volatile compounds begin to evaporate promptly; in the case of diluted bitumen, a dense, viscous material with a strong tendency to adhere to surfaces begins to form as a residue. For this reason, spills of diluted bitumen pose particular challenges when they reach water bodies. In some cases, the residues can submerge or sink to the bottom of the water body. Importantly, the density of the residual oil does not necessarily need to reach or exceed the density of the surrounding water for this to occur. The crude oil may combine with particles present in the water column to submerge, and then remain in suspension or sink. These factors are important to consider for Spill Response Planning and Implementation (Chapter 4). Spills of diluted bitumen into a body of water initially float and spread while evaporation of volatile compounds may present health and explosion hazards, as occurs with nearly all crude oils. It is the subsequent weathering effects, unique to diluted bitumen, that merit special response strategies and tactics. For example, the time windows during which dispersants and in situ burning can be used effectively are significantly shorter for diluted bitumen than for other commonly transported crudes. In cases where traditional removal or containment techniques are not immediately successful, the possibility of submerged and sunken oil increases. This situation is highly problematic for spill response because (1) there are few effective techniques for detection, containment, and recovery of oil that is submerged in the water column, and (2) available techniques for responding to oil that has sunk to the bottom have variable effectiveness depending on the spill conditions. When Comparing Properties Affecting Transport, Fate, Effects, and Response (Chapter 5), several key properties emerge. Figure S-1 illustrates the properties relevant to transport, fate, and effects and the

4 SPILLS OF DILUTED BITUMEN FROM PIPELINES

potential environmental outcomes following a crude oil spill. Based on the similarities and differences between diluted bitumen (in pipeline and weathered forms) and other commonly transported crudes, the comparative levels of concern associated with these properties are highlighted. The majority of the properties and outcomes that differ from commonly transported crudes are associated not with freshly spilled diluted bitumen, but with the weathering products that form within days after a spill. Given these greater levels of concern for weathered diluted bitumen, spills of diluted bitumen should elicit unique, immediate actions in response. Based on the differences identified previously, a review of theRegula - tions Governing Spill Response Planning (Chapter 6) was conducted. Of particular focus was Part 194 of the Pipeline and Hazardous Materials Safety Administration (PHMSA) regulations, which governs the planning of responses to spills from transmission pipelines. In addition, because the scope of the task was broadly defined to address “regulations governing spill response plans, spill preparedness, or cleanup,” relevant U.S. Environmental Protection Agency (USEPA) and U.S. Coast Guard (USCG) regulations were reviewed, primarily for comparison to PHMSA regulations. It is clear that PHMSA takes a substantially different approach from USEPA and USCG when setting expectations for and reviewing spill response plans. Notably, PHMSA reviews plans for completeness in terms of the regulatory requirements only, while USEPA and USCG review plans for both completeness and adequacy for response. Broadly, regulations and agency practices do not take the unique properties of diluted bitumen into account, nor do they encourage effective planning for spills of diluted bitumen. In light of the aforementioned analysis, comparisons, and review of the regulations, it is clear that the differences in the chemical and physical properties relevant to environmental impact warrant modifications to the regulations governing diluted bitumen spill response plans, preparedness, and cleanup. The concern associated with these differences is summarized in Figure S-1 for both diluted bitumen and weathered diluted bitumen. Each property that is relevant to environmental transport, fate, and effects is identified with the potential outcomes and a qualitative level of concern compared to other commonly transported crudes. The most notable changes observed are in the comparison between diluted bitumen and weathered diluted bitumen. For example, the level of concern goes from the same to more (or less) concern between the weathered and non-weathered material for ten of the properties in Figure S-1 and all techniques identified in Figure S-2.

SUMMARY 5

FIGURE S-1 Spill hazards: diluted bitumen relative to commonly transported crude oils. Acronyms: BTEX: benzene, toluene, ethylbenzene, xylenes; HMW: high molecular weight; LMW: low molecular weight.

6 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE S-2 Response operations: diluted bitumen relative to commonly transported crude oils. Acronym: VOCs: volatile organic compounds.

RECOMMENDATIONS Diluted bitumen has unique properties, differing from those of commonly transported crude oils, which affect the behavior of diluted bitumen in the environment following a spill. This behavior differs from that of the light and medium crudes typically considered when planning responses to spills. Of greatest significance are the physical and chemical changes that diluted bitumen undergoes during weathering. A more comprehensive and focused approach to diluted bitumen across the oil industry and the relevant federal agencies is necessary to improve preparedness for spills of diluted bitumen and to spur more effective cleanup

SUMMARY 7

and mitigation measures when these spills occur. The recommendations presented here are designed to achieve this goal.

Oil Spill Response Planning Recommendation 1: To strengthen the preparedness for pipeline releases of oil from pipelines, the Part 194 regulations implemented by PHMSA should be modified so that spill response plans are effective in anticipating and ensuring an adequate response to spills of diluted bitumen. These modifications should

a. Require the plan to identify all of the transported crude oils using industry-standard names, such as Cold Lake Blend, and to include safety data sheets for each of the named crude oils. Both the plan and the associated safety data sheets should include spill-relevant properties and considerations; b. Require that plans adequately describe the areas most sensitive to the effects of a diluted bitumen spill, including the water bodies potentially at risk; c. Require that plans describe in sufficient detail response activities and resources to mitigate the impacts of spills of diluted bitumen, including capabilities for detection, containment, and recovery of submerged and sunken oil; d. Require that PHMSA consult with USEPA and/or USCG to obtain their input on whether response plans are adequate for spills of diluted bitumen; e. Require that PHMSA conduct reviews of both the completeness and the adequacy of spill response plans for pipelines carrying diluted bitumen; f. Require operators to provide to PHMSA, and to make publicly available on their websites, annual reports that indicate the volumes of diluted bitumen, light, medium, heavy, and any other crude oils carried by individual pipelines and the pipeline sections transporting them; and g. Require that plans specify procedures by which the pipeline operator will (i) identify the source and industry-standard name of any spilled diluted bitumen to a designated Federal On-Scene Coordinator, or equivalent state official, within 6 hours after detection of a spill and (ii) if requested, provide a 1-L sample drawn from the batch of oil spilled within 24 hours of the spill, together with specific compositional information on the diluent.

8 SPILLS OF DILUTED BITUMEN FROM PIPELINES

Oil Spill Response Recommendation 2: USEPA, USCG, and the oil and pipeline industry should support the development of effective techniques for detection, containment, and recovery of submerged and sunken oils in aquatic environments.

Recommendation 3: USEPA, USCG, and state and local governments should adopt the use of industry-standard names for crude oils, including diluted bitumen, in their oversight of oil spill response planning.

USCG Classification System Recommendation 4: USCG should revise its oil-grouping classifications to more accurately reflect the properties of diluted bitumen and to recognize it as a potentially nonfloating oil after evaporation of the diluent. PHMSA and USEPA should incorporate these revisions into their planning and regulations.

Advanced Predictive Modeling Recommendation 5: NOAA should lead an effort to acquire all data that are relevant to advanced predictive modeling for spills of diluted bitumen being transported by pipeline.

Improved Coordination Recommendation 6: USEPA, USCG, PHMSA, and state and local governments should increase coordination and share lessons learned to improve the area contingency planning process and to strengthen preparedness for spills of diluted bitumen. These agencies should jointly conduct announced and unannounced exercises for spills of diluted bitumen.

Improved Understanding of Adhesion Recommendation 7: USEPA should develop a standard for quantifying and reporting adhesion because it is a key property of fresh and weathered diluted bitumen. The procedure should be compatible with the quantity of the custodial sample collected by pipeline operators.

SUMMARY 9

RESEARCH PRIORITIES Although many differences between diluted bitumen and commonly transported crudes are well established, there remain areas of uncertainty that hamper effective spill response planning and response to spills. These uncertainties span a range of issues, including diluted bitumen’s behavior in the environment under different conditions, its detection when submerged or sunken, and the best response strategies for mitigating the impacts of submerged and sunken oil. These research priorities, discussed in Chapter 7, apply broadly to the research community. Major topics for future research include

• Transport and fate in the environment, • Ecological and human health risks of weathered diluted bitumen, • Detection and quantification of submerged and sunken oil, • Techniques to intercept and recover submerged oil on the move, and • Alternatives to dredging to recover sunken oil.

Introduction

BACKGROUND Over the past decade, production of “unconventional” oil in North America has surged as technological improvements and cost reductions have made these crudes competitive in the North American market. Unconventional oil in North America derives from two sources. In the U.S., hydraulic fracturing technologies have been widely applied to extract oil from shale formations or other typically inaccessible, low- permeability rocks. In Canada, petroleum products have been extracted from “oil sands” or “tar sands.” Together, these streams have increased North American production of crude oils by 46% since 2008.4 The oil sands yield bitumen, a highly viscous form of petroleum that is produced by surface mining or by in situ recovery. Surface mining is preferred for deposits within 75 m of the surface.5 In situ recovery, in which steam is injected to mobilize bitumen underground, is used for deeper deposits. Production of bitumen is predicted to increase 2.5-fold from 2013 to 20305a although future production trends may be influenced by declining crude prices in world markets. After separation from the host rock, bitumen is modified for transport. Commonly, it is combined with lower-density hydrocarbon mixtures (condensates, synthetic crude, or a mixture of both) to obtain a product with an acceptable viscosity and density for transport to refineries via pipeline. This engineered fluid is referred to as diluted bitumen. Common names refer to subtypes (e.g., dilbit, synbit, railbit, and dilsynbit) but, for

12 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE 1-1 Existing and proposed Canadian and U.S. crude oil pipelines. SOURCE: Canadian Association of Petroleum Producers5a

simplicity, the term diluted bitumen as used in this report encompasses all bitumen blends that have been mixed with lighter products. Diluted bitumen has been transported by pipeline in the U.S. for more than 40 years, with the amount increasing recently as a result of improved extraction technologies and resulting increases in production and exportation of Canadian diluted bitumen. The increased importation of Canadian diluted bitumen to the United States has strained the existing pipeline capacity and contributed to the expansion of pipeline mileage over the past 5 years. Although rising North American crude production has resulted in greater transport of crude oil by rail or tanker, oil pipelines continue to deliver the vast majority of crude oil supplies to U.S. refineries. Most of the pipeline systems that are currently transporting diluted bitumen originate near extraction sites in Alberta, Canada (see Figure 1-1). To accommodate increased export volumes, additional pipelines are being proposed and developed. Proposals include (i) the Keystone XL.,i which would deliver diluted bitumen to Cushing, Oklahoma, in the U.S.; (ii)

i President Obama announced on November 6, 2015 the decision to deny a Presidential Permit for the construction of the Keystone XL Pipeline.

INTRODUCTION 13

the Energy East, which would transport products to eastern Canada and its refineries; (iii) the Northern Gateway (“Enbridge Gateway” in Figure 1-1); and (iv) the Kinder Morgan Trans Mountain (“TM Expansion” in Figure 1-1). Both of the latter pipelines would transport products within Canada from Alberta to West Coast terminals. In the event of a spill, impacts and cleanup procedures depend strongly on the environmental setting. Figure 1-1 indicates that the transmission pipelines transporting diluted bitumen are currently located onshore, which includes passage across terrestrial and freshwater environments, and near shore, which includes the marine waters near the coastline. Deepwater environments are not presently pertinent to pipeline transport of diluted bitumen and are not considered in this report.

EFFECTS OF DILUTED BITUMEN ON CRUDE OIL TRANSMISSION PIPELINES In January 2012, the Secretary of Transportation was tasked by Con- gress to “determine whether the regulations are sufficient to regulate pipeline facilities used for the transportation of diluted bitumen . . . and whether any increase in the risk of release exists for pipeline facilities transporting diluted bitumen.”2 The U.S. Department of Transporta- tion’s (USDOT) Pipeline and Hazardous Materials Safety Administration (PHMSA) contracted with the National Academies of Sciences, Engineer- ing, and Medicine (the Academies) to assemble a committee of experts to analyze whether the likelihood of release was greater for the transportation of diluted bitumen compared to that for other commonly transported crudes via U.S. transmission pipelines.2 An expert committee completed a comprehensive analysis and review of the available data on the chemical and physical properties of shipments of diluted bitumen and other crudes, examined pipeline incident statistics and investigations, and consulted experts in pipeline corrosion, cracking, and other causes of releases. The analysis covered many aspects of pipeline transportation including an explanation of the U.S. pipeline system; pipeline construction, maintenance, and alerts; incident data reported to PHMSA; and a discussion of bitumen production. Ultimately, after detailed analysis, the committee report, issued in 2013, “did not find any causes of pipeline failure unique to the transportation of diluted bitumen.”2 Environmental consequences of spills of diluted bitumen from pipelines were not within the scope of the Effects of Diluted Bitumen on Crude Oil Transmission Pipelines report. Following the release of the 2013 study, Congress tasked USDOT in 2014 to undertake a study to better understand the environmental impacts of spills of diluted bitumen from transmission pipelines and the adequacy of spill response planning.3

14 SPILLS OF DILUTED BITUMEN FROM PIPELINES

CHARGE TO THE COMMITTEE Based on this direction from Congress, PHMSA returned to the Acad- emies with a request to assemble an ad hoc committee to analyze whether the relevant properties of diluted bitumen differ sufficiently from those of other crude oils commonly transported in U.S. transmission pipelines to warrant modifications of the regulations governing spill response plans, spill preparedness, and/or cleanup. The committee’s statement of task is provided in Box 1-1. This report focuses primarily on spills of crude oil from U.S. transmission pipelines. Over the course of producing this report, however, it became clear that the utilization of other modes of transportation for crude oil such as rail, truck, and tanker have increased and are worth consideration. While this report does not address any of the particular aspects of those transportation modes, many of the environmental effects of spilled oil are independent of the method of transporta-

BOX 1-1 Statement of Task

An ad hoc committee will analyze whether the properties of diluted bitumen differ sufficiently from those of other crude oils commonly transported in U.S. transmission pipelines to warrant modifications of the regulations governing spill response plans, spill preparedness, or cleanup.

The committee will 1. Review the available literature and data, including any available data from oil spill responses or cleanup, to determine the current state of knowledge of the transport, fate, and effects of diluted bitumen once spilled into the environment (onshore and offshore); 2. Identify the relevant properties and characteristics that influence the transport, fate, and effects of commonly transported crude oils, including diluted bitumen, in the environment; 3. Make a comparison of the relevant properties identified in item 2 between diluted bitumen and a representative set of crude oils that are commonly transported via pipeline; and 4. Based on the comparison in item 3, analyze and make a determination as to whether the differences between the environmental properties of diluted bitumen and those of other crude oils warrant modifications to the regulations governing spill response plans, spill preparedness, or cleanup.

If the committee finds that there is not sufficient information to make a comparison of the environmental properties between diluted bitumen and other crude oils, the committee may make recommendations as to the additional data that would be needed to make such a determination.

INTRODUCTION 15

tion and therefore this report can provide useful insight into areas beyond pipeline transportation.

ADDRESSING THE STATEMENT OF TASK To understand the potential consequences of spills of diluted bitumen, knowledge regarding its chemical properties and environmental behavior during and after a spill in various spill environments is required. To date, several reports have been published that examine the properties,6 toxicity,7 and composition of diluted bitumen products derived from the Canadian oil sands.8 Other recent reports focus on the behavior and fate of spills in marine5b,8-9 and freshwater environments.5b,9a,10 Many of these reports were prepared as a result of the release of diluted bitumen into the Kalamazoo River by a break in the Enbridge 6B pipeline in Marshall, Michigan, on July 25, 2010. The total release was estimated to be 843,444 gallons, one of the largest freshwater oil spills in North American history, with cleanup costs exceeding $1.2 billion.11 The Marshall release attracted attention because of the broad extent and consequences of the release and the unprecedented scale of impact.12 The data and information gathered from experts and reports have been critical to addressing the statement of task and supporting the recommendations found herein. Nonetheless, the current knowledge base is limited, and a better understanding of the chemical constituents and behavior of diluted bitumen spills in diverse environmental settings would be helpful to inform response plans and actions.

Data Gathering To make a comparative analysis of diluted bitumen and crudes commonly transported by U.S. transmission pipelines, the committee gathered information from a variety of experts and stakeholders from government, nongovernmental organizations (NGOs), industry, and academia. A list of those experts and stakeholders can be found in the Acknowledgments. Technical information on properties of crude oil and on the behavior, fate, and environmental impacts of spills of diluted bitumen was provided directly or presented during one of several data-gathering meetings. In addition, discussions with agencies, individuals, and groups concerned with development of plans for responses to oil spills were extremely valuable. A subgroup from the committee conducted a site visit to the incident command post for the Refugio spill in Santa Barbara, California, in May 2015 to observe a response in action and to hold discussions with participants. A questionnaire requesting data was submitted to the American Petroleum Institute (API) and to the Association of Oil Pipe Lines (AOPL)

16 SPILLS OF DILUTED BITUMEN FROM PIPELINES

with a request for help obtaining data from the pipeline industry but drew no response.

Defining Commonly Transported Crude Oil in the U.S. Pipeline System In 2013, the United States produced 2,720 million barrels13 of crude oil, an increase of about 35% since 2000.14 The United States also imported a total of 2,820 million barrels13 of crude oil in 2013 from all countries. The total volume of produced and imported crude oil to the United States for 2013, the most recent year for which complete data are available, was thus 5,540 million barrels. Of these, 3,190 million barrels14 (58%) were delivered to refineries by pipeline. This amount includes both domestic crude (2,200 million barrels) and imported crude (992 million barrels).14 A summary of products transported in the U.S. pipeline system is presented in Table 1-1. Taken in broad strokes, the majority of the crude oil transported in the U.S. pipeline system in 2013 was conventional light and medium crude (~71%). With the recently increased production of light crude oil in the U.S., it is expected that light and medium crudes will remain dominant in the U.S. pipeline system. In the United States, production of heavy crude oil has been roughly constant even as production of light crude oil production has, in recent years, grown rapidly.15 The principal domestic sources of heavy crude oil are in California, which produced 199 million barrels in 2013,16 most of that being transported by pipeline within the state.17 Because heated

TABLE 1-1 Types and Quantities of Crude Oil in the U.S. Pipeline System in 2013 Volume in 2013 % by volume in Type of Crude Oil (million barrels) pipeline system Diluted Bitumen 250 8% Undiluted Conventional Heavya 199 6% Diluted Conventional Heavyb 273 9% Conventional Medium and Light 2,278c 71% Synthetic Crude 190 6% Total 3,190 aDomestic production of conventional heavy crude oil, API < 27°. bImported conventional heavy crude oil, API < 25°. cConventional medium and light crude oil = Reported Total – (Diluted Bitumen + Diluted and Undiluted Conventional Heavy + Synthetic Crude). SOURCES: National Energy Board and U.S. Energy Information Administration15, 19

INTRODUCTION 17

pipelines are used, the heavy crude oils in California do not require blend- ing with lighter products and are hence termed “undiluted conventional heavy” crude oil. Most of the other heavy crude oil that is transported by pipeline in the United States is from Canada. These heavy crude oils are diluted with lighter hydrocarbons and, hence, are referred to as “diluted conventional heavy” crude oil. Some imported heavy crude oil also comes from Mexico and Venezuela but those products arrive by tanker18 and are not typically transported by U.S. transmission pipelines. The remaining categories of crude oil transported via pipeline are diluted bitumen and synthetic crude. In 2013, the National Energy Board (NEB) of Canada reported an export of 250 million barrels of diluted bitumen to the United States.20 The NEB has defined diluted bitumen as bitumen blended with light hydrocarbons and/or synthetic crude oil. Although there has been an increase in rail transportation of diluted bitumen, petroleum products from Canada, including diluted bitumen, are transported mainly by pipeline. By this analysis, diluted bitumen made up 8% of the crude oil carried in the U.S. pipeline system in 2013.20 The volume of diluted bitumen imported from Canada increased by ~20% in 2014. The Canadian diluted bitumen transported in transmission pipelines to the U.S. typically contains 50-70% bitumen by volume with lighter hydrocarbons accounting for the remainder.2,5b The quantity of diluents added is typically the minimum needed to meet pipeline specifications. The most common specifications for pipeline inputs are a maximum density of 0.94 grams per cubic centimeter (g/cm3) and a maximum viscosity of 350 centistokes (cSt).2,6 Bitumen blended with synthetic crude usually has a mixture of about 50% bitumen and 50% synthetic crude,5b whereas bitumen blended with naphtha-based oils derived from conventional crudes or from condensates derived from natural gas typically contains a mixture of about 70% bitumen and 30% light oils.2 Bitumen blends also vary seasonally in order to meet specifications for density and viscosity at the temperature of the pipeline. Synthetic crude oil can be upgraded bitumen, upgraded heavy crude oil, or a mixture of those products, and makes up 6% of the total oil being transported by pipeline. “Upgrading” refers to inefficient, but cost-effective, refining procedures implemented at or near the site of production rather than after transport to a refinery. The total volumes of Canadian crudes imported to the U.S. by pipeline are presented in Table 1-2. The statement of task seeks a comparison between diluted bitumen and “crude oils commonly transported in U.S. transmission pipelines.” A definition of commonly transported crudes is thus required. Figure 1-2 graphically depicts the volume percentages of types of crude oil transported by the U.S. pipeline system in 2013 and includes both imported and domestic oil. It shows that light and medium crude oils are the

18 SPILLS OF DILUTED BITUMEN FROM PIPELINES

TABLE 1-2 Volumes of Canadian Crudes Imported to the U.S. by Pipeline in 2013 Type of Crude Oil Volume (million barrels) Diluted Bitumena 250 Conventional Diluted Heavyb 273 Conventional Medium and Light 206 Synthetic Crudec 190 aBitumen blended with light hydrocarbons and/or synthetic crude oil. bAPI gravity < 25°. cUpgraded bitumen or upgraded heavy crude oil of any API gravity. SOURCE: National Energy Board19

predominant crude oil products being transported in the U.S. transmission pipeline system and account for nearly three-quarters of the crude oil transported. Further, a significant fraction of the transport of heavy crude oils occurs in a single state (California), whereas other crude oils are transported throughout the contiguous United States. This is the basis for identifying light and medium crudes as commonly transported and indicating that a comparison between diluted bitumen and these crude classes provides a meaningful basis for addressing the statement of task. Accord- ingly, for the purposes of this report, commonly transported crudes are defined as conventional light and medium crude oils (Figure 1-2). Key terms used to describe the types of crude oils used throughout this report are highlighted in Figure 1-3.

FIGURE 1-2 Percentages of crude oil types by volume in the U.S. pipeline system in 2013. See Table 1-1 for details.

INTRODUCTION 19

FIGURE 1-3 Key terms used in this report.

ORGANIZATION OF THE REPORT Chapter 2 discusses the chemical and physical properties of crude oils that are relevant to environmental impact. Properties discussed in detail include density, viscosity, flash point, and adhesion. The effects of weathering on these properties are also highlighted and presented in a series of tables which are organized to compare light, medium, and heavy crude oils with diluted bitumen. The chapter concludes by identifying key differences between the properties of those products. Chapter 3 examines the environmental transport, fate, and effects of spills of crude oils with a focus on properties unique to diluted bitumen before and after the diluent has been lost to volatilization. It also reviews relevant crude oil spills and considers potential spills in a variety of environmental settings including land, groundwater, inland waters, and coastal zones. Chapter 4 describes the current planning and implementation of response to spills of crude oil. Each spill is unique and its characteristics depend on the chemical and physical properties of the oil and on the environment in which the spill has occurred. Predictions of the behavior of spilled oil and of its effects on health and safety are described in this

20 SPILLS OF DILUTED BITUMEN FROM PIPELINES

chapter and are pertinent to how a spill response will be implemented and what type of tools and equipment will be employed. The chapter reviews general response tactics and techniques for floating oils as well as tactics for detection, containment, and recovery of spills that have a higher tendency to submerge. Considering the distinctions between light and medium crudes compared to that of diluted bitumen described in Chapters 2 and 3, a descriptive table about recovery techniques for diluted bitumen spills concludes the chapter. Chapter 5 synthesizes the information presented in the previous chapters. The differences between commonly transported crude oils and diluted bitumen are presented in three separate tables organized in terms of environmental transport, fate, effects, and spill response describing the relevant properties, potential outcomes, and levels of concern. Chapter 6 focuses on the adequacy of current regulations governing spill response plans, preparedness, and cleanup. The chapter provides an overview of the federal spill planning and response framework for crude oil spills. Weaknesses of the current pipeline spill response planning and response framework for addressing spills of diluted bitumen are discussed. Chapter 7 presents specific recommendations to stakeholders involved in spill response based on the committee’s analysis and assessment of the statement of task. While the focus of these recommendations is on how to increase the effectiveness of spill response planning and response for spills of diluted bitumen, the committee’s recommendations are relevant to other oils that share physical and chemical properties with diluted bitumen (i.e., heavy oils), although non-bitumen heavy oils are beyond the scope of this report.

Chemical and Physical Properties of Crude Oils

INTRODUCTION Crude oil derives, by way of geological processing, from organic material initially buried in sediments at the bottom of ancient lakes and oceans. Crude oil formed at depth in a sedimentary basin migrates upward because of lower density. Many such migrations end with the oil collecting beneath a layer of impermeable rock, also referred to as a “trap,” and forming a reservoir that can be tapped by drilling. If the oil approaches the surface, it cools and comes in contact with groundwater. At the oil-water interface, anaerobic microorganisms degrade the oil in the absence of oxygen. The progressive loss of metabolizable molecules from the oil leads to an increase in viscosity and eventually results in a tarry residue that clogs the pores of the strata through which the oil had been migrating. Over a long duration and with adequate sources of oil from below, enormous deposits of biodegraded oil residue can accumulate. This sequence is how the Alberta oil sands21 and other oil-sand deposits were formed. Bitumen is separated from the host rock or sand by heating, which reduces its viscosity so that it can flow to a collection point. Once collected, it is mixed with a diluent so that its viscosity is low enough to allow transport in a transmission pipeline. Such mixtures are called diluted bitumen. Diluted bitumen are engineered to resemble other crude oils that are transported via pipeline and processed in the same refineries. The composition of diluted bitumen is dependent on several factors, particularly the

22 SPILLS OF DILUTED BITUMEN FROM PIPELINES

diluent or diluents chosen and the diluent-to-bitumen ratio. As a result, diluted bitumen has dimensions of variability significantly exceeding those of crude oil from a given source region.21

CHEMICAL COMPOSITION OF DILUTED BITUMEN Diluted bitumen and other crude oils generally contain the same classes of compounds, but the relative abundances of those classes vary widely. Those variations are associated in turn with wide differences in physical and chemical properties. Industry-standard analyses group compounds into four main classes, namely saturated hydrocarbons, aromatic hydrocarbons, resins, and asphaltenes. Saturated hydrocarbons are most abundant in light crude oils, which are the least dense and least viscous. Denser and more viscous crude oils have greater concentrations of other components, including resins and asphaltenes, which contain more polar compounds, often including “heteroatoms” of nitrogen, sulfur, and oxygen as well as carbon and hydrogen. Even among light or medium crude oils, the relative abundances of specific compounds can vary significantly. The relative abundances will depend on the precise composition of the organic material delivered to the source sediments, the rate and length of time over which the source rock was heated, which inorganic minerals—potential catalysts of specific chemical reactions—were present in the source rock, the distance and details of the migration pathway, and conditions in the reservoir. In Table 2-1 and Figure 2-1, North American crude oils of each type for which data are readily available are provided as representative examples.22 From light, to medium and heavy crudes, and on to diluted bitumen, the abundance of saturated hydrocarbons drops 4-fold and the combined abundances of resins and asphaltenes increase 50-fold. These differences

TABLE 2-1 Major Classes of Compounds in Crude Oils, Percentages by Weight Type of Crude Oil Saturates Aromatics Resins Asphaltenes Light Crudea 92 8 1 0 Medium Crudeb 78 15 6 1 Heavy Crudec 38 29 20 13 Diluted Bitumend 25 22 33 20 aScotia Light. bWest Texas Intermediate. cSockeye Sour. dCold Lake Blend. SOURCE: Hollebone22

CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 23

FIGURE 2-1 Components of typical crude oils.

are attributable mainly to the great influence of biodegradation on the heavier crude oil and bitumen.

Saturated Hydrocarbons Under the anaerobic conditions prevailing during formation of the oil sands, the saturated hydrocarbons are mostly biodegradable, the aromatic hydrocarbons much less so, and the resins and asphaltenes not at all. A heavy crude, or the bitumen from an oil sand, is composed of the residue from a very protracted process whereby microbial action consumes most of the metabolizable saturates. The saturated hydrocarbon fraction in diluted bitumen thus differs from that in other crude oils because the readily metabolizable molecules are missing. This is seen most dramatically in chemical analyses that reveal the distribution of individual compounds in the crude oil. For example, the graphs in Figure 2-2 show results of parallel analyses of samples of Cold Lake Blend diluted bitumen and Bakken crude oil.23 The latter is dominated by a strong series of peaks representing its abundant, straight-chain, saturated hydrocarbons. The diluted bitumen, in contrast, is dominated by a hump representing the profusion of branched and cyclic hydrocarbons that are more resistant to biodegradation. These are so numerous and varied that their peaks overlap and they cannot be resolved by this gas chromatographic analysis. The diluted bitumen has a small series of peaks indicating the presence of some straight-chain hydrocarbons that derive from the diluent.

24 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE 2-2 Gas chromatography (GC) comparison of Cold Lake Blend diluted bitumen and Bakken crude oil, a light crude, from North Dakota (FID, flame ionization detector). SOURCE: Swarthout, et al.23

Aromatic Hydrocarbons Crude oils contain aromatic hydrocarbons possessing one or more aromatic rings. Those with more than one ring are commonly referred to as polycyclic aromatic hydrocarbons (PAHs). The one-ring compounds are most abundant and are referred to collectively as BTEX, an acronym based on the chemical names of benzene, toluene, ethyl benzene, and xylenes. The most common aromatic hydrocarbons with two rings are naphthalenes. Other commonly measured groups include the three-ring phenanthrenes, dibenzothiophenes, and fluorenes and also the four-ring chrysenes. The napthalenes and the even larger phenanthrenes are progressively less volatile and soluble compared to BTEX. PAHs are present as unsubstituted or parent forms but the vast majority are alkyl substituted PAHs. The aromatic hydrocarbons are of interest because of their toxicity. Specific properties and risks are discussed in Chapter 3. In Table 2-2, abundances of commonly measured PAHs in crude oils and in diluted

CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 25

bitumen are listed. Values listed in red mark cases in which the concentration in diluted bitumen exceeds that in the other crude oils indicated.

Resins and Asphaltenes The resins and asphaltenes characteristic of heavy crudes and diluted bitumen can precipitate from the oil as black sludge and cause numerous problems: clogging well bores, pipelines, and apparatus.24 Moreover, refining costs increase with the abundances of resins and asphaltenes.24 For all of those reasons, light and medium crudes have been favored. With increasing pressure on supplies, and with continued improvements in refining processes, heavy crude oils have come into broader use. As shown by Table 2-1, the content of resins and asphaltenes in light and medium crude oils is very much lower than that in heavy crude oils, and lower still than that in diluted bitumen. The resins and asphaltenes have presented major challenges to chemical analysts.24 The range of structures and the tendency of the molecules to cluster in larger, multimolecular aggregates make it difficult to determine even rudimentary properties like molecular weight. It has been shown only recently25 that most individual molecules in the heavy residues have from 30 to 70 carbon atoms. They comprise a complex mixture of polycyclic molecular structures in that range. These molecules tend to stick together, not only in bulk (the property that makes asphalt an attractive paving material) but even at the low concentrations prevailing when samples are injected into analytical instruments. The resulting “nanoaggregates” have masses two to five times higher than those of the molecules of which they are composed. The apparent molecular weights are accordingly higher than the true molecular weights. Heteroatoms (mainly nitrogen, sulfur, and oxygen) and metals (mainly nickel, vanadium, and iron) are also present in higher relative abundances in the resin and asphaltene fractions than in the saturate fraction. As a result, the heteroatom content of bitumen is higher than that of other crude oils.

Chemical Composition of Diluents The density and viscosity of raw bitumen are too great to allow transportation by transmission oil pipeline without heating or alteration of the material. To reduce the viscosity and density, a diluent must be added to bitumen to produce an engineered mixture with a density of less than 0.94 g/cm3 and a viscosity of less than 350 cSt. Additional industry-standard specifications that are largely a function of the operating temperature of the pipeline vary seasonally. Diluents alone do not confer unique chemical or toxicological properties to diluted bitumen;

26 SPILLS OF DILUTED BITUMEN FROM PIPELINES

TABLE 2-2 Concentrations of Parent and Alkylated PAHs, EPA Priority PAHs, and Total Aromatic Compounds in Various Crude Oils Diluted Oil Type Light Medium Heavy Bitumen Oil Sample Scotia Light West Texas Sockeye27 Cold Lake oil26 oil27 Blend28

Specific Conc. (µg/g) Conc. (µg/g) Conc. (µg/g) Conc. (µg/g) Compounds

Total EPA 16 139 514 218 176 Priority PAHs* (µg/g)

Total Aromatic 3,504 7947 5,231 5,384 Compounds (µg/g)

Parent and Alkylated PAHs Sum Naphthalene* 2692 5172 3422 2099 C0-41 Sum 351 1295 1078 1242 Phenanthrene* C0-4 Sum 16.3 816 403 1250 Dibenzothiophene C0-3 Sum Fluorene* 358 458 184 535 C0-3 Sum Chrysene* 17.9 100 60 200 C0-3 Total parent and 3,434 7841 5,147 5,326 alkylated PAHs Biphenyl 25.9 68.5 34.23 6.58 Acenaphthylene* 3.91 11.08 6.72 2.16 Acenaphthene* 24.2 8.84 7.7 6.93 Anthracene* 1.57 1.00 2.2 N.D. Fluoranthene* 2.93 2.12 1.22 4.31 Pyrene* 2.55 6.72 5.01 11.3

CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 27

TABLE 2-2 Continued Diluted Oil Type Light Medium Heavy Bitumen Benz[a] 1.41 1.24 3.18 2.52 anthracene* Benzo[b] 1.29 1.37 0.98 4.06 fluoranthene* Benzo[k] 0.30 0.37 0.40 0.81 fluoranthene* Benzo[e]pyrene 1.33 3.48 1.59 4.10 Benzo[a]pyrene* 0.74 0.25 0.49 3.01 Perylene 1.10 0.12 19.32 7.16 Indeno[1,2,3-cd] 0.38 0.18 N.D. 1.89 pyrene* Dibenzo[ah] 0.32 0.18 0.12 0.73 anthracene* Benzo[ghi] 1.29 0.50 0.86 2.24 perylene* 1 N.D. = not detected. Red values indicate levels in Cold Lake Bitumen that are higher in comparison to Scotia Light and West Texas crude oils. C0 are parent unsubstituted PAHs and C1-C4 are the alkyl PAHs. *Denotes the 16 EPA priority PAHs (naphthalene, phenanthrene, fluorene, and chrysene use the C0 parent levels only).

all crude oils contain similar, light end components. The compositions of diluents, however, can strongly affect the weathering behavior of diluted bitumen, chiefly because the evaporation of a highly volatile diluent will more readily produce a heavy residue. The individual selection of diluents varies depending on the desired outcome, the current cost of acquiring and transporting the diluent to the bitumen source, and other internal considerations of pipeline operators. Specific information about the diluents used is typically not publicly available. In general, diluents used fall into two broad categories: naturally occurring mixtures of light hydrocarbons, synthetic crude oil, or both. Synthetic crude oil is produced by upgrading bitumen to reduce its density and viscosity for transport by pipeline. When mixed with bitumen to obtain the required viscosity and density, synthetic crude oils yield a product that can be handled efficiently and economically by conventional heavy oil refineries. A drawback is that supplies of synthetic crudes are limited by the availability of upgraders at the source of extraction and that roughly a 50:50 mixture of bitumen with synthetic crude oil is required to obtain the desired density and viscosity.

28 SPILLS OF DILUTED BITUMEN FROM PIPELINES

The alternative, and more commonly used, diluents are naturally occurring mixtures of light hydrocarbons. These light hydrocarbons are acquired from two sources: ultralight crude oils and gas condensates. Gas

condensates are produced by separating most of the C3 and all of the C4 and higher hydrocarbons from natural gas. Because the ultralight crudes and gas condensates are less dense and less viscous than synthetic crude oil, diluent-to-bitumen ratios are roughly 30:70. The particular mixture of light hydrocarbons in the diluent can be important in spill response. If

the diluent is dominated by lighter compounds (C4-C8), it can evaporate more readily in the event of a spill, yielding a dense and viscous residue that must be accounted for in response.

WEATHERING AND ITS EFFECTS ON PHYSICAL PROPERTIES The behavior of a crude oil or diluted bitumen released into the environment is shaped not only by its chemical composition but also by its physical properties. Those of particular interest are density, viscosity, flash point, and adhesion. Oil spilled into the environment undergoes a series of physical and chemical changes that in combination are termed weathering. Weathering processes occur at different rates, but they begin as soon as oil is spilled and usually proceed most rapidly immediately after the spill. Most weathering processes are highly temperature dependent and will slow to insignificant rates as temperatures approach freezing. The most important weathering process is evaporation,29 which accounts for the greatest losses of material. Over a period of several days, a light fuel such as gasoline evaporates completely at temperatures above freezing, whereas only a small percentage of bitumen evaporates. Importantly, properties of the residual oil change as the light components of the oil are removed.

Density Given that the density of fresh water is 1.00 g/cm3 at environmental temperatures and the densities of crude oils commonly range from 0.7 to 0.99 g/cm3 (see Table 2-3), most oils will float on freshwater. Because the density of seawater is 1.03 g/cm3, even the heaviest oils will usually float on seawater. But evaporative losses of light components can lead to significant increases in density of the residual oil. The densities of some weathered, diluted bitumen and of undiluted bitumen can approach and possibly exceed that of freshwater. Accordingly, those materials can submerge and may sink to the bottom. In this respect, diluted bitumen differs not only from light and medium crude oils, but even from most conventional heavy crude oils. Details are shown in Table 2-3.

CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 29

TABLE 2-3 Density Comparison of Typical Crude Oilsa Density After Density After Initial Additional Weathering Weathering Density (mass % loss in (mass % lost in Type of Crude Oil Before Release weathering) weathering) Light Crudeb 0.77 0.80 (25%) 0.84 (64%) Medium Crudec 0.85 0.87 (10%) 0.90 (32%) Heavy Cruded 0.94 0.97 (10%) 0.98 (19%) Diluted Bitumene 0.92 0.98 (15%) 1.002 (30%) Bitumen 0.998 1.002 (1%) 1.004 (2%) aData in g/cm3 at 15°C; freshwater has a density of 1.00, seawater of 1.03. bScotia Light. cWest Texas Intermediate. dSockeye Sour. eCold Lake Blend. SOURCE: Hollebone22

Importantly, as discussed in Chapter 3, as the density of a weathering oil approaches that of water, contact with even small amounts of sand, clay, or other suspended sediment can trigger submergence. For this reason, the density of the oil residue itself (i.e., not including any associated natural particulate matter) does not need to exceed that of water for the residue to sink from the surface.

Viscosity Viscosity is defined as the resistance to flow of a liquid: the lower the viscosity, the more readily a liquid flows. For example, water has a low viscosity and flows readily, whereas honey, with a high viscosity, flows poorly. The viscosity of oil is largely determined by its content of large, polar molecules, namely resins and asphaltenes. The greater the percentage of light components such as saturates and the lower the amount of asphaltenes, the lower the viscosity. Temperature also affects viscosity, with a lower temperature resulting in a higher viscosity. The variations with temperature are commonly large. Oil that flows readily at 40ºC can become a slow-moving, viscous mass at 10ºC. Evaporative losses selectively remove lighter components and, consequently, increase the viscosity of the residual oil, as illustrated in Table 2-4.

30 SPILLS OF DILUTED BITUMEN FROM PIPELINES

TABLE 2-4 Viscosity Comparison of Typical Crude Oilsa Viscosity After Viscosity After Additional Initial Weathering Weathering Viscosity Before (mass % loss in (mass % lost in Type of Crude Oil Release weathering) weathering) Light Crudeb 1 2 (25%) 5 (64%) Medium Crudec 9 16 (10%) 112 (32%) Heavy Cruded 820 8,700 (10%) 475,000 (19%) Diluted Bitumene 270 6,300 (15%) 50,000 (30%) Bitumen 260,000 300,000 (1%) 400,000 (2%) aData in mPa·s. bScotia Light. cWest Texas Intermediate. dSockeye Sour. eCold Lake Blend. SOURCE: Hollebone22

Flash Point The flash point of oil is the temperature at which the liquid produces vapors sufficient for ignition by an open flame. A liquid is considered to be flammable if its flash point is less than 60oC. There is a broad range of flash points for oils and petroleum products, many of which are considered flammable, especially when freshly spilled. Gasoline, which is flammable under all ambient conditions, poses a serious hazard when spilled. Many fresh crude oils and diluted bitumen have an abundance of volatile components and may be flammable for a day or longer after being spilled, depending on the rate at which highly volatile components are lost by evaporation. On the other hand, undiluted bitumen and heavy crude oils typically are not flammable. Table 2-5 provides a quantitative summary of these variations.

Adhesion The adhesion or “stickiness” of some crude oils has been noted as a problem at several spills. The adhesion of a crude oil to the surfaces of rocks, built surfaces, and vegetation can greatly impede cleanup. Although important in the context of oil spill response, adhesion is a property that is not measured during industry-standard analyses of crude oils. However, a quantitative measure of adhesion has been developed30 and a comparison of some values appears in Table 2-6. The test measures the mass of oil, or of weathered oil, that will adhere to a steel needle that has been immersed in the sample for 30 min and then allowed to drain for

CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 31

TABLE 2-5 Flash Point Comparison of Typical Crude Oilsa Flash Point Flash Point After After Additional Initial Weathering Weathering Flash Point (mass % loss in (mass % lost in Type of Crude Oil Before Release weathering) weathering) Light Crudeb <−30 23 (25%) 95 (64%) Medium Crudec −10 33 (10%) >110 (32%) Heavy Cruded −3 67 (10%) >95 (19%) Diluted Bitumene <−35 >60 (15%) >70 (30%) Bitumen >100 >100 (1%) >110 (2%) aData in °C. bScotia Light. cWest Texas Intermediate. dSockeye Sour. eCold Lake Blend. SOURCE: Hollebone22

30 min. As can be seen, diluted bitumen is much more strongly adhesive than light or medium crude oils, or their evaporated residues. The contrast is even greater than it may appear. Not only is the diluted bitumen residue more adhesive, there is much more of it relative to the discharge for a given spill, due to the greater abundances of resins and asphaltenes in diluted bitumen. A comparison of the adhesion for various crude oils is listed in Table 2-6.

TABLE 2-6 Adhesion Comparison of Typical Crude Oilsa Adhesion After Adhesion After Additional Initial Weathering Weathering Adhesion (mass % loss in (mass % lost in Type of Crude Oil Before Release weathering) weathering) Light Crudeb 0 2 (25%) 9 (64%) Medium Crudec 12 17 (10%) 33 (32%) Heavy Cruded 75 98 (10%) 605 (19%) Diluted Bitumene 98 146 (6%) 1580 (20%) Bitumen 575 aData in g/m2. bScotia Light. cWest Texas Intermediate. dSockeye Sour. eCold Lake Blend. SOURCE: Hollebone22

32 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE 2-3 The relative proportions of light versus residual components in crude oils. SOURCES: Hollebone22 and Environment Canada31

TABLE 2-7 Comparison of Important Crude Oil Properties Adhesion Density Viscosity Flash Type of Crude Oil (g/m2) (g/cm3) (mPa·s) point (ºC) Light Crudea 0 0.77 1 −30 Weatheredb Light Crude 9 0.84 5 95 Medium Crudec 12 0.85 8 −10 Weathered Medium Crude 33 0.90 112 >110 Heavy Cruded 75 0.94 820 −3 Weathered Heavy Crude 600 0.98 475,000 >95 Diluted Bitumene 98 0.92 270 −35 Weathered Diluted Bitumen 1,580 1.002 50,000 >70 aScotia Light. bAfter additional weathering. cWest Texas Intermediate. dSockeye Sour. eCold Lake Blend. SOURCES: Hollebone22 and Environment Canada31

CHEMICAL AND PHYSICAL PROPERTIES OF CRUDE OILS 33

CONCLUSION Crude oils are mixtures of hydrocarbon compounds ranging from smaller, volatile compounds to very large, nonvolatile compounds. The hydrocarbon structures found in oil include saturates, aromatics, and polar compounds that include resins and asphaltenes. The resins and asphaltenes are largely recalcitrant in the environment. They evaporate, dissolve, and degrade poorly and thus may accumulate as residues after a spill. The percentage of the saturates and aromatics—herein called the light components, in comparison to the heavy, residue-forming resins and asphaltenes—varies with oil type and is summarized in Figure 2-3. The physical and chemical properties of diluted bitumen differ substantially from those of other crude oils, with key differences highlighted in Table 2-7. The distinct physical and chemical properties of diluted bitumen arise from two components: the bitumen provides the high- molecular-weight components that contribute most to density, viscosity, and adhesion; and the diluent contributes the low-molecular-weight compounds that confer volatility and flammability, and that determine the rate at which evaporation increases the density of the residual oil. Because diluted bitumen has higher concentrations of resins and asphaltenes than most crude oils, spills of diluted bitumen products will produce relatively larger volumes of persistent residues. Such residues may be produced relatively rapidly when gas condensate has been used as the diluent, and these weathered residues display striking differences in behavior compared to other oils: exceptionally high levels of adhesion, density, and viscosity.22 Contrasts between diluted bitumen and other crude oils are strongly enhanced by weathering. Weathered heavy crude and especially weathered diluted bitumen are, for example, much more adhesive than the other oils. The densities of the oils also vary, with weathered heavy oils approaching the density of fresh water and weathered diluted bitumen possibly exceeding the density of fresh water.

Environmental Processes, Behavior, and Toxicity of Diluted Bitumen

INTRODUCTION This chapter is concerned with what happens to crude oil after it is released from a pipeline, and with the potential environmental and ecological consequences of that release. It thus considers the chemical and physical processes affecting the oil and its residues, the resulting behaviors that manifest across various environmental settings, and the toxicity of the spilled and eventual residual oil. The discussion further distinguishes ways in which the behavior of diluted bitumen is similar to or distinct from that of the light and medium crude oils that are commonly transported in U.S. pipelines.

ENVIRONMENTAL PROCESSES As mentioned in the previous chapter, crude oil released to the environment experiences a host of chemical and physical changes—processes collectively referred to as “weathering.” For the purposes of this chapter, weathering processes are divided into three categories based on the nature of the chemical and physical effects on the oil. Chemical processes include photooxidation and biodegradation and cause alteration to molecular structures through the cleavage and formation of covalent bonds, the linkages between atoms. Physical-chemical partitioning processes include evaporation and dissolution. These act without changing molecule structures and partition material, for example, between the atmosphere and a liquid phase. Physical processes include spreading, dispersion, emul-

36 SPILLS OF DILUTED BITUMEN FROM PIPELINES

sification, adhesion, and sedimentation. All of these change the physical properties and behavior of the oil but do not always partition it between phases or change its molecular structure. Processes occurring for a crude oil spill on water are outlined in Fig- ure 3-1. The details of each process are discussed in this section. Given the statement of task, the focus is on the processes most relevant to oil spilled in subaerial and aquatic continental and coastal environments, including groundwater, lakes, estuaries, and streams of all sizes. By tracing the flow of the oil, where the origin of the spill represents day 0 of the spill, a general sequence of processes that occur can be summarized.

Chemical Processes Although many processes act on spilled oil, few processes lead to chemical decomposition. The two processes of greatest relevance to oil spills in the environment are photochemical oxidation and biodegradation. These processes tend to occur slowly over a period of weeks to years and represent the breakdown of oil at the molecular level.

Photooxidation Photochemical processes result from exposure of spilled oil to sunlight, leading to cleavage and formation of covalent bonds. Oxygen is typically incorporated into the products and thus the term photooxidation is commonly used. These oxidized products include both carbon dioxide and other oxygenated compounds.32 Typically, aromatic hydrocarbons are transformed more rapidly than alkanes,33 thereby increasing the relative abundance of resins and asphaltenes in the residual oil.32,33c In one set of laboratory experiments, the photooxidation of crude oil in fresh water under direct ultraviolet irradiation showed oxidation of 5% of the branched alkanes, 9% of the linear alkanes, and 37% of the aromatic hydrocarbons.34 Although photooxidation can be important when spills receive intense solar exposure, the photooxidation of diluted bitumen in the environment has hardly been studied and its role in the weathering of spilled diluted bitumen is not well understood. Notably, one potential outcome of photooxidation of crude oils is the production of persistent molecules in the environment32 containing carboxylic acids and alcohols, which may be soluble in water. Thus, these photoproducts may be transported in surface waters or groundwater.35 Photochemical enhancement of toxicity has been demonstrated for some polycyclic aromatic hydrocarbons (PAHs) via photomodification and photosensitization. In one study,36 organisms exposed to PAHs from

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 37

Figure 3-1 Processes affecting the composition, amount, and behavior of diluted bitumen.

38 SPILLS OF DILUTED BITUMEN FROM PIPELINES

spilled oil experienced up to a 48-fold greater toxicity when illuminated with natural sunlight instead of standard laboratory light. This mechanism of toxicity is important for early-life-stage and translucent organisms that often accumulate PAHs in their tissues and inhabit surface waters.

Biodegradation Biodegradation is a process by which living organisms, mainly bacteria, degrade hydrocarbons.37 Biodegradation can occur either aerobi- cally or anaerobically, with aerobic processes typically occurring more rapidly and extensively. Biodegradation is accelerated in the presence of abundant oxygen and nutrients,38 moderate temperature and salinity, and reasonable oil-water interfacial surface area.39 For diluted bitumen, the extent of biodegradation depends on a combination of environmental factors, the proportions of bitumen and diluent, the nature of the diluent, and how fast the diluent is lost to evaporation. Since the deposits from which bitumen is extracted are themselves residues remaining after extensive anaerobic biodegradation, a spill of diluted bitumen may be less susceptible to biodegradation than a comparable spill of light or medium crude oil. However, biodegradation of diluted bitumen in environments containing aerobic bacteria and nutrients warrants further study.40 The main classes of crude oil components highlighted in Chapter 2 (see Figure 2-1)—saturates, aromatics, resins, and asphaltenes—provide a useful framework for understanding the relevance of biodegradation to the environmental fate of diluted bitumen. There are no quantitative field studies on the biodegradation of spills of diluted bitumen, but saturates and aromatics are expected to biodegrade within weeks to years. Conversely, resins and asphaltenes in the bitumen are expected to remain recalcitrant for a longer time. The U.S. Environmental Protection Agency (USEPA) studied short- term biodegradation in the laboratory using residual oil in sediment from the Kalamazoo River that was collected 19-20 months after a diluted bitumen spill.41 Over 28 days of aerobic incubation of sediment slurries with inorganic nutrients added, about 25% of the total petroleum hydrocarbons degraded, mostly in the first 14 days. The decreasing rate of biodegradation over the 28-day period suggested that the majority of the spilled oil would not degrade over time scales of at least a few months in spite of the experiment’s favorable conditions for bacterial activity.

Physical-Chemical Partitioning Processes Once crude oil is spilled from a transmission pipeline, its composition can be affected by evaporation of volatile compounds and aqueous disso-

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 39

lution of water-soluble compounds. These processes tend to occur rapidly and strongly impact the composition and behavior of residual spilled oil.

Evaporation Following a spill that brings oil into contact with the atmosphere, light components will evaporate at relative rates that depend on their volatility. As a result, compounds with greater volatility tend to evaporate from oil more rapidly than those with lesser volatility. This relative relationship tends to hold across various environments but absolute rates vary substantially based on concentrations of the volatile compounds in the oil and on ambient conditions including exposed surface area and volume; temperature of the oil, water, and air; and velocities of the wind current.42 Fingas43 has argued that the importance of the wind speed is moderated by the fact that the supply of hydrocarbon molecules to the oil-atmosphere interface is often a limiting step for evaporation. For diluted bitumen, as with other crude oils, evaporation of light components can occur readily. The relatively light natural-gas condensates often used as the diluent in diluted bitumen are particularly volatile. The loss of volatiles thus leads to a residue strongly resembling the original bitumen, and this is a key behavior that distinguishes diluted bitumen from other commonly transported crude oils. As noted in Chapter 2, such processes increase both density and adhesion of the residual oil. The relationship between density and submergence is considered further here, whereas adhesion is considered in the context of physical processes. The increase in density that occurs with evaporative loss of the diluent increases the likelihood that the residual oil will submerge beneath the water surface and potentially sink to the bottom.8,9b The rate at which density increases will depend on the composition of the diluted bitumen and especially on the nature of the diluent, but significant density increases have been observed to occur over in the first 1-2 weeks of a spill. Diluted bitumen with relatively high proportions of light and heavy

hydrocarbons and a paucity of compounds in the C15-C25 range, such as Access Western Blend (AWB), Christina Lake, and Borealis Heavy Blend, are expected to achieve a higher density more rapidly with evaporation. As evidence, King et al.9b conducted oil weathering studies in an experimental tank placed outside in Dartmouth, Nova Scotia, Canada (Figure 3-2). As the curves fit to the experimental data in Figure 3-2 show, within 13 days, the density of residues derived from Access Western Blend exceeded that of fresh water, and the oil sank until reaching an underlying layer of saline water. The maximum density of AWB (1.008 g/cm3) is greater than that of the density of weathered residues of Cold Lake Blend (1.0014 g/cm3) and it takes less time for AWB to approach its final density

40 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE 3-2 Observed increases in density for two diluted bitumen (AWB, Access Western Blend; CLB, Cold Lake Blend) added to a flume tank containing seawater, compared to the typical densities of fresh water and seawater. Curved lines show model fits to the observations. SOURCE: King, et al.9b

in comparison with CLB. The time difference could be important when responding to an oil spill, where a difference of 12 hours could result in submergence of the oil.

Dissolution If the spilled oil is in contact with water, components that are at least slightly soluble in water will be lost by dissolution. For diluted bitumen, the most soluble components originate from the diluent and, because of their volatility, also tend to be readily lost by evaporation. There- fore, unless the oil propagated or spread under water for an extended time, such as in the Deepwater Horizon oil spill,44 evaporative losses are expected to be larger than dissolution losses. However, evaporative losses may be slower than dissolution if a spill spread under the ice in an ice- covered lake or river. Also, dissolution may become important for diluted bitumen that has percolated to or been released beneath the water table.

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 41

Physical Processes Oil released to the environment may experience changes due to a host of physical processes including spreading, dispersion, emulsification, adhesion, and sedimentation. While these processes do not affect the molecular composition of oil directly, they do result in complex, environment-dependent interactions with the chemical and physical-chemical partitioning processes.

Spreading On land, spreading of spilled oil is often limited, but when oil reaches a water surface it starts to spread immediately.45 Unless constrained, the oil will continue to spread out into a thin film, or slick, due mostly to interfacial tension.45 Spreading rapidly increases the footprint of oil in the environment and can make recovery efforts more difficult, but it also makes the oil more exposed to photooxidation and evaporation. Small amounts of crude oil can spread into a very thin layer, or “sheen,” that is readily visible. Such sheening represents an impairment of water quality that can determine the degree of oil recovery that is required, as happened in the case of the spill in Marshall, Michigan (Box 3-1, Figure 3-3). Submerged or sunken oil residues, mostly out of sight, can still serve as a source of a sheen. The effect can continue for long periods, either spontaneously in response to temperature and water- level changes, or as a result of disturbance of the sediments by animals, boats, etc.

Dispersion In the context of oil spills, dispersion refers to the entrainment of oil droplets in the water column. The extent of oil dispersion depends on the interfacial tension between oil and water, oil viscosity, and the mixing energy that may be driven by wind, currents, or tides.46 The interfacial tension between the oil and water does not vary widely among oil types. The range is typically less than twofold. When applied, chemical dispersants can decrease interfacial tension by 10- to 200-fold,47 allowing a greater proportion of oil to disperse into the water column. The mixing energy varies across environments as well as over space and time in a particular environment (e.g., rivers passing through dams or cataracts, and windy versus calm weather on lakes and coastal marine waters). The droplet size distribution (DSD) of oil dispersed in water plays an important role in the behavior of oil in the aquatic environment. Larger droplets are more buoyant than smaller droplets and thus rise to the water surface, regardless of whether they were released underwater or released

42 SPILLS OF DILUTED BITUMEN FROM PIPELINES

BOX 3-1 Marshall, Michigan: Enbridge

The largest release of diluted bitumen into the environment occurred in July 2010 when the Line 6B pipeline operated by Enbridge Energy Partners LLC ruptured and released an estimated 843,000 gallons of oil into a tributary of the Kalamazoo River near Marshall, Michigan.12, 52 This spill is recognized as one of the largest, and most costly, inland spills in North America, with estimated costs exceeding $1 billion. Oil flowed a couple of miles down the tributary and entered the Kalamazoo River, ultimately affecting about 40 miles of stream and river channels. The release occurred when the river was at flood stage and had temporarily inundated its floodplain; shortly thereafter, falling water levels left oil stranded on vegetation and soils up to about a meter above the normal summer river level. The river carries a lower suspended sediment load (i.e., less turbid) than most U.S. rivers and is not particularly turbulent because it has a low elevational gradient. The U.S. Environmental Protection Agency (USEPA) quickly assumed control of the emergency response under the National Contingency Plan (NCP) and worked with Enbridge and a host of federal, state, and local government agencies.12 The USEPA remained in this function through 2014, an exceptionally protracted response period. The Michigan Department of Environmental Quality is still engaged in remediation as of 2015. Initial response focused on capturing and collecting floating oil using conventional techniques such as conventional and sorbent boom, and the majority of the oil was recovered as floating oil or deposits on land, including from the wetland at the source of the release. A major wildlife rehabilitation effort commenced, ultimately cleaning and releasing over 3,000 turtles as well as some mammals and birds. Visible oil on floodplain and riparian vegetation was removed, mostly manually. Within a few weeks it became apparent that significant amounts of the oil had sunk to the bottom of the river. Recovery of sunken oil increasingly became the focus of response efforts after the initial autumn, although some heavily oiled islands and floodplain areas still required excavation. Oil accumulated on the bot-

on the water surface and entrained into the water column by turbulence. In the Deepwater Horizon spill in the Gulf of Mexico, where the oil was released at depth, large droplets (> 1.0 mm) rose almost vertically and reached the surface within hours,44c,48 while small droplets rose more slowly or became permanently entrained.49 Studies of surface oil slicks50 predicted rapid transport of large oil droplets and slow transport of smaller droplets, causing the formation of a comet-shaped oil slick on the water surface. The DSD affects not only the transport but also the fate and toxicity of spilled oil. Increasing the proportion of small droplets results in an increase in the surface area per unit mass of oil, which enhances the

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 43

tom wherever the river flow slowed down, but particularly behind three areas with manmade dams, and in some oxbows (abandoned river channels). Detection and quantification of the sunken oil was challenging.9a Laboratory measurements of total petroleum hydrocarbons suffered from interferences by natural organic matter. Sunken oil was mapped out using a method called poling, in which the sediments were disturbed with a disk on the end of a pole and the amount of sheen and floating globules that appeared on the surface was estimated. Poling showed that the oil tended to accumulate in areas of slow or no flow and fine, often highly organic sediment. Chemical fingerprinting, requiring an expensive set of measurements, was employed selectively to establish that the sunken oil and the sheen it produced were in fact from the pipeline release and not a legacy from other pollution sources. With the exception of dredging, proven techniques for recovery of this type of sunken oil in a riverine setting were lacking. Sediment agitation and collection of the resultant sheen was employed extensively in an attempt to recover the sunken oil without sediment removal, but eventually that approach was shown to be inadequate and dredging was conducted in the most extensive contaminated areas. In the later stages of the cleanup, a net environmental benefits analysis was conducted to examine the distribution of sunken oil, as indicated by poling, and to determine whether further recovery attempts were justified against the environmental disturbance of recovery operations.12 Many small areas of sheen-producing sediments were left alone but continue to be monitored. The USEPA estimates that as much as 80,000 gallons of oil may remain in the sediments, but major recovery efforts ended in 2014. The toxicity of the spilled oil to fish and wildlife received some study. A his- topathological study of fishes conducted shortly after the spill showed evidence for toxic effects, whereas short-term bioassays with invertebrates conducted in later years were less conclusive. Sampling of fishes and benthic invertebrates indicated that communities returned to normal after the first year, although the apparent decline in the first year was clearest for invertebrates, and it is difficult to distinguish the relative roles of direct toxicity of the spilled oil from the effects of cleanup activities (particularly sediment disturbance).

dissolution of hydrocarbons in the water column44b and their potential for bioaccumulation via transport across biological membranes. The time for a slick to break up and reach a stable droplet size distribution (the equilibrium DSD) is also an important factor because the delay enables other processes to act on the oil in a differential way. For example, turbulence could change during this time, leading toward another equilibrium DSD, or particles of a certain size might preferentially interact with mineral or organic particles, thus forming oil-particle aggregates. Figure 3-4 shows how the median droplet size varies with time for oils with various viscosities subjected to agitation at an energy dissipation

44 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE 3-3 Sheening in the Kalamazoo River, the site of a diluted bitumen spill (see Box 3-1). Left: Sheen emanating from an island after a modest rise in river level (the spill occurred at a higher river level). Right: Sheen generated from sunken oil upon disturbance of the sediments. Photo credit: USEPA.

FIGURE 3-4 Evolution of the median droplet diameter (d50) with time as a function of the oil viscosity, based on model simulations.53 High-viscosity oils would require hours to disperse into an equilibrium droplet size distribution.

rate (ε, W/kg) representative of spilling breaking waves. Even under these conditions, which can be expected to be rare in inland waters, but more likely in a coastal environment, oils with viscosities >10,000 cP require many hours to approach equilibrium, leaving ample opportunity for other processes to act.

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 45

Emulsification Emulsification is the process by which one liquid is dispersed into another one in the form of small droplets.51 To be called an emulsion, the product must have some physical stability; otherwise the process is called water uptake rather than emulsification. Water droplets can remain in an emulsified oil layer in a stable form and the resulting material, sometimes called mousse or chocolate mousse because of its appearance, has properties differing strongly from those of the parent oil. Water can be present in oil in five distinct ways. First, some oils contain about 1% water as soluble water. This water does not significantly change the physical or chemical properties of the oil. Second, oil can contain water droplets without forming a stable emulsion. They are formed when water droplets are incorporated into oil by wave action and the mixture is not viscous enough to prevent droplets from separating from the oil. Mesostable emulsions represent the third way that water can be present in oil and are formed when the small droplets of water are stabilized by a combination of the viscosity of the oil and the interfacial action of asphaltenes and resins. The viscosities of mesostable emulsions are 20-80 times higher than that of the starting oil. These emulsions generally break down within a few days into oil and water or sometimes into water, oil, and emulsion remnants.51 Mesostable emulsions are viscous liquids that are reddish brown in color. The fourth way that water exists in oil is in the form of stable emulsions, which form in a way similar to mesostable emulsions and persist because the concentration of asphaltenes and resins is high enough to stabilize the oil-water interface. The viscosity of stable emulsions is 800- 1,000 times higher than that of the starting oil. This emulsion will remain stable for weeks and even months after formation. Stable emulsions are reddish brown in color and appear to be nearly solid. Because of their high viscosity and near solidity, these emulsions do not spread and tend to remain in lumps or mats on water or shore. The fifth way that oil can contain water is by viscosity entrainment. If the viscosity of the oil is such that droplets can penetrate, but will only slowly migrate downward, the oil can contain about 30% to 40% water as long as it is in a turbulent water body. Once the water calms or the oil is removed, the entrained water slowly drains. Typically most of the water will be gone before about 2 days. The formation of emulsions is an important event in an oil spill. First, and most importantly, emulsification substantially increases the actual volume of the spill. Emulsions of all types initially contain about 50% to 70% water and thus, when emulsions are formed, the volume of the spill can be more than tripled. Even more significantly, the viscosity of the oil

46 SPILLS OF DILUTED BITUMEN FROM PIPELINES

increases by as much as 1,000 times, depending on the type of emulsion formed. For example, oil that has the viscosity of a motor oil can triple in volume and become almost solid through the process of emulsification. These increases in volume and viscosity make cleanup operations more difficult. Oil in stable emulsions is difficult or impossible to disperse, to recover with skimmers, or to burn. Mesostable emulsions are relatively easy to break down, whereas stable emulsions may take months or years to break down naturally. Emulsion formation also changes the fate of the oil. When oil forms stable or mesostable emulsions, evaporation slows considerably. Biodegradation also slows. The dissolution of soluble components from oil may also cease once emulsification has occurred.

Adhesion Adhesion was considered in Chapter 2 as a physical property of oil, and diluted bitumen tends to be more adhesive than other commonly transported crude oils. In this section, adhesion is further considered in the context of weathering, as an environmental process that changes the physical properties and behavior of oil. When diluted bitumen is spilled and the diluent evaporates, the residual oil will increasingly adhere to surfaces. Many inland and coastal waters contain submerged, floating, and emergent aquatic vegetation and debris that would provide surfaces for adhesion of bitumen. In addition, various animals may also become coated with oil, including turtles, amphibians, insects, and mammalian species (see Figure 3-5). Residual bitumen can be difficult to remove from tree trunks and other biota, rocks, concrete, and manufactured surfaces (Figure 3-5). Strongly adhesive behavior is not unique to bitumen, as some heavy oils with high contents of resins and asphaltenes can also become highly adhesive following weathering. However, diluted bitumen is unique in terms of the rate at which its physical properties can begin to revert back to those of bitumen, due to the evaporative loss of volatile hydrocarbons that comprise the diluent. The emergence of strong adhesion following the evaporative loss of volatile components can impede recovery efforts and, as discussed further below, is expected to increase the tendency of the residue to adhere to particulate matter and sink.

Sedimentation In the context of this report, sedimentation refers to oil that sinks and comes to rest on the underlying bed in an aquatic setting. Sedimentation may occur by several routes, including an increase in density of the oil through physical-chemical partitioning or chemical processes, the adhe-

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 47

(a) (b)

FIGURE 3-5 Biota coated in adhesive oil: (a) Oil that coated the shoreline after the Marshall spill into the Kalamazoo River in Michigan (Box 3-1) as the river level fell in August 2010. Photo credit: S. Hamilton. (b) Residual bitumen oil on a tree along the Kalamazoo River in September 2015, over 5 years after the spill. Photo credit: S. Hamilton. (c) Oiled map turtle from the Kalamazoo River, collected shortly after the spill. Photo credit: Michigan Department of Natural Resources. (d) Ladybird beetle collected at a natural hydrocarbon seep off Coal Oil Point, Santa Barbara, CA. Like bitumen, the seep oil is highly adhesive because of its high resin and asphaltene content. Photo credit: D. Valentine and R. Harwood.

sion of entrained (dispersed) droplets of oil to the bed, and formation of oil-particle aggregates (OPAs) of sufficient density to submerge. This section focuses on OPAs, although the high density of weathered diluted bitumen is expected to increase sedimentation by all three routes, relative to commonly transported crude oils. Aggregation of oil with natural particulate matter can cause submergence of an initially floating oil.9a There are two major types of OPAs: oil droplets coated by small particles54 and oil trapped within or adhering to large particles (Figure 3-6). The first type is more common and has

48 SPILLS OF DILUTED BITUMEN FROM PIPELINES

FIGURE 3-6 Left: Types of oil-particle aggregates (OPAs): A, single and multiple oil droplets (yellow) aggregated with natural particles (blue); B, solid aggregate of a large, usually elongated mass of oil with interior particles (dashed blue circles); and C, flake aggregate of thin layers of clay that incorporate oil and fold up (c.f. Stoffyn-Egli and Lee55 and Fitzpatrick, et al56). Right: Kalamazoo River sediment spiked with weathered source oil from the Marhsall, MI spill after 48 hours, under ultraviolet epifluorescence microscopy at 320× magnification.57 Images from Fitzpatrick et al.9a

been studied in some detail. The particles on the oil droplet surfaces prevent coalescence with other droplets, thus stabilizing the suspension. The OPAs tend to separate from the oil droplets from which they form, and then they may sink because the aggregate of residual oil (density near, or slightly less than, that of water) with inorganic particles (density at least twice that of water) yields an aggregate particle density that is neutrally or negatively buoyant in water. The formation of aggregates depends on the viscosity of the oil droplet, the surface areas and mineralogy of the particles, and the salinity of the water.58 Salinity enhances the formation of OPAs, but the onset of aggregation is very steep, becoming important at salinities as low as 1/200 that of seawater.54b, 58b The extent of adsorption also depends on the surface properties of the sediment particles and other particles such as algal cells. The relative quantity of residual oil produced by weathering is from 5- to 50-fold greater than the quantity produced by weathering of equal volumes of light and medium crude oils. In quiescent water, the density of the OPA determines whether it remains at the surface or sinks to the bottom. However, when turbulence is present, the suspension of the OPA in the water is due to the balance between hydrodynamic forces that tend to lift the OPA and keep it in suspension versus the gravitational force on the OPA. This means that density alone does not determine the location of the OPA; its size and shape

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 49

are also important. A large OPA would be less prone to displacement by turbulence than a smaller one and, thus, tends to settle faster than a smaller OPA. Because viscous oils tend to break into large droplets,59 the OPAs resulting from these oils would tend to be large, and if their density is greater than that of the receiving water, they would settle to the bottom. This formation of large OPAs is a major distinction between diluted bitumen, which rapidly weathers to a viscous residue, and commonly transported crude oils. Two recent laboratory studies dealt specifically with the formation and sedimentation of oil-particle aggregates in fresh water. The results of the first of these studies60 are fully consistent with the second, far more detailed report, by Waterman et al.,61 which focused specifically on the mechanisms underlying the submergence and deposition of diluted bitumen residues in the bed of the Kalamazoo River. Cold Lake Blend, a diluted bitumen, was added to pure water and agitated by bubbling air at 21ºC. The study found that it lost at most 17.4% of its mass and reached a density of 0.993 g/cm3, thus remaining buoyant.61 However, when the diluted bitumen was added to a mixture of Kalamazoo River water and sediment collected upstream from the spill site, the results were very different. Abundant OPAs formed quickly, on time scales of minutes to an hour, and included both inorganic and organic particles. Measured settling velocities ranged from 1 to 11 mm/s, with most being around 2 mm/s. This observation—that aggregates formed readily in a natural, fresh water—is not in conflict with laboratory studies showing that aggregates form slowly, if at all, in nonsaline water. The natural water is definitely “fresh,” but its content of ions derived from clays and soils is high enough to allow the formation of aggregates. Specifically, the electrical conductance of the water from the Kalamazoo River was 640 µS/cm, typical of North American river waters and indicating a salinity about 1/100 that of seawater. Water temperature and salinity are important determinants of the propensity of residual bitumen to submerge. Short62 reviewed existing experimental studies of the submergence of weathered bitumen and concluded that the studies had not sufficiently considered the differential effects of temperature on the densities of bitumen products versus that of freshwater or seawater. The density of bitumen increases faster with decreasing temperature than does the density of water, and therefore a weathered bitumen may sink in cooler water while floating in warmer water, yet the studies did not investigate a range of temperatures. Fur- thermore, Short62 noted that where there is salinity stratification with fresh or brackish water overlying seawater, as is particularly common at freshwater inlets to coastal marine zones, submerged oil may accumulate at density interfaces beneath the surface. Such interfaces can be dynamic

50 SPILLS OF DILUTED BITUMEN FROM PIPELINES

under the influence of winds, tides, and changing freshwater discharge, presenting a particularly challenging situation for crude oil detection and recovery. In sum, in a diluted bitumen spill subject to weathering, there is much more residue and its density is much closer to that of water, a combination that is likely to translate to enhanced OPA formation and oil submergence relative to commonly transported crude oils.

ENVIRONMENTAL BEHAVIOR This section addresses how oil spills behave in and potentially impact specific environments, with particular attention to how spills of diluted bitumen may be similar to or distinct from those of other commonly transported crude oils. The toxicity of diluted bitumen is reviewed in a following section. Water bodies merit particular attention. Pipelines traverse innumer- able inland water bodies ranging from headwater streams to wetlands to larger rivers and lakes. Spills can enter water directly or indirectly via overland flow or transport in groundwater. These inland waters are immensely diverse. Once an oil spill contacts water bodies, it may be more difficult to recover, it can potentially be transported quickly from the site of release, it presents a hazard to aquatic life (see toxicity discussion below), and it may impair human uses of the water body. The land-water interface often presents special challenges for oil spill cleanup. Water levels in lakes, rivers, and coastal zones vary under the influence of precipitation and runoff, dam operations, and/or tides. Accordingly, oil floating on the water can subsequently become stranded in intermittently flooded areas, from which it could be remobilized later. The land-water interface is often disproportionately important to its area, serving as habitat for fish and wildlife and providing recreational and aesthetic values for people.

Spills on Land and into Groundwater When diluted bitumen or other crude oil is released onto land, or into soils beneath the ground surface, its movement in the ground depends on the soil type, oil viscosity, and the depth to the water table. Oil perco- lates deeper in coarse sediments, such as gravel, than in fine sediments, such as fine sand, silt, and clay. Extensive studies of a spill of crude oil at Bemidji, Minnesota (Box 3-2), have provided an in-depth understanding of the behavior of crude oil in a subsurface setting. If crude oil is sufficiently fluid to move in the subsurface, it would migrate downward until reaching the water table, where it will sit above the water and could

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 51

BOX 3-2 Groundwater Contamination in Bemidji, Minnesota

One of the most comprehensively studied sites for crude oil transport and fate in groundwater has been at Bemidji, Minnesota, where 10,700 barrels of crude oil, characterized as light to medium,67 spilled from a failed weld on a pipeline in August 1979. About 2,500 barrels seeped into the subsurface and could not be recovered. Remediation of this site remains a challenge today. In 2011, reviewing three decades of remediation and investigation, scientists from the U.S. Geologi- cal Survey wrote, “Considerable volumes of NAPL oil [non-aqueous-phase liquid oil] still remain in the subsurface despite 30 years of volatilization, dissolution, and biodegradation, and 5 years of pump-and-skim remediation.”68 The current approach for the contaminated Bemidji aquifer is to rely on natural attenuation, including both aerobic and anaerobic biodegradation by microorganisms.64 Some of the liquid oil reached the groundwater table, where it has floated on top of the water, while some was retained by sediments in the unsaturated zone. As depicted in the cross-section figure below, various processes have affected the amount and composition of the subsurface oil, including differential sorption to the soils and underlying glacial deposits, evaporation of volatile components into the soil air space, dissolution of the more soluble components into groundwater, and biodegradation. Natural attenuation by biodegradation has substantially reduced the amount of oil and slowed its movement relative to the underlying groundwater, even though rates are limited by the availability of oxygen and other oxidants. The oil has gradually migrated in the direction of groundwater flow, yielding a plume of liquid oil, an associated vapor plume in the overlying unsaturated zone, and a plume of dissolved constituents that moves with the groundwater. The different zones in the figure below have distinct concentrations of petroleum contaminants, dissolved oxygen, and the products of microbial reactions that degrade the petroleum compounds. Research at the Bemidji site has provided a wealth of information on the rates and limitations of the microbial processes that biodegrade oil, and this information has been integrated into a mathematical model that informs decisions about remediation of subsurface oil spills.

SOURCE: Delin et al.69

52 SPILLS OF DILUTED BITUMEN FROM PIPELINES

propagate horizontally along the groundwater flow gradient.63 In most instances, the underground migration of the free phase oil (or diluted bitumen) is limited, and many small crude oil spills onto land have been remediated by excavating the contaminated soils. In the case of diluted bitumen, if spilled oil is exposed to the atmosphere, the evaporation of the diluent and the light fractions increases the oil viscosity and thus limits oil migration in the ground. However, if not lost to evaporation, the light components of crude oil, especially those that are more soluble in water such as benzene, toluene, ethyl benzene, xylenes (BTEX), and naphthalenes, could migrate relatively easily in the aquifer, especially if the aquifer material contains only small amounts of organic matter. Examples of such materials include gravel and sand aquifers, typical of glacial outwash. For organic-rich soils, water-soluble oil components could be adsorbed onto sediments, depending on their water-to-octanol 63 (KOW) partition coefficients. A concern with crude oil spills underground is the mobilization of trace metals such as arsenic, as has been observed at the Bemidji site in Minnesota.64 If oil were spilled into karst aquifers, where transmission of groundwater can be exceptionally rapid, its behavior might be expected to be comparable to that in surface waters. Rapid migration of the oil could pose a threat to water supplies for drinking or irrigation. However, to date there is no experience with diluted bitumen spills into karst aquifers. The surface soils at the site of the release of diluted bitumen from a pipeline near Marshall, MI, as well as the downstream floodplains of the Kalamazoo River, were significantly contaminated (Box 3-1). At the release site, diluted bitumen saturated soils in a small wetland before overflowing into a creek. Large amounts of residual oil were deposited on downstream floodplains as river levels fell. Much adhered to vegetation and other surfaces as the diluent evaporated. Residual oil did not move long distances in groundwater, presumably because the groundwater flows to the river from the surroundings in almost all of the affected reaches.65 Had the flows been from the river to the alluvium, a common situation, groundwater contamination may have been more pervasive. The ecological impacts of the oil deposited on land at the Marshall, MI spill were not studied in depth. Because remobilization of oil by subsequent floods was a major concern, oil-coated vegetation and soils were largely removed within months wherever oil deposits were visible. Little immediate mortality of wildlife was reported. Removal of oily vegetation and soils severely disturbed the riparian and floodplain ecosystems, even though this was done manually in all but the most heavily oiled locations, where excavation was necessary. In terrestrial ecosystems, spills of diluted bitumen and of light or medium crude oils present similar challenges for cleanup. Contaminated

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 53

soils, vegetation, and man-made surfaces must be removed or cleaned. The most distinctive feature of diluted bitumen in this regard is its tendency to be strongly adhesive and to coat surfaces, and the difficulty of washing it off. Contact with the sticky bitumen by wildlife and people would increase exposure although toxicity implications are not clear given limited data (see Toxicity section, Chapter 3). The bitumen would be expected to be highly persistent due to its abundance and resistance to dissolution or biodegradation.

Inland Lakes and Reservoirs Inland lakes and reservoirs vary greatly in depth, exposure to wind and currents, suspended matter, and submerged and emergent aquatic vegetation, all of which potentially affect the behavior and impacts of oil spilled into these environments. Spilled oil may enter a lake or reservoir along a shoreline, or from tributaries. Submerged oil carried in suspension (including as OPAs) by streams and rivers may accumulate in depositional areas once the water enters a lake or reservoir, and this is particularly pertinent to spills of diluted bitumen because of their greater propensity to form OPAs, as discussed above. Lakes and reservoirs are often used as water supplies and for recreation, thus raising the possibility of risks to the public, and they provide important habitat for fish and wildlife. The 2010 spill in Marshall, MI (Box 3-1) reached a reservoir known as Morrow Lake, and that location provides our only experience to date with a diluted bitumen spill in a reservoir. The extensive delta at the upper end of the reservoir, which begins 60 km downstream of the pipeline release, accumulated submerged oil. Sheening was observed during warm months for 3 years after the spill. A hydrodynamic model indicated the potential for resuspension of the sediment containing oil and its movement further into the reservoir.66 The most heavily oiled sediments had to be dredged in 2014. The Great Lakes system of the U.S. and Canada has distinct characteristics that would affect the behavior and impacts of an oil spill. Transmis- sion pipelines capable of transporting diluted bitumen products� cross the Great Lakes system at two points: the Straits of Mackinac between Lake Michigan and Lake Huron,70 and the St. Clair River upstream of Detroit and Lake Erie. A release at either the Mackinac Straits or the St. Clair River would lead to movement of oil into the lakes. Additionally, pipelines cross many streams and rivers that flow short distances to either the southwestern shores of Lake Superior or the southern shores of Lake Michigan. Currents can be complex in the Great Lakes, with currents in the Straits of Mackinac depending on relative water levels of Lakes Michigan

54 SPILLS OF DILUTED BITUMEN FROM PIPELINES

and Huron as well as on wind speed and direction. It could be very difficult to anticipate the movement of the spilled oil and to recover the oil, even at the surface, due to the expansive area and potential for strong wave action. Ice cover during winter could impede detection and recovery of spilled oil.

Wetlands Pipelines often traverse wetlands, which are common landscapes in the U.S. in all but the arid West. Wetlands are highly variable ecosystems in terms of their hydrology (water sources, flow, and connectivity with water bodies) and ecology (vegetation and animal life). An oil release into a wetland may affect a limited area if the system is hydrologically isolated or has only seasonal flow. However, wetlands commonly occur along land-water interfaces of lakes, rivers, and coastal marine waters, and spilled oil could spread from wetlands into adjacent waters, or vice versa, if they are connected via surface water flow. Wetlands are valued and protected because they provide important ecosystem services and can harbor rare or threatened plants and animals. Oil spills into wetlands present challenges because of the sensitivity of these ecosystems;71 for example, in Marshall, MI (Box 3-1), oil trapped in floodplain swamps and marshes had to be recovered manually by cutting vegetation and scraping soils wherever possible, and care had to be taken to not alter the natural hydrology (e.g., by building access roads). Unless the water is deep enough that submergence of oil poses an issue, diluted bitumen and commonly transported crude oils pose many similar challenges following release in wetlands, but three factors specific to diluted bitumen can be mentioned. First, the amount of residual oil will be large compared to those produced by spills of conventional crude oils. Second, the increased level of adhesion of bitumen may complicate operations in a wetland environment. And third, diluted bitumen residues may persist longer in wetlands, as in other environments, because of their resistance to biodegradation.

Streams and Rivers Streams and rivers vary in several ways that are important for the transport and fate of oil released into them. Key factors include gradients and velocities of flow; the type and concentration of suspended particulate matter; the types and abundance of underwater algae and plants; the extents and types of zones in which oil may be deposited, such as sedimentation in impoundments and side channels; and varia-

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 55

tions in flow and water levels. Submerged oil carried by streams and rivers can continue to move downstream until reaching depositional zones in water bodies or floodplains, where it can settle in relatively extensive areas. The transport of oil in rivers could be also affected by whether the reach is gaining or losing water by exchange with groundwater. Rivers may transport sediment (and associated crude oil) by wash load, suspended load, and bed load.72 Wash load consists of very fine particles that are relatively evenly distributed through the water column; examples include suspended clays and organic matter. The wash load may represent those particles that would interact with spilled diluted bitumen and influence its initial behavior. Knowledge of bed material composition does not allow one to predict wash load transport. Suspended load is the load that is suspended in the water column but still interacts with the bed. The interaction usually occurs at riffles causing the shear stress to increase and sediments to be suspended from the lee side of the ripples. Bed load has continuous contact with the bed and has a direct relation to the turbulence along the bottom in rivers. While both suspended and bed loads are commonly predicted using sediment transport models, the wash load is not predicted by these models, limiting the prediction of OPA formation in rivers. Furthermore, prediction requires a better understanding of which types of sediment are most likely aggregate with oil to form OPAs, and how those OPAs are likely to be transported in streams and rivers. The answers to these questions are likely to be different for diluted bitumen compared to other commonly transported crude oils.

Estuaries and Coastal Zones Pipeline spills can readily reach coastal zones if they occur in tributaries or along estuaries, or at coastal processing facilities. Coastal zones in the U.S. where diluted bitumen spills could occur in the future as diluted bitumen pipelines and infrastructure expand include Anacortes in the Pacific Northwest, the Gulf Coast, and Portland, Maine. Each of these regions has distinct physical and ecological characteristics that would influence the fate and ecological effects of spilled diluted bitumen. Many oil spills have occurred in coastal areas29 and have included heavy refined oil products, though until recently the only one in North America that involved diluted bitumen occurred in Burnaby, British Columbia. The recent Refugio oil spill in California also entered a coastal area and all active cleanup for this spill ended August 31, 2015 (Box 3-3). Estuaries occur at the interface of terrestrial and marine systems. Estuaries are characterized by periodic reversals of flow due to tides and

56 SPILLS OF DILUTED BITUMEN FROM PIPELINES

Box 3-3 Santa Barbara, California: Refugio Pipeline Spill

On May 19, 2015, a pipeline rupture occurred near Santa Barbara, California, on Line 901, operated by Plains All American Pipeline L.P. The heavy crude oil was discharged from a heated transmission pipeline and comprised a blend of oils from four nearby platforms. Despite the chemical and physical differences between the heavy crude oil spilled at Refugio and diluted bitumen, several aspects of the event provide for a valuable point of comparison for this study, which led to a site visit from members of the study committee.

Unanticipated Complexities

The May 19, 2015, rupture to Line 901 occurred on the inland side of a four-lane highway during daylight hours, though an earthen berm paralleling the highway prevented motorists from seeing the oil. The discharged oil pooled in a local de- pression and then overflowed to a drainage culvert that ran under the berm. The oil flowed unseen to a second drainage culvert that ran underneath the freeway, to a drainage that led under a railway line, down a gulley to a cliff, and then to a sand and gravel beach from which it entered the ocean. The total discharge was estimated at 101,000 gallons, with 20,000 gallons estimated to have reached the ocean. The partitioning of the discharge between land and ocean dictated response actions from both USEPA and the U.S. Coast Guard, and a joint command structure was established in which two federal On-Scene Coordinators, one from each agency, shared the lead responsibilities for the response. The bifurcation of discharge and the resulting structure of the Unified Command underscore the importance of response planning for diluted bitumen spills in a terrestrial setting that might subsequently affect lacustrine, riverine, and/or coastal marine environments. This incident further underscores the need for coordinated planning among federal agencies for response to diluted bitumen spills.

by the presence of freshwater-saltwater mixing zones. Large estuaries can be strongly affected by wind and waves. The interaction of fresh and saline waters in estuaries is important not only for physical transport of spilled crude oil but also because mixing of these waters often results in the flocculation and settling of clay particles. Where crude oil is associated with these particles, salinity gradients can be important in controlling the sinking rate. Coastal zones are characterized by three-dimensional flows due to waves, longshore drift, and rip currents. Although there are numerous regional ocean models, their highest spatial resolution tends to be around

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 57

Coastal Zone Transport and Submerged Oil

The product spilled from Line 901 was a heavy crude oil and much of the mass floated at the sea surface. The heavy crude oil that reached the ocean was initially transported along the coast to the east, driven by longshore drift, and opposite to the direction of the prevailing offshore current. This transport gave way to heavy oiling of beaches extending several miles to the east of the rupture. Transport offshore by the prevailing current and then to the southeast caused oiling events and beach closures in Ventura and Los Angeles Counties, up to 100 miles away. Scientists identified submerged and sunken oil in near-shore reefs and kelp beds, near the most heavily oiled beaches. The submerged oil was presumably ballasted by mineral and organic material to which it had adhered while on the beach and in the surf zone. The documentation of submerged oil by scientists triggered response officials to conduct a formal assessment of submerged oil, the result of which was a finding that quantities were too low to warrant removal efforts.

Photo credit: D. Valentine

1.0 km, a scale that is too large to quantify the processes occurring in the coastal zone. While waves in coastal areas tend to bring the crude oil to the beach, there are situations where regional, longshore, and rip currents transport crude oil offshore and parallel to the shoreline and deposit it miles away from where it was released (Box 3-3). One distinct characteristic of coastal waters (including estuaries) that affects the potential for crude oil to submerge is salinity, which ranges from fresh to oceanic levels and can fluctuate seasonally or daily due to tides and river discharge. Salinity affects the density of the water, and as noted earlier, OPAs that may sink in brackish water can remain in suspen-

58 SPILLS OF DILUTED BITUMEN FROM PIPELINES

sion in more saline water. Water flow in coastal zones is often complex and this extends to spatial and temporal patterns of salinity. Tides cause bidirectional flow of water in channelized areas and coastal currents are also commonly induced by waves. Wave action can be produced by local winds and also ocean swells. Freshwater inputs may lie above saltwater. Suspended particulate matter, composed of mixtures of inorganic sediment, detrital organic particles, and algae, also is spatially and temporally variable in coastal waters but often occurs at high concentrations compared to offshore waters. All of these features complicate the prediction of the behavior of an oil spill in these environments, including its propensity to form OPAs and become submerged. There has been only one significant release of diluted bitumen into a coastal marine zone—the Burnaby Inlet in British Columbia (Box 3-4)— and in that case most of the spilled oil was recovered quickly. However, there is a great deal of experience with crude oil spills in coastal zones because of the high tanker, barge, pipeline, and refinery activities in close proximity. These have included spills where the crude oil became submerged or sank mostly as a result of oil-particle interactions.46a

Beaches Crude oil penetration and persistence in beaches (or river banks and bars) is affected by both crude oil properties and beach hydrodynamics. The mobility of crude oil in porous media such as sand is inversely proportional to its viscosity and, thus, high-viscosity oils (such as weathered diluted bitumen) tend to penetrate very little within the sediment. High adhesion of heavy crude and weathered diluted bitumen oils further reduces their mobility in porous media, making them as amenable to recovery as other crude oils. In tidally influenced beaches, crude oil deposition is usually highest in the upper intertidal zone and decreases moving toward the lower intertidal zone. However, crude oil buried in the upper intertidal zone tends to weather faster than that in the lower intertidal zone due to high water exchange resulting from tidal action and often from local groundwater inputs that replenish oxygen and nutrients.73 For example, groundwater movement was found to be a major factor in the persistence of the Exxon Valdez oil in Alaska beaches.38b

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 59

BOX 3-4 Burnaby, British Columbia: Kinder Morgan

On July 24, 2007, approximately 1,400 barrels (58,800 gallons) of Albian heavy synthetic crude oil leaked from the Westridge Transfer Line in Burnaby, British Columbia. The spill resulted from an excavator bucket striking the pipeline during excavation for a new storm sewer line. The pipeline was operated by Kinder Morgan Canada and owned by Trans Mountain Pipeline L.P. The pipeline linked the Burnaby terminal to the Westridge Dock, where oil could be loaded to tankers.37b After the oil was spilled, it flowed in Burnaby’s storm sewer systems until it reached Burrard Inlet.37b In total, 11 houses were sprayed by the rupture, 50 properties were affected, 250 residents voluntarily left, and the Burrard Inlet’s marine environment and 1,200 m of shoreline were affected by the spill.37b Cleanup took months and cost roughly $15 million and resulted in the recovery of approximately 1,321 barrels of oil.75 Responders used three distinct methods to recover the oil, based on the oil’s location. In residential areas, peat moss was used successfully to absorb oil on land. In storm sewers, oil in the storm sewers was vacuumed up. Much of the oil was collected in the pump station. And finally, in Burrard Inlet, booming was used to contain oil around the release points, skimmers and absorbent pads were used to remove oil, and tarballs, debris, and oiled macroalgae were manually removed. To treat the oil that had adhered to the shoreline, responders successfully used the chemical shoreline cleaner Corexit 9580.76 An estimated 35 barrels of oil were not recovered and were considered to be released to the marine environment. The recovery effort during the Burnaby Harbor spill was relatively successful. Because the synthetic crude traveled on a predictable path through the storm sewer system, the responders were able to set up booms in a quick and efficient manner. The effects of the spill were limited due to favorable conditions for recovery:77

• There was sunny weather with little stormwater flow (slowed the movement of oil in storm drains and facilitated evaporation of oil). • The spill occurred on slack tide (incoming tide helped keep the oil near shore while booms were placed, and helped limit the movement of oil in the Inlet). • It occurred outside the primary migration and overwintering period for birds, and after the breeding bird season. • It occurred prior to the main period of salmon migration to creeks and rivers for spawning.

There were no reports of the oil sinking or becoming submerged in the water column.

60 SPILLS OF DILUTED BITUMEN FROM PIPELINES

TOXICITY

Toxicity of Commonly Transported Crude Oils As described in Chapter 2 and displayed in Table 2-1, crude oils contain a variety of chemical constituents whose percentage compositions differ widely depending upon the specific product or blend, and thus generalizations about toxicity are difficult. However, the toxicity of typical constituents of crude oil has been demonstrated by many studies. Specific toxic compounds include the monoaromatic hydrocarbons comprising BTEX. The acute toxicity, respiratory issues, and potential carcinogenic properties of BTEX are of concern regarding the health of humans and of wildlife. Various diluted bitumen products show lower BTEX levels (percentage volume)74 compared to light and medium crudes, with values similar to those of heavy crudes (Table 3-1). Other crude oil components of interest include the low-molecular- weight linear alkanes, other monoaromatics, and 2-ring PAHs that are often of acute aquatic toxicity concern, mainly due to acute narcosis-based mechanisms of toxicity. Also of concern are the 3- to 5-ring (unsubstituted

TABLE 3-1 Five-Year Average Concentrations and Ranges of BTEX in Various Crude Oils in Percent Volumeii Light Crude Medium Crude Heavy Crude Diluted Bitumen Benzene 0.22 0.42 0.12 0.16 (<0.01-0.38) (0.14-0.77) (0.02-0.21) (0.06-0.28) Ethylbenzene 0.26 0.35 0.11 0.07 (0.24-0.27) (0.19-0.60) (0.05-0.18) (0.04-0.11) Toluene 0.67 0.87 0.26 0.32 (0.03-1.14) (0.29-1.34) (0.12-0.45) (0.18-0.47) Xylenes 0.99 0.77 0.36 0.33 ± 0.05 (0.18-1.46) (0.43-1.09) (0.23-0.49) (0.27-0.43) TOTAL BTEX 2.10 2.34 0.84 0.89 (0.24-3.26) (0.97-3.11) (0.56-1.26) (0.64-1.16) ii Data are from the Crudemonitor database accessed May 2015. Values in parentheses list the range of concentrations across oils in each category. All crude oils with data spanning at least two years were included for comparison. The crude oils include five light crude oils (BC light, Boundary Lake, Koch Alberta, Pembina light sour, and Scotian light); four medium crude oils (Medium Gibson sour, Midale, Peace Pipe Sour, and West Texas Intermediate); eight heavy crude oils (Lloyd blend, Wabasca Heavy, Western Canadian Blend, Bow River North, Bow River South, Fosterton, Seal Heavy, and Sockeye 2000) and eight diluted bitumen (Access Western Blend, Cold Lake Crude, Western Canadian Select, Christina dilbit blend, Borealis Heavy Blend, Kearl Lake).

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 61

and alkylated) PAHs, which have been demonstrated to cause both acute and sublethal toxicity via numerous mechanisms of action that are important in delayed responses and in long-term residual and chronic effects, such as carcinogenesis, reproductive failures, developmental deformities, and immune suppression. A summary of PAH concentrations in some commonly transported crude oils and diluted bitumen is presented in the previous chapter (Table 2-2). There is a wealth of toxicological literature describing the sublethal effects of commonly transported crude oil components.29 The individual- level acute and sublethal effects include early-life-stage developmental defects, reduced growth and reproductive capacity, behavioral impairment, altered bioenergetics, genetic damage, impaired immune function and hence disease resistance, and enzymatic and hormonal changes including endocrine and hypothalamic-pituitary-adrenal axis implications, which can lead to population- and ecosystem-level impacts (e.g., changes to the base of food webs that affect higher-level consumers). In addition to direct, chemically based mechanisms, crude oils can also result in acute and sublethal effects via physical mechanisms. The physical coating of biological surfaces impedes an organism’s movement and can alter behavior and/or hamper respiration. An example of this would be the coating of gills and permeable skin surfaces of fish.

Toxicity of Diluted Bitumen As noted in Chapter 2, many of the chemical compounds in diluted bitumen are found in other crude oils, and thus toxic properties are expected to be similar in many respects, although the relative proportions differ. The bitumen is a concentrate of the high-molecular-weight components of a conventional crude, while the diluent is a concentrate of the low-molecular-weight components. In contrast to other crude oils, very limited data exist on the toxicity of diluted bitumen, although much is known about the diluent components as they are commonly found in other crude oils. The potential for diluted bitumen to significantly weather, adhere to particles, submerge, and possibly sink in quiescent areas, coupled with its high content of recalcitrant resins and asphaltenes, can result in unrecoverable sunken oil and thus prolonged chronic exposure of benthic organisms. Without a more detailed chemical characterization of this initial chemical pool, coupled with characterization of weathering or biodegradation products the comparison of diluted bitumen toxicity to other commonly transported crudes cannot be completed. As discussed in Chapter 2, a large fraction of diluted bitumen consists of an array of currently uncharacterized chemicals. This situation is not unique to diluted bitumen and

62 SPILLS OF DILUTED BITUMEN FROM PIPELINES

applies to other crude oils; however, diluted bitumen has a larger number of unknown polar compounds. Furthermore, the adhesion of residual bitumen oil to biological surfaces may lead to physical coating of organisms (Figure 3-5), impacting movement, behavior, feeding, thermoregula- tion, and respiration. There are very few published laboratory experiments specifically investigating the toxicity of diluted bitumen, despite its use and transport in North America for over 40 years. Currently, there is only one laboratory study investigating the toxicity of a diluted bitumen, working with larval fish (Japanese medaka, Oryzias latipes).78 Sublethal concentrations of soluble components of the oil or “water-accumulated fraction” derived from Access Western Blend caused an increased prevalence of blue sac disease, impaired development, and abnormal swim bladders upon hatching. In addition, exposures resulted in various genetic markers for physiological stress that are commonly observed with exposure to other crude oils. It has been well established that exposure to conventional crude oils can cause embryo toxicity in fish primarily due to the 3- to 5-ring alkylated PAHs. Concentrations of 3- and 4-ring unsubstituted parent and alkyl PAHs are similar or higher in Cold Lake Blend in comparison to other commonly transported crudes as listed in Table 2-2 in Chapter 2, although that blend contains higher levels of total phenanthrenes, fluorenes and chrysenes compared with light and medium crude oils. Therefore, based on the limited available research, it is expected that exposure to diluted bitumen would cause similar or higher chronic toxicity to fish embryos, although further chemical characterization of diluted bitumen along with more toxicity studies in other fish species would be required to confirm this. Delayed effects from acute or chronic exposure or chronic toxicity studies of the residual bitumen component have also not been investigated. Other additional mechanisms of action and sub-lethal effects to other species and life stages have also not been studied for diluted bitumen. While there have been few experimental dose-response investigations of diluted bitumen toxicity, the toxicity of diluted bitumen spilled into the environment has been inferred from post-spill field observations and investigations as well as laboratory studies using field samples from recent diluted bitumen spills. In addition, organisms collected near the oil sands deposits have been examined.79 Natural bitumen deposits are exposed in the banks of rivers in the Athabasca oil sands region, and studies of wild fish from these locations have found sublethal biochemical and hormonal responses, including the classic response to exposures of PAHs, namely increased levels of ethoxyresorufin-O-deethylase activity and a reduction in steroid production in comparison to fish from reference areas.79d For early life stages of fish, these biochemical responses can

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 63

be related to the observed deformities in embryos and larvae following exposure to waters affected by oil sands.79c The significance of the field observations was confirmed by laboratory-based studies of sediment toxicity in water bodies of the oil sands region.79a,79b Compared to controls, fish eggs and embryos of fathead minnow (Pimephales promelas) and white sucker (Catostomus commersonii) showed increased mortality, reduced hatching success, delayed timing of hatching, abnormal development of embryos, and deformations and reduced size of larvae. The same effects have been observed during exposures to weathered conventional crude oils. Bitumen contains several metals that are potentially toxic and are discussed in further detail in the human toxicity section below. However, potential bioavailability and toxicity of metals in the diluted bitumen transported by pipelines remains little studied. Studies in the vicinity of the oil sands mining sites in Alberta have documented increased concentrations of cadmium, copper, lead, mercury, nickel, silver, and zinc in snowmelt (reflecting atmospheric emissions and redeposition) and river waters, reaching levels of concern for the protection of aquatic life.7b Com- monly measured metals, such as vanadium and nickel are found at higher levels in bitumen (and diluted bitumen) compared to other crude oils but these metals are predominantly bound in organic-metal porphyrin structures that are less bioavailable. Few data address this issue, however, and weathering and biodegradation processes have the potential to release these metals. Overall the toxicity (chronic and acute) of fresh and weathered diluted bitumen and its residues to freshwater, estuarine, and marine species at various life stages is at this time very understudied compared with other commonly transported light and medium crude oils. Furthermore, bioaccumulation and impacts to the food web and trophic transfer issues have not received attention for diluted bitumen in contrast to that of commonly transported crudes. In the case of the spill of diluted bitumen into the Kalamazoo River (Box 3-1), toxicological effects on fishes were studied by a team from the U.S. Geological Survey within the framework of the Natural Resource Damage Assessment (NRDA).80 Fish were sampled within a few weeks of the spill to obtain a gross pathological assessment of general health. When fish in oiled reaches were compared to fish in an upstream reference reach, significant adverse changes were evident. These differences were not observed in subsequent years. The likelihood of submergence and sinking of weathered diluted bitumen, often as OPAs, merits particular attention because it presents distinct routes of exposure to the biota. In addition, the sunken oil may not be recoverable, thus resulting in protracted periods of exposure. Many aquatic animals also consume particles directly or indirectly from the bot-

64 SPILLS OF DILUTED BITUMEN FROM PIPELINES

tom sediment, underwater plant or macroalgal surfaces, or the water column, which may include oil droplets and OPAs. Contact of droplets with respiratory surfaces (e.g., fish gills) or with permeable dermal surfaces can interfere with respiration. Organisms feeding on oil-contaminated material can in turn be consumed by predators, which can pass contaminants up the food web.

Human Health The major ways in which humans may be exposed to chemicals following an oil spill are via (i) inhalation of volatile organics, (ii) dermal (skin) exposure through direct physical contact with the oil, (iii) ingestion of contaminated drinking water, and (iv) ingestion of contaminated food. This section aims to compare potentially hazardous characteristics of diluted bitumen with those of other commonly transported crude oils and identify whether diluted bitumen may pose similar or distinct human health risks. Inhalation and dermal exposure are typically immediate concerns for the first responders and the public in the vicinity of the spill. Dermal exposure can be addressed by use of appropriate personal protective equipment, and so the focus here is on inhalation hazards and contamination of drinking water and food sources. While the issue of drinking water contamination can also be immediate, such concerns may persist for some time and may occur well beyond the initial site of the spill, depending on the fate and transport of the spilled oil. Similarly, contamination of agricultural produce (e.g., by irrigation water) and fisheries may be a concern in the near term. However, there may also be longer-term concern for food safety. During a spill, water bodies for human recreational use (swimming, boating, and fishing) are closed until deemed safe. For these reasons, the immediate human health issues are considered first and then the potential longer-term human health concerns are considered separately.

Initial human health concerns During the initial days of spill response, the major components of concern to human health in crude oils include the volatile compounds— benzene, toluene, ethylbenzene, and xylenes (collectively called BTEX)

and hydrogen sulfide (H2S)—that can result in acute and sublethal effects via inhalation exposure. Benzene is also a well-known human carcinogen. Benzene is typically present in crude oils and is frequently monitored to assess both inhalation and drinking water supplies. Health and safety

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 65

concerns regarding exposure to volatile organic compounds have also been addressed in the context of response (see Chapter 4). Oil spills can occur in populated areas and result in immediate human exposure; the 2013 Mayflower, Arkansas, oil spill in a residential area exemplifies the importance of an immediate, effective response to mitigate human health hazards (Box 3-5). Table 3-1 shows the 5-year average BTEX concentrations and ranges for light, medium, heavy and diluted bitumen crude including the four representative crude oils described in Chapter 2.74 The data indicate that the BTEX constituents in diluted bitumen (reported in % by volume) are not distinctly different from other crude oils. The average BTEX in diluted bitumen at 0.89 % vol was similar to the heavy crude oils at 0.84 % vol, whereas light and medium crude oils were 2.56 and 2.80 % vol respectively. The light crude oils category represented the highest variability. By comparison, the 5-year average values reported for all condensates listed in the crude monitor database averaged 4.22% BTEX by volume.74

Exposure to hydrogen sulfide (H2S) gas is also of concern as it can immediately damage the central nervous system and act as a chemical asphyxiant.81 Given that hydrogen sulfide is corrosive, its levels were discussed in the previous report.2 Diluted bitumen typically contains similar or lower levels of hydrogen sulfide compared to the other crude oils. Another immediate (and potentially longer-term) concern relative to human health is the contamination of drinking water in the vicinity of or downstream of the spill. The more soluble BTEX components, especially benzene, are of most concern for drinking water and are also discussed in Chapter 4. In water, the solubilities of BTEX compounds range from approximately 150 to 1,800 mg/L82 making them significantly more soluble than most other hydrocarbons. The USEPA has established water quality standards for BTEX compounds that are regulated by the Safe Drinking Water Act (SDWA). USEPA’s National Primary Drinking Water Regulations set maximum contaminant level (MCLs). For example, MCLs for benzene are 0.005 mg/L (5 parts per billion or ppb). In a 2015 pipeline spill of Bakken crude oil into the Yellowstone River in Montana, oil related volatile organic contaminants (e.g., benzene) were detected at levels of concern at a drinking water treatment plant forcing the closure of the water intakes. Given the similar concentrations of BTEX in diluted bitumen and other commonly transported crude oils in the U.S. pipeline system, it is not expected that the BTEX components in diluted bitumen will pose a higher immediate risk to human health during the initial spill response phase. Of additional concern with respect to immediate human (and any exposed organism) toxicity is the presence of elevated concentrations of

66 SPILLS OF DILUTED BITUMEN FROM PIPELINES

Box 3-5 Mayflower, Arkansas: ExxonMobil

On March 29, 2013, the 20-in. Pegasus oil pipeline, constructed in 1947-1948 and operated by ExxonMobil Pipeline Corporation, ruptured in Mayflower, Arkan- sas, releasing 3,190 barrels of Wabasca Heavy crude oil in a residential area. The Pegasus pipeline is buried 24 in. underground at the release site. The oil flowed down the street and into a drainage ditch and tributary that led to Dawson Cove, an arm of Lake Conway. The upper part of Dawson Cove was heavily forested and flooded at the time of the spill. The lower part of Dawson Cove is separated from the lake by a road with open culverts. Response crews were on site within 30 min of detection of the release, including ExxonMobil Pipeline employees and federal, state, and local responders. Staff from the Arkansas Game and Fish Commission quickly constructed earthen berms at the head of Dawson Cove, which allowed the oil to be contained within the cove, with no oil documented as entering Lake Conway. Wabasca Heavy crude oil is a blend typically composed of ~80% bitumen obtained by polymer injection and water flooding from the Athabasca region and 20% diluent; this blend typically has a 19-20 API gravity, ~1% BTEX, and 4.15% sulfur.74 Twenty-two homes were evacuated. Air quality monitoring in residential areas in the first week or so after the release reported benzene below detection (0.05 ppm) but volatile organic compounds (VOCs) of up to 29 parts per million (ppm) on the day of the spill and 3 ppm by the third day. Higher levels of both benzene and VOCs were measured in work areas, and workers wore respiratory protection as specified in the health and safety plan. There were no reports of oil sinking in the quiet waters of Dawson Cove. However, the dense vegetation had to be removed to access the oil, and intensive mechanical methods were used to remove trees, shrubs, rootballs, and other or-

unknown volatile and/or water-soluble compounds. This concern is not unique to diluted bitumen and is of concern for all crude oils.21 However, the concentrations of polar compounds such as those containing nitrogen are higher in diluted bitumen. Therefore, there may be chemicals of toxicological concern unique to diluted bitumen that have not yet been characterized. This consideration is an important caveat for an assessment of relative human health risk for diluted bitumen spills compared to spills of commonly transported crude oils: an assessment based only on known BTEX concentrations must be considered incomplete and therefore tentative.

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 67

ganic debris from about 14.5 acres of forested wetland. The intensive mechanical operations mixed oil into the soils in the cove area. The post-cleanup assessment found low levels of PAHs in the soils that were mostly below sediment toxicity concerns. Nonetheless, because of chronic sheening in the cove, in August 2014 ExxonMobil was required to (i) excavate 800 yd3 of sediment from the 1,300-ft- long tributary upstream of the cove, and backfill as needed with clean sediment; (ii) place an organoclay soil amendment on 2 acres of sediment surface within the remaining vegetated area; and (iii) place a reactive cap on about 4.5 acres of open water. This spill shows that quick response can be very effective. Diluted bitumen products will initially float with the lack of turbulence in fresh water, and at that time they can be cleaned to levels of low toxicity, but are persistent and can cause chronic sheening that can trigger the need for extensive treatment, particularly in inland areas where natural removal processes are slow.

Photo Credit: U.S. Environmental Protection Agency

Longer-term human health concerns In addressing longer term human health concerns, protection of water supplies is a focus of spill response activities. For reasons described earlier in the report, weathered diluted bitumen has a greater potential to submerge or sink, presenting a greater potential for chronic contamination of a water supply that may result in a long closure time for drinking water sources. Another serious outcome in the case of incomplete removal of sunken weathered bitumen could be a longer lasting impairment of a surface-water source of drinking water.

68 SPILLS OF DILUTED BITUMEN FROM PIPELINES

In the United States, groundwater supplies about 32% of drinking water83 and there have been studies of crude oil contamination of groundwater, such as the site in Bemidji, MN (Box 3-2). Given the potential for longer travel time of crude oil in groundwater systems, the impacts may be detected later than for surface waters and may be inherently more difficult to remediate than for surface water bodies. The environmental behavior of a diluted bitumen spill with respect to groundwater contamination is discussed above. There are various delineations of groundwater zones in which limits are set for both residential and industrial zones84 for acceptable levels of Total Petroleum Hydrocarbons (TPH) and BTEX components. In the recent diluted bitumen spill in Marshall, MI, over 150 drinking water wells were monitored for a variety of inorganic and organic oil (and non- oil) related chemical contaminants. The contaminants monitored included nickel, vanadium, and organic compounds, including BTEX. TPH in the

gasoline range (GRO, ranging in carbon atoms from C6-C10) and the diesel range (DRO, >C10-C28) were measured. The public health assessment report released in 201385 stated that there were no detections of crude oil related chemicals in any of the sampled wells, other than iron and nickel. These elevated levels of iron and nickel were only found in a few of the over 150 wells sampled and were deemed to be a natural occurrence for the wells in Calhoun and Kalamazoo Counties. The absence of groundwater contamination along the river was not unexpected given that groundwater flows toward the river and its floodplain in almost all of the affected reach.86 When there is deemed to be a risk of human exposure via food consumption, the collection of fish and other food items (e.g., shellfish) may be prohibited over space and time based on sheening and/or the presence of targeted compounds in the tissues of food items. The main class of compounds that are measured for food safety is the polycyclic aromatic hydrocarbons (PAHs), although others, including metals may be monitored. As Table 2-2 in Chapter 2 shows, there are a number of higher molecular weight PAHs that the USEPA has listed as probable human carcinogens for a representative set of light, medium, and heavy crudes, in addition to diluted bitumen. Out of the 16 EPA priority PAHs listed, 11 have slightly higher to over 3 times higher levels of these carcinogenic compounds in Cold Lake diluted bitumen compared to the four other crude oils. For example, benzo[a]pyrene levels in Cold Lake diluted bitumen are 3.01 µg/g compared to the light and medium crudes ranging from 0.25–0.74 µg/g. As mentioned above there also are many unknown compounds in crude oils for which we do not know the environmental fate, bioavailability and potential impact to organisms. Further chemical composition details are required, together with assessments on bioavail-

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 69

ability and toxicity to organisms. Furthermore, details on how and to what the chemical constituents are biodegraded when diluted bitumen is spilled into the environment or metabolized to by organisms that bioac- cumulate oil constituents are also required for a full understanding of the toxicity of diluted bitumen. Accumulation of oil-derived metals into the food chain and ultimately into food for human consumption is also possible. However, levels of metals are usually very low in crude oils and only a few are measured (typically, nickel, vanadium, copper, cadmium and lead). Levels of vanadium and nickel in the various crude oils are reported as these are usually the metals of highest concentration compared to other trace metals. Using the 5 year average levels of vanadium and nickel in the same oils as detailed for the BTEX comparison shows that diluted bitumen concentrations are over 7 and 2.5 times higher than in the example light and medium crude oils respectively. For example, five year average levels of nickel and vanadium in diluted bitumen are 60 and 152 mg/L respectively compared to 8 and 20 mg/L in light and 23 and 60 mg/L in medium crudes. The bioavailability and toxicity of metals can also be very dependent upon the specific form (speciation) of the metal, which is dependent upon a variety of environmental parameters, including redox status (related to the presence or absence of oxygen) and pH content, both of which can be modified due to an oil spill. However, these measurements do not take the bioavailability of these metals into consideration and they are likely to be less bioavailable given that they are commonly found in tightly bound organic structures (porphyrins) in diluted bitumen. Although evidence of food web contamination from a diluted bitumen spill is lacking, risks for benthic organisms and their consumers cannot be ruled out given the limited data available regarding chemical composition and how biodegradation and other weathering processes might change the chemical composition of the residual oil. A more complete understanding of the chemical constituents of all crude oils and their weathering products is necessary to support a more thorough toxicological assessment. Based on the limited available evidence for diluted bitumen, however, it appears that it would pose a lower or similar hazard to human health in the short- term for the chemicals currently monitored and assessed. However, the potential toxicological risks to humans and animals, particularly longer- term exposures, are currently unknown compared to commonly transported crude oils.

CONCLUSIONS In some respects the environmental effects of diluted bitumen spills resemble those of spills of other commonly transported crude oils, as long

70 SPILLS OF DILUTED BITUMEN FROM PIPELINES

as the diluent and bitumen remain mixed and in their original proportions. The movement of the oil on land and on the water surface, and its toxicity to wildlife and people, are similar at the outset. Once a spill occurs, however, exposure of diluted bitumen to the atmosphere allows the lighter diluent fraction to evaporate, resulting in residual bitumen that has several distinctive characteristics, being particularly dense, viscous, and with a strong tendency to adhere to surfaces and to submerge beneath the water surface and potentially sink to the sediments. In light of these characteristics, diluted bitumen spills in the environment pose particular challenges when they reach water bodies. Progres- sive evaporative loss of the diluent leaves behind the relatively dense and viscous bitumen, which can then become submerged, perhaps first by adhering to particles, and ultimately sink to the sediments. The density of the residual oil need not exceed that of the water to submerge if conditions are conducive to the formation of oil-particle aggregates with densities greater than water, and this may be a common situation in inland and coastal waters where suspended particulate matter abounds. The loss of the lighter fraction and resultant potential for submergence of residual oil manifests more quickly and will involve a greater fraction of the spilled oil than in the case of light and medium crude oils. Toxicity of the residual bitumen has received little study, although toxic effects of both organic substances and associated metals have been observed in the vicinity of oil sands deposits in western Canada. The difficulty of recovering sunken oil and the recalcitrant nature of bitumen mean that aquatic biota may be exposed to the material for longer periods than in the case of lighter oils that sometimes sink to the bottom but are relatively biodegradable. Impacts of diluted bitumen spills are expected to vary across the great diversity of inland and coastal water settings, which present varying scenarios for transport and submergence of the oil, for the feasibility of recovering oil that has sunk, for the nature of the aquatic biota that would be exposed, and for human uses of the water and water bodies. Flow and mixing patterns, turbulence, and the nature and abundance of natural particulate matter are among the most important considerations. The toxicity of diluted bitumen has scarcely been studied using a direct, experimental dose-response approach in the laboratory, although it has been inferred from studies in surface waters draining the oil sands deposits, as well by post-spill sampling and bioassays. Evidence so far suggests that the BTEX in the diluent—as well as certain PAHs in the bitumen—can have toxic effects, but these same compounds also occur in other commonly transported crude oils. Many of the PAHs known to cause chronic and sublethal effects, such as alkyl PAHs and EPA priority PAHs, are at similar or higher levels than those of commonly transported crude oils. Diluted bitumen or its weathered residues may contain

ENVIRONMENTAL PROCESSES, BEHAVIOR, AND TOXICITY 71

other, uncharacterized compounds with toxic properties, but this awaits further investigation. There may also be various metabolites produced from diluted bitumen components that are also currently unknown and uncharacterized for their toxicological consequences. Until there is more toxicological research specifically targeting diluted bitumen, the acute, chronic, sub-lethal and longer-term toxicities of diluted bitumen relative to conventional crude oils will be poorly known.

Spill Response Planning and Implementation

INTRODUCTION The identification of an oil spill triggers the mobilization of personnel and equipment to protect the health and safety of the public, and to detect, contain, and recover the spilled oil while minimizing impacts to communities and the environment. Many characterizations of risk include probabilistic considerations of the magnitude and frequency of the hazard itself; for example, in the case of hurricane storm surge, these factors are estimated based on historic data. The multiple factors that can contribute to spill occurrence can include unpredictable accidents resulting from human actions; for example, the 2007 diluted bitumen spill in Burnaby, British Columbia (Box 3-4), was the result of construction activities unre- lated to pipeline operations.37b The vulnerability of communities and environments potentially affected can be assessed in advance, however, and these factors become a key component of spill response planning. For crude oil spills from transmission pipelines, the corridor of potential release is at least fixed and the impact areas are predictable. The types of environments, communities, and facilities that could potentially be impacted, and their sensitivities and vulnerabilities, can thus be identified in advance and are essential elements of spill response planning. Where the characteristics of the right of way are such that it is more or less sensitive to spills of different materials (e.g., diluted bitumen versus medium or light crude oil), these factors can be considered in advance; however, the specific characteristics of the material may change over time in any transmission pipeline operations, particularly for pipelines transporting diluted bitumen. 73

74 SPILLS OF DILUTED BITUMEN FROM PIPELINES

The transport, fate, and effects of spilled oil depend not only on the characteristics of the oil but also on the environments and conditions at the time and place of the spill. The consequences of a spill of diluted bitumen into a stream at base flow will differ from those at flood conditions based on the effect of turbulence on suspended particle formation, the availability of sediment particles for adhesion, the extent of the riparian or floodplain zone at risk, and the length of stream affected within the first few days. Similarly, the public safety issues surrounding volatilization of lighter fractions would be different in a recreational community on a holiday weekend compared to midweek during low season. Every spill presents a unique combination of conditions. Responders on scene must use their experience to adjust the response plan to the circumstances that confront them. Given this uncertainty regarding the magnitude and character of any specific incident, spill response planning for pipelines is based on the concept of the “Worst Case Discharge,” which is the largest foreseeable discharge of oil, including a discharge from fire or explosion, in adverse weather conditions. This is calculated as follows:113 The Worst Case Discharge is the largest volume, in barrels (cubic meters), of the following:

1. The pipeline’s maximum release time in hours, plus the maximum shutdown response time in hours (based on historic discharge data or, in the absence of such historic data, the operator’s best estimate), multiplied by the maximum flow rate expressed in barrels per hour (based on the maximum daily capacity of the pipeline), plus the largest line drainage volume after shutdown of the line section(s) in the response zone expressed in barrels (or cubic meters); or 2. The largest foreseeable discharge for the line section(s) within a response zone, expressed in barrels (or cubic meters), based on the maximum historic discharge, if one exists, adjusted for any subsequent corrective or preventive action taken; or 3. If the response zone contains one or more breakout tanks, the capacity of the single largest tank or battery of tanks within a single secondary containment system, adjusted for the capacity or size of the secondary containment system, expressed in barrels (or cubic meters).

In addition, pipeline operators may claim prevention credits for breakout tank secondary containment and other specific spill prevention measures.

SPILL RESPONSE PLANNING AND IMPLEMENTATION 75

This chapter outlines the main elements of spill response planning relevant to the consideration of diluted bitumen, describing the types of plans developed under the National Contingency Plan (NCP) and considering potential protection priorities for diluted bitumen spills. The roles and responsibilities of various agencies and entities in the development of these plans, including the NCP process, are described in Chapter 6. The next section focuses on activities that occur during actual spills and how these may need to vary for spills of diluted bitumen compared to crude oils. The chapter concludes with a summary of the specific challenges for spill response planning and implementation presented by the transport of diluted bitumen in pipelines.

IMPLEMENTATION OF PLANS

Predicting the Behavior of Spilled Oil In framing and scaling an actual incident, responders ask the following questions:

• What and how much was/is being spilled? • Where will it go? • What are the resources at risk? • What are the likely impacts? • What should be done to reduce these impacts?

The observed type and volume of the spilled product drives initial actions related to the safety of responders and the public, mobilization of equipment, and estimates of the crude oil’s likely behavior and pathway in the environment. Any delay in access to accurate information about the composition and properties of the spilled product can significantly affect the effectiveness of the response. Access to such information is critical, but specific compositions of products being transported by pipelines are usually not promptly available from pipeline operators or from the sources of the products they transport, and often vary over time in a particular pipeline. Compositional information is particularly important for diluted bitumen because the types and concentrations of diluents vary in ways that strongly affect the behavior of the spill and thus response strategies. Safety Data Sheets (SDSs) for crude oils are usually generic and provide ranges in reported properties, such as density; they do not provide information that responders need, such as the specific type of crude oil, density after weathering over time, chemical composition, and adhesion (which is rarely provided), among others. Because of their generic nature,

76 SPILLS OF DILUTED BITUMEN FROM PIPELINES

responders seldom obtain the data they need from an SDS. If the crude oil name is provided and the crude oil is a standard type, readily available databases, such as Crudemonitor.ca, the Environment Canada oil properties database, and the National Oceanic and Atmospheric Administration (NOAA) Automated Data Inquiry for Oil Spills (ADIOS) oil library, can be consulted to obtain some of these data. However, if the spilled material is a blend that does not have a standard composition, but rather may change significantly from batch to batch, these databases may provide incomplete or inaccurate information. In such cases, batch-specific information is needed. Modeling for guiding response activities is typically done for short durations and, thus, differs from modeling conducted to evaluate long- term impact. NOAA ADIOS is designed to provide oil weathering information for only 5 days. In addition, the NOAA Environmental Response Division uses the General NOAA Operational Modeling Environment (GNOME) to obtain modeling results of oil transport on the water surface, and the main purpose of these results is assisting the Unified Com- mand of a spill in making the appropriate response decision. Such models would need to run with a minimal number of parameters and to attempt to capture the salient features of the release in terms of direction and magnitude. While existing oil spill models can be used for the response to diluted bitumen spills, the main parameters are typically calibrated to conventional oils. For example, the windage factor, which provides the transport speed of oil, is typically equal to 3% to 4% in the early stages of a conventional oil spill,87 and it is later decreased further as the oil weathers and forms emulsions. For diluted bitumen, the residual oil density can increase rapidly with the evaporation of the volatile diluent components. Since diluted bitumen does not promote the formation of emulsions, the windage factor of a diluted bitumen is initially low (e.g., 3%) and is not expected to decrease further with time. Another challenge in using existing oil spill models for diluted bitumen is the lack of sufficient experi- mentally obtained data to calibrate the modules with diluted bitumen.

Health and Safety Concerns Because volatile organic compounds (VOCs) can evaporate rapidly when crude oil is released to the environment, public and worker safety related to air quality must be considered during the early stages of the response.88 Light crude oils, particularly those produced during hydraulic fracturing of shales (e.g., Bakken and Eagle Ford) can pose significant air quality and explosion risks early in the response. Because benzene is a known carcinogen that is present in many petroleum products, it has the

SPILL RESPONSE PLANNING AND IMPLEMENTATION 77

highest ranking in terms of potential for exceeding occupational exposure levels and community-exposure guidelines.89 However, there may also be concerns about and monitoring of other VOCs, hydrogen sulfide, and explosion hazards (Chapter 3). When air quality is a particular concern, there may be the need to establish a Public Health Unit within the Planning Section of the Unified Command to develop criteria for evacuations and reoccupation by the public. In addition, the Site Safety Plan for responders will have to follow Occupational Safety and Health Administration (OSHA) and American Conference of Industrial Hygienists guidelines for VOCs in general, and benzene, toluene, ethyl benzene, and xylenes in particular, as well as for hydrogen sulfide and other contaminants of concern. There may also be a need for real-time measurements of concentrations in work areas and for use by workers of passive air-monitoring and dosimeter badges, which are sent for analysis in order to monitor exposure. A program of this kind was implemented during the recent spills of diluted bitumen in Marshall, MI (Box 3-1),12 and Mayflower, AR (Box 3-5), where the oil spread to areas in close proximity to residential areas. Effects on water quality can also be significant. Spills of crude oil that reach water bodies can result in either closure of the affected water body to public use or posting of advisories to avoid oiled areas until there is no longer a potential for exposure. Such closures and advisories are likely to be longer when the spilled oil sinks in the water body and generates chronic sheening. For example, the Kalamazoo River and a reservoir known as Morrow Lake were closed for nearly 2 years after the Enbridge Pipeline spill in July 2010 (Box 3-1). Drinking water intakes may be shut down until testing determines that the water is safe to use, or the raw water may require additional treatment such as aeration and carbon filtration, as was conducted during the 2015 spill of crude oil from the Bridger Pipeline in Glendive, Montana, into the ice-covered Yellowstone River.90

Cleanup Endpoints Cleanup endpoints are the criteria against which the response actions are measured, to determine if the goals and objectives have been met. Cleanup endpoints generally are set for water, shorelines, and soils. For spills in coastal and marine habitats, cleanup endpoints are usually based on the following guidelines91 rather than analysis of samples for measurement of the concentration of selected contaminants:

• No oil observed: not detectable by sight, smell, or feel; • Visible oil but no more than background amounts of oil;

78 SPILLS OF DILUTED BITUMEN FROM PIPELINES

• No longer generates sheens that will affect sensitive areas, wildlife, or human health; • No longer rubs off on contact; and • Oil removal to allow recovery/recolonization without causing more harm than natural removal of oil residues.

Cleanup endpoints for a specific spill are developed through consensus among the stakeholders, which can include public health officials. Key considerations are the trade-offs between aggressive techniques that remove the oil but also cause additional damage versus less intrusive techniques that rely on natural processes to remove, dilute, or bury residual oil (collectively known as natural attenuation). Cleanup endpoints for inland oil spills tend to be more stringent than those applied to spills in the marine environment and often require the use of more intensive cleanup methods that carry a risk of increased ecological impacts92 for the following reasons:91

• Inland habitats often lack some of the physical processes (such as waves and tidal currents) that can speed the rate of natural removal of oil residues after treatment operations are terminated and can affect smaller water bodies where there are slower rates of dilution and degradation. • The direct human uses of inland habitats, such as for drinking water, recreation, industrial use, and irrigation, require a higher degree of treatment compared to marine environments to avoid human health and socioeconomic impacts. • Spills in close proximity to where people live, work, or recreate often require treatment to a higher level. • There may be large-scale differences in water levels during the response, causing oil to be stranded well above normal levels where it can pose hazards to wildlife as well as humans using these areas. • Many states have sediment quality guidelines that must be met as part of the remediation phase after the emergency response is completed.

Table 4-1, which has been adapted from Whelan et al.,92 lists guidelines for establishing cleanup endpoints for spills in inland habitats. Achieving consensus on cleanup endpoints for spills of diluted bitumen can be challenging if the crude oil sinks and continues to generate sheens in areas of high public use, or where the residual crude oil adhering to substrates is difficult to remove.

79 Continued Less aggressive removal removal Less aggressive during seasonal low- use periods could allow to work natural processes Consider how long before the oil weathers Public information campaign concerning staining remaining required Usually determined by Usually determined by manager or land resource manager experts Case studies that show habitat damage from treatment aggressive Particular sensitivity of a species or habitat habitat Inability to replace When oil residues are no no are When oil residues to human longer a threat health and safety odor Falls below threshold limits or exposure • • • • • • • • • Guidelines for No Further Guidelines for No Further Determination Treatment

a No visible oil than 20% No more stain or coat No free-floating No free-floating black oil or mousse on the water surface No accessible oiled debris No oil in sediments used that are for nesting, hibernating, for food grubbing No visible oil No detectable oil (sight or smell) Example Primary • • • • • • • Cleanup Endpoints Wipe, high high Wipe, high pressure, flush, temperature cut, remove/replace Gross oil removal oil removal Gross using vacuum, skimming, manual using removal in walking boards soft substrates of Passive recovery sheens Whatever needed Whatever needed threats: to remove excavate, cut, flush, remove/replace Treatment Methods Treatment • • • • Hard substrates substrates Hard such as bedrock, gravel, seawalls, riprap Beaches Vegetation Debris areas, T&E species T&E species areas, habitat, wildlife national refuges, parks, other areas protected High public use High public use areas Residential areas Groundwater supplies Applicable Habitats • • • • Wetlands, bird nesting nesting bird Wetlands, • • • Guidelines for Selecting Cleanup Methods and Endpoints for Different Inland Habitats Guidelines for Selecting Cleanup Methods and Endpoints Different

4-1 TABLE Basis for Treatment Removing Aesthetic Removing Aesthetic Impacts in High-Use Areas Protection of Sensitive of Sensitive Protection and Habitats Resources Protection of Public of Public Protection Health and Safety

80 High risk of erosion or or High risk of erosion excessive sedimentation Unacceptable changes in surface topography excessive change in Avoid e.g., sediment/soil quality, matter content, organic grain size Potential permanent change to the habitat type e.g., wetland to open water Education on Education on considerations between and removal aggressive sheens chronic Education on considerations between and removal aggressive sheens chronic Site specific studies to risk assess receptor Consider seasonal use and Consider seasonal use and (e.g., flooding processes that speed natural removal) Education on considerations between and removal aggressive sheens chronic Site specific studies to risk assess receptor Consider how long before Consider how long before the oil will weather to a nonsticky stain or coat excessive vegetation Avoid removal Falls below known limits for hazards Public information campaign concerning staining remaining required • • • • In low-use areas: • In high-use areas: • • In low-use areas: • In high-use areas: • • • • • • Guidelines for No Further Guidelines for No Further Determination Treatment

a No visual oil No visual oil than stain or greater coat Does not release black oil when disturbed or Agriculture for human pasture use may need a ppm endpoint No longer generates No longer generates sheens that affect sensitive resources No longer generates No longer generates sheens that affect sensitive resources No longer releases black oil or mousse during flushing operations No longer generates black oil or mousse during high-water events No longer rubs off off No longer rubs on contact off No oil that rubs on sorbents Example Primary • • • • • • • • • Cleanup Endpoints Acutely remove the the Acutely remove contamination gross (excavate, dredge, flush, cut, remove/ replace) Passively contain/ remobilized recover oil with booms and sorbents In situ techniques such as aeration, tilling, phytoremediation, adding nutrients Acutely remove the the Acutely remove of major sources sheens (excavate, flush, cut, dredge, remove/replace) Passively contain/ sheens with recover booms and sorbents Acutely remove the the Acutely remove of major sources sheens (excavate, flush, cut, dredge, remove/replace) Passively contain/ sheens with recover booms and sorbents Wipe, flush, cut, flush, cut, Wipe, sorbent barriers, remove/replace Treatment Methods Treatment • • • • • • • • 92 Upland soils River/lake bed sediments sediments Wetland Rivers, streams, Rivers, streams, other flowing water bodies Lakes, ponds, other standing water bodies Seasonally flooded wetlands Rivers, streams, Rivers, streams, other flowing water bodies Lakes, ponds, other standing water bodies Seasonally flooded wetlands Hard substrates substrates Hard such as seawalls, riprap, bedrock Vegetation Debris Soil Applicable Habitats • • • • • • • • • • • • •

Continued Secondary Cleanup Endpoints should include “Or, as low as reasonably practicable considering net environmental benefit.” practicable considering net environmental as low reasonably Secondary Cleanup Endpoints should include “Or, 4-1 TABLE SOURCE: Adapted from Whelan et al. Adapted from SOURCE: Basis for Treatment Mitigating Sediment/Soil Mitigating Sediment/Soil Contamination Mitigating Intermittent Mitigating Intermittent by Sheens (triggered rainfall, temperature changes, etc.) a Mitigating Persistent Mitigating Persistent Sheens Removing Contact Removing Contact (both humans Hazard and wildlife)

81 High risk of erosion or or High risk of erosion excessive sedimentation Unacceptable changes in surface topography excessive change in Avoid e.g., sediment/soil quality, matter content, organic grain size Potential permanent change to the habitat type e.g., wetland to open water Education on Education on considerations between and removal aggressive sheens chronic Education on considerations between and removal aggressive sheens chronic Site specific studies to risk assess receptor Consider seasonal use and Consider seasonal use and (e.g., flooding processes that speed natural removal) Education on considerations between and removal aggressive sheens chronic Site specific studies to risk assess receptor Consider how long before Consider how long before the oil will weather to a nonsticky stain or coat excessive vegetation Avoid removal Falls below known limits for hazards Public information campaign concerning staining remaining required • • • • In low-use areas: • In high-use areas: • • In low-use areas: • In high-use areas: • • • • • • Guidelines for No Further Guidelines for No Further Determination Treatment a No visual oil No visual oil than stain or greater coat Does not release black oil when disturbed or Agriculture for human pasture use may need a ppm endpoint No longer generates No longer generates sheens that affect sensitive resources No longer generates No longer generates sheens that affect sensitive resources No longer releases black oil or mousse during flushing operations No longer generates black oil or mousse during high-water events No longer rubs off off No longer rubs on contact off No oil that rubs on sorbents • • • • Example Primary • • • • • Cleanup Endpoints Acutely remove the the Acutely remove contamination gross (excavate, dredge, flush, cut, remove/ replace) Passively contain/ remobilized recover oil with booms and sorbents In situ techniques such as aeration, tilling, phytoremediation, adding nutrients Acutely remove the the Acutely remove of major sources sheens (excavate, flush, cut, dredge, remove/replace) Passively contain/ sheens with recover booms and sorbents Acutely remove the the Acutely remove of major sources sheens (excavate, flush, cut, dredge, remove/replace) Passively contain/ sheens with recover booms and sorbents Wipe, flush, cut, flush, cut, Wipe, sorbent barriers, remove/replace • • • • • Treatment Methods Treatment • • • 92 Upland soils River/lake bed sediments sediments Wetland Rivers, streams, Rivers, streams, other flowing water bodies Lakes, ponds, other standing water bodies Seasonally flooded wetlands Rivers, streams, Rivers, streams, other flowing water bodies Lakes, ponds, other standing water bodies Seasonally flooded wetlands Hard substrates substrates Hard such as seawalls, riprap, bedrock Vegetation Debris Soil • • • • • • Applicable Habitats • • • • • • •

Secondary Cleanup Endpoints should include “Or, as low as reasonably practicable considering net environmental benefit.” practicable considering net environmental as low reasonably Secondary Cleanup Endpoints should include “Or, SOURCE: Adapted from Whelan et al. Adapted from SOURCE: Mitigating Sediment/Soil Mitigating Sediment/Soil Contamination Mitigating Intermittent Mitigating Intermittent by Sheens (triggered rainfall, temperature changes, etc.) a Basis for Treatment Mitigating Persistent Mitigating Persistent Sheens and wildlife) Removing Contact Removing Contact (both humans Hazard

82 SPILLS OF DILUTED BITUMEN FROM PIPELINES

TACTICS FOR DETECTION, CONTAINMENT, AND RECOVERY OF SPILLS OF DILUTED BITUMEN

Spills to Land Spills on dry land, if detected early, are often readily contained and recovered before extensive contamination of soil or groundwater (Chapter 3). One recent study93 found that diluted bitumen penetrated a sand column more slowly than light, medium, and heavy conventional crude oils, indicating diluted bitumen soaked into sandy substrates may be no more difficult to recover than other crude oils. Problems occur when a light crude oil is released underground and not detected for days to months, or when the release is into a highly permeable substrate. When a pipeline released light crude oil underground in a gravel-outwash plain near Bemidji, Minnesota (Box 3-2), the combination of both of these factors led to one of the most extensive (and studied) incidents of groundwater contamination.68 A more recent example is the subsurface release of about 20,000 barrels of Bakken crude oil from a pipeline in agricultural land near Tioga, ND. The light crude oil penetrated more than 30 ft into the ground. Extensive excavation and treatment of soil is required and is expected to take 2 years to complete. Recovery methods for spills on land include manual and mechanical removal followed by offsite disposal, burning of oil that is pooled on the surface or in depressions and ditches, high-temperature thermal desorption or incineration (followed by return of treated soils to the spill site once they meet endpoints), and bioremediation of residual oils after gross oil removal. Cleanup of land spills is usually completed in weeks to months. When groundwaters are contaminated, cleanup is far more challenging and can extend over decades of time, with associated high costs.

Spills to Water and Wetlands

Detection Floating crude oil is detected mostly by aerial observations, ground and water surveys, and, depending on the spill size and characteristics, remote sensing. These methods are well established and effective for any floating crude oil. These methods fail, however, when the crude oil submerges or sinks. Methods of detection employed in such cases have included (i) sonar systems, (ii) underwater cameras and videos, (iii) diver observations, (iv) sorbents, (v) laser fluorosensors, (vi) visual observations from the water surface, (vii) bottom sampling, (viii) water-column sampling, and (ix) the combination of in situ mass spectrometry with autonomous underwater vehicles.94 These methods are not well estab-

SPILL RESPONSE PLANNING AND IMPLEMENTATION 83

lished, are relatively slow, often provide only a snapshot of a small area, and suffer from many limitations depending on conditions such as wave height, water depth and currents, water turbidity, and ability to detect buried crude oil. Disturbance of sediments with a disk on a pole, followed by observations of floating crude oil globules and sheen appearing at the surface, was the preferred method for field detection and mapping of submerged oil in the case of the diluted bitumen spill into the Kalamazoo River near Marshall, MI.

FLOATING OIL RESPONSE TACTICS

On Water Containment and Recovery Because most crude oils that could be released from pipelines are expected to float initially, the first response actions are to deploy booms to contain the crude oil and protect sensitive areas, and use of skimming, vacuum, and sorbents to recover the contained crude oil. When booms are deployed quickly and well, a large amount of the floating crude oil can be recovered, particularly in streams and rivers where the crude oil is contained between the banks. However, there are many conditions where floating crude oil cannot be effectively contained and recovered, including high-flow and turbulent conditions in rivers; strong winds and large waves, especially in estuaries and coastal waters; coverage of a water body by snow and ice; and limited access, such as in remote areas, floodplains, other wetlands, and difficult terrain. Under these conditions, responders look for downstream or downcurrent locations where response equipment can be effectively deployed. On rivers, these can include impoundments, bridge crossings, and boat ramps. Temporary roads may have to be constructed to gain access, particularly to wetlands and small streams. As diluted bitumen weathers and the diluent is lost by volatilization, the floating bitumen will become highly viscous and require specialized skimming and pumping systems capable of handling such high-viscosity oils. Mesocosm tests with Cold Lake Winter Blend (CLWB) with an initial thickness of 30 mm floating on water in outdoor tanks, where the oil increased in viscosity over time up to 30,000 cP, showed that conventional heavy oil skimmers were effective.9c However, under real-world conditions, depending on the source and temperature, weathered diluted bitumen can increase in viscosity up to 1,000,000 cP (see Chapter 2).94 Moreover, the residue may not continue to float (Chapter 3), a possibility not addressed by most spill response plans that exist today. Weathered diluted bitumen adheres strongly to shorelines, vegetation, and debris and will be more difficult to remove from these surfaces

84 SPILLS OF DILUTED BITUMEN FROM PIPELINES

for on water recovery by flushing methods, compared to conventional crude oils. The adhered oil will also pose a threat of fouling of habitats and wildlife because it will more quickly weather into a viscous, sticky residue.

Dispersants The efficacy of dispersants is related to the viscosity of the spilled crude oil. Based on laboratory95 and mesocosm studies9c with the diluted bitumens CLWB and Access Western Blend (AWB), interactions with dispersants exhibit roughly the same dependence on viscosity. Accordingly, dispersants are moderately effective at 10,000 cP but have little to no effectiveness at viscosities >20,000 cP. Diluted bitumen, for which the viscosity thresholds are reached within 6-12 hours under mild to moderate open water conditions and at temperatures of 15°C to 20°C, therefore have a narrower window of opportunity for effective use of dispersants than conventional crude oils. In comparison, medium crude oils are expected to reach these thresholds within 24-72 hours in temperate conditions and possibly within 12-24 hours during the winter.46a

In situ Burning Mesoscale tests9c showed that burning is viable on diluted bitumen weathered up to 1 day, with removal efficiencies of 50% to 75%. The burn residues were sticky and easily submerged; thus, there would need to be rapid removal of the residues to prevent sinking. In comparison, medium and heavy crude oils are expected to burn at 85% to 99% removal efficiencies over longer periods of weathering, thus generating much lower amounts of burn residue. Formations of stable oil emulsions containing >25% water are difficult to ignite and burn less efficiently,96 regardless of the oil type. However, the mesocosm studies showed that the water uptake in both AWB and CLWB was as a mechanically mixed and unstable oil-water combination, and not as a stable, uniform emulsion (see discussion of emulsification in Chapter 3). Such combinations would likely break up during calm-water periods, which might increase the effectiveness of a burn.

Surface Washing Agents Because diluted bitumen spills are expected to adhere strongly to surfaces, tests have been conducted using chemical agents designed to enhance crude oil removal and listed on the National Product Schedule under the heading of Surface Washing Agents. Studies have shown that

SPILL RESPONSE PLANNING AND IMPLEMENTATION 85

CLWB that had weathered on granite tiles under various conditions (on water, in sun, or in shade) could not be removed by low-pressure, ambient-temperature water, but could be removed by high-pressure, high- temperature flushing when used in combination with a surface washing agent after up to 5 days of weathering.9c Similar results were obtained during the response to the Burnaby spill (Box 3-4).76 As in all spills, early application of surface washing agents increases their effectiveness. In fact, their use has been preapproved by Regional Response Team 6 since 2003.97 During the Refugio spill, responders reported success in removing weathered oil from surfaces using dry-ice blasting, a technique that may also find application with surface cleaning of diluted bitumen (Box 3-3).

RESPONSE TO NONFLOATING CRUDE OIL AND ITS RESIDUES When crude oil is suspended in the water column or sinks to the bottom, response tactics must change. There are no known, effective strategies for recovery of crude oil that is suspended in the water column, particularly where it occurs as droplets or oil-particle aggregates. Accord- ingly, the objectives are to track the suspended material and to predict where it may sink to the bottom. Nets with various mesh sizes and towed at varying speeds have been tested to determine the adhesion and leak rates for diluted bitumen and its residues.98 Submerged material adhered to nets that extended to 0.5 m depth with minimal leakage at tow speeds of 0.3 m/s (0.6 knots) for fine and medium mesh sizes. When full, the nets weighed 25 kg/m2, making them difficult to recover by hand, and 25% to 50% of the oil leaked out when the nets were removed from the water. The recovered material stuck so firmly that the nets could not be reused. Submerged material deeper in the water column was swept under the net at water flow rates of 0.3 m/s.99 Similar results have been reported for heavy oils, indicating that the use of nets as a removal method for any type of oil suspended in the water will be of very limited effectiveness. Ideal conditions would be in low-flow, relatively small rivers or streams where the nets could be placed across the water body and readily replaced before they failed. In open water environments, submerged oil would first have to be located and the nets then deployed quickly and effectively, which seems unlikely. Other tactics for removal of oil suspended in the water column include various types of filter fences, such as gabions (wire cages) stuffed with sorbents (usually “Oil Snare,” a polypropylene adsorbent) placed on the bottom downstream from the release or snares attached to frames placed downstream. None of these tactics has been documented as effective. Tactics for removal of sunken crude oil include suction dredge, diver directed pumping and vacuuming, mechanical removal, manual removal,

86 SPILLS OF DILUTED BITUMEN FROM PIPELINES

sorbents, trawls and nets, and agitation/refloating. Suction dredging is a standard technique for removing sediments from the bottom of a water body, and it has been used to recover sunken crude oil during at least five spills and most extensively after the Enbridge pipeline spill in the Kalamazoo River (Box 3-1). This method generates large volumes of water and sediment that have to be treated and properly disposed of. Thus, it works best for surgical removal of small concentrated areas of sunken crude oil. The method used most frequently for removal of bulk crude oil that has accumulated at the bottom of a water body is diver directed pumping and vacuuming. Divers can target the sunken oil and regulate the flow to minimize removal of ancillary water and sediment. The rate of pumping must be adjusted to prevent shearing and emulsification of the oil, and to effectively move highly viscous oils. When the sunken crude oil is solid or semisolid, removal using an excavator, clamshell dredge, or other mechanical equipment can be effective and, under favorable conditions, generates little additional water or sediment for handling and disposal. This equipment is readily available and, if deposits are near shore, can be operated from land. Deployment from barges is also feasible, but there are depth restrictions (< 6 m), the equipment is large and heavy, and the rate of recovery is slow. Where the sunken oil occurs in discrete patches, manual removal in shallow water by wading, or in deeper water by divers, can be effective and allow selective recovery of crude oil if visibility is adequate. However, it is labor intensive and slow, and requires specialized gear for diving in contaminated water and special procedures and supplies for decontamination of divers. Where the sunken material consists of oil-particle aggregates, it may be possible to refloat the crude oil by agitation of the bottom. Agitation using rakes or similar tools, injection of water using water wands, and injection of air using equipment such as pond aerators were all used during the cleanup of the Enbridge pipeline spill in the Kalamazoo River (Box 3-1).9a The refloated crude oil was recovered using skimmers or sorbents. However, depending on the conditions, a significant amount of the crude oil or oiled sediment sinks back to the bottom. The agitation can also simply mix the crude oil more deeply into the sediment. Careful testing is needed to determine the effectiveness of these methods. Sometimes, crude oil from sunken aggregates returns to the surface as the water warms and the oil becomes less viscous and is able to separate from the sediment; it is not likely that the increased temperature affects the oil density relative to the water density.100 Gas bubbles released naturally from the sediment can also result in oil transport to the surface, through a process known as ebullition.101

SPILL RESPONSE PLANNING AND IMPLEMENTATION 87

WASTE MANAGEMENT AND DISPOSAL Management and minimization of wastes are key challenges during response to a spill. Any activities that increase the volume of oily waste will have a large impact on cost. For spills where the crude oil initially floats, then sinks, the response team will be faced with the management and disposal of conventional waste materials, such as sorbents, protective gear, skimmed oil, oiled solids removed from land, oiled debris, and oily liquids, as well as any materials collected during detection and recovery of sunken crude oil. If the recovery of sunken crude oil involves methods such as pumping, vacuuming, or dredging, very large volumes of crude oil, water, and sediment will be generated, requiring separation into different waste streams for further treatment prior to disposal. For example, over 237,000 yd3 of materials were removed from the Kalamazoo River and its floodplain in 2010-2014 after the Enbridge pipeline spill.102 Because the collected crude oil may either float or sink and the character of the waste stream will vary widely over time, decanting systems tend to be custom designed. Waste management is typically divided into three phases: (i) separation and treatment of solids, (ii) separation and treatment of liquids, and, where allowed, (iii) final polishing of liquids prior to release at the spill site. Where wastes can be treated on land, methods such as dewatering using geotubes (requiring a large footprint for the treatment area) and carbon treatment of water are used. Geotubes are sediment-filled sleeves of geotextile fabric. Where wastes can or must be treated at the site, a series of decanting tanks (onshore) or barges (on the water) is used, often with the goal of being able to discharge the treated water back into the spill site. Table 4-2 provides a summary of the effectiveness of selected response tactics for spills of conventional crude oils compared to spills of diluted bitumen.

CONCLUSIONS Spills of diluted bitumen will initially float regardless of the water density; thus, the first response actions are similar to those employed after spills of conventional crude oil. However, as the diluted bitumen weathers, its properties change (see Chapters 2 and 3) in ways that can affect the response. The time windows during which dispersants or in situ burning can be used effectively are much shorter for spills of diluted bitumen than for spills of conventional crude oils. The strong adhesion of diluted bitumen to surfaces requires higher pressures and temperatures when using flushing techniques. Because it is already highly degraded, natural attenuation of residual diluted bitumen is less likely to be effective, which can trigger the need for more aggressive removal actions.

88 SPILLS OF DILUTED BITUMEN FROM PIPELINES

TABLE 4-2 Effectiveness of Selected Response Tactics for Conventional Crude Oils Compared to Diluted Bitumen in Seawater Light Medium Heavy Diluted Tactic Crude Crude Crude Bitumen Dispersant 50% to 90% 10% to 75% 0%a ~50% at 6 hr Effectiveness up to 72 hr up to 72 hr ~0% after 12 hr In Situ 99% at 96 hr 99% at 96 hr 90% at 96 hr 50% to 75% up Burning to 24 hr; not effective after 96 hr Removal of Washing with Washing Washing Washing with Oil Adhered low pressure, with higher with high- high-pressure, to Substrates ambient pressure, pressure, hot hot water; may temperature and higher water; may require use of temperatures require use surface washing of surface agents; possibly washing dry-ice blasting agents; dry- ice blasting Waste Lowest Moderate High Potentially Generation because of highest, if high natural benthic sediment removal removal is processes required aBased on review of laboratory dispersant effectiveness tests reported by Environment Canada in the online Oil Properties Database. SOURCE: Environment Canada31

Most spill response tactics are based on the assumption that the crude oil will float. When a significant fraction of the spilled crude oil becomes suspended in the water column or sinks to the bottom, the response becomes more complex because there are few proven techniques in the responder “tool box” for detection, containment, and recovery. Recovery of sunken crude oil often generates large amounts of water and sediments that require complex logistics for handling, separation, treatment, and proper disposal of wastes. When sunken crude oil refloats spontaneously over a protected period, it can trigger the need for aggressive removal to mitigate the threats to water intakes, the public, fish, and wildlife. All of these threats are greater for spills of diluted bitumen than for spills of commonly transported crude oils. They drive the need for more complete removal of spills in inland areas where cleanup endpoints are usually more stringent. Every spill presents a unique combination of materials and conditions. Better documentation of the behavior of diluted bitumen when spilled and of effective recovery methods is needed, so that the response community can benefit from these experiences.

Comparing Properties Affecting Transport, Fate, Effects, and Response

The preceding chapters have outlined differences between commonly transported crude oils and diluted bitumen in terms of compositions, properties, and likely fates in the environment. Using a potential spill from a transmission pipeline as a framework, the comparison can be extended to consider how those differences relate to the responses that would be necessary. Three potential products of a spill via transmission pipeline that are analyzed include (i) a spill of crude oil (commonly light or medium crude), (ii) a spill of diluted bitumen, and (iii) the residue produced by weathering of diluted bitumen, a few days after a spill. The third point is important because, relative to a light or medium crude oil, diluted bitumen is likely to produce a heavy residue both more promptly and in greater quantity.

POTENTIAL OUTCOMES AND LEVEL OF CONCERN Three sets of hazards can be identified. The first relates to the transport or movement of spilled products in the environment. The second relates to the fates of those products (sinking, evaporation, persistence, etc.). The third relates to the effects of those products (impaired water quality, toxicity, air pollution, etc.). For each set of hazards, drawing on information summarized in Chapters 2 and 3, key properties were identified that could be related to the level of concern regarding that hazard. Finally, the levels of concern associated with each of the three products of interest were compared. Overall, this chapter represents an effort to

90 SPILLS OF DILUTED BITUMEN FROM PIPELINES

distill the accumulated information, doing so in ways that are relevant to the regulatory framework.

TRANSPORT Environmental hazards related to the release of diluted bitumen, and its eventual conversion to weathered diluted bitumen, are summarized and compared with other commonly transported crude oils in Figure 5-1. In Figure 5-1 the first column lists properties affecting transport, for example, adhesion and solubility. The second column graphically summarizes how commonly transported crude oils (CTC), diluted bitumen (D), and the residue resulting from weathering of diluted bitumen (WD) compare in terms of that property. The second column in the figure shows, for example, the density of commonly transported crude oils is lower than that of diluted bitumen and that the density of weathered diluted bitumen is considerably higher. The third column lists potential outcomes related to each property of interest. For example, density is an important factor in determining whether a product will submerge and move, for example, in a river or stream, in suspension, or as part of the bed load. Finally, the fourth and fifth columns tabulate the level of concern relative to that associated with a commonly transported crude oil. For example, the densities of commonly transported crudes and of diluted bitumen are so similar that the levels of concern regarding the related potential outcomes are approximately equal (noted in Figure 5-1 as “Same”). In contrast, the higher density of weathered diluted bitumen causes an increased level of concern (“More”). Columns 4 and 5 are color-coded, indicating whether a high or low value would exacerbate the risk associated with a potential outcome. Crude oil that floats on water is transported by different mechanisms than crude oil that submerges, often sorbed onto sediments, and is transported in suspension or in the bed load of streams and rivers. The greater density of weathered bitumen results in a greater level of concern that weathered bitumen will become submerged in an aquatic environment (Chapter 3). Even in the first days of a spill, the greater adhesive properties of diluted bitumen compared to commonly transported crude oils result in a greater level of concern. This concern derives from impacts on wildlife and vegetation and from the associated public reaction, as volun- teers mobilize to rescue contaminated wildlife, for example. The greater level of concern for weathered bitumen also reflects the potential magnitude of the long-term effects of a spill that reaches a water body. Given the known composition of diluted bitumen, a much greater proportion of the material released can be expected to become denser than water and/or adhere to sediments, thereby sinking and entering

91 Diluted bitumen relative to commonly transported crude oils: considerations related to transport in the environment. to transport in the environment. oils: considerations related to commonly transported crude Diluted bitumen relative

5-1 FIGURE xylenes. ethylbenzene, toluene, benzene, BTEX: Acronyms:

92 SPILLS OF DILUTED BITUMEN FROM PIPELINES

the bed load and sediments of riverine, wetland, and coastal environments. Furthermore, once the weathered bitumen becomes incorporated into the bed load, it may be deposited some distance from the initial spill and remobilized in a future storm or flood. Thus, the benefits of being prepared to contain the diluted bitumen early during the response to a spill are substantial. Very little is known about the risks associated with a subsurface release of diluted bitumen (i.e., into groundwater or a deep water column), particularly in terms of the risks of dissolution of the light, relatively water-soluble monoaromatics such as benzene, toluene, ethyl benzene, and xylenes (BTEX) into groundwater, where loss by volatilization and microbial degradation are likely to be slow. The other properties noted in Figure 5-1, namely viscosity, solubility, and concentrations of the BTEX compounds, relate to other modes of transport and to toxicity. During the first few days after a spill, while substantial portions of the diluent are likely to remain, the diluted bitumen can be expected to be transported in a manner similar to that for commonly transported crude oil. The spill responses that are standard for crude oils can be applied effectively to diluted bitumen prior to weathering. On the other hand, once sufficient weathering has occurred, there is a lower level of concern that weathered bitumen will be transported by spreading compared to commonly transported crude oils or to diluted bitumen. Lower levels of concern also apply to other outcomes that are particularly important in spills of commonly transported crude oil, such as contamination of groundwater and release of toxic volatile compounds (e.g., BTEX).

FATE Figure 5-2 summarizes estimates of the risks associated with a variety of fates for the spilled material. Those fates, or outcomes, such as sinking, surface coating, and burn residue are listed in the third column of the figure. The effects of density and adhesion on sinking and burial are similar to those on transport noted in Figure 5-1. The viscosity of diluted bitumen, or especially that of its weathered residues, however, is high enough that the risk of penetration in a soil profile is lower than that of a commonly transported crude oil. While the diluent is retained, concentrations of the very lightest, most volatile, and most flammable hydrocarbons in a diluted bitumen (particularly if the diluent is a gas condensate) may be similar to or less than those of a light or medium crude. Accordingly, the risks associated with spills of diluted bitumen are approximately the same as those for spills of commonly transported crude oils. On the other hand, the weathered

93 Diluted bitumen relative to commonly transported crude oils: considerations related to fate in the environment. environment. the in fate to related considerations oils: crude transported commonly to relative bitumen Diluted

5-2 FIGURE

94 SPILLS OF DILUTED BITUMEN FROM PIPELINES

residues resulting from spills of diluted bitumen are less volatile and less flammable than those of commonly transported crude oils. The presence in diluted bitumen, and particularly in its weathered residues, of large quantities of resins and asphaltenes heightens the level of concern about long-term persistence in the environment. Compared to commonly transported crude oils, a larger proportion of diluted bitumen and its weathered residues have the potential to become retained in sediments or in soils. Additionally, both the greater density and adhesive properties of weathered bitumen contribute to the greater likelihood of such an unfavorable outcome. If partially weathered bitumen becomes buried or entrained in sediments it can be more difficult to delineate and remove. This outcome can lead to a protracted period of exposures to the biota. For commonly transported crude oils, there is a reasonable prospect that some portion of the oil not collected by cleanup activities will be biologically degraded in the water column or in soil and sediment. Figure 5-2 summarizes estimates of the risks associated with a variety of fates for the spilled material. Those fates, or outcomes, such as sinking, surface coating, and residue after burning, are listed in the third column of the figure. The effects of density and adhesion on sinking and burial are similar to those on transport noted in Figure 5-1. Another fate issue is the comparison of the effectiveness of oil removal by in situ burning. Studies have shown9c the diluted bitumen has a shorter time window in which burning can be effective and a lower burn efficiency, compared to commonly transported crude oils. Thus, burning could leave a larger amount of residues for removal. Overall, the level of concern for environmental persistence of a substantial portion of the released material is greater for diluted bitumen and much greater for weathered bitumen

EFFECTS Effects, or outcomes, such as impaired water quality and hazardous air pollution, are considered in Figure 5-3. Density and adhesion again heighten the risks associated with diluted bitumen and especially with its weathered residues. Concentrations of the BTEX compounds in diluents are commonly high enough that related risks associated with diluted bitumen are similar to those with commonly transported crude oils, though the near absence of those compounds in the weathered residues yields a lowered risk. For commonly transported crude oils, BTEX compounds are generally an immediate concern and this is true to a similar (or lesser) extent for diluted bitumen, which would correspond to the first phase of spill

95 Diluted bitumen relative to commonly transported crude oils: considerations related to effects in the environment. environment. the in effects to related considerations oils: crude transported commonly to relative bitumen Diluted

5-3 Acronyms: BTEX: benzene, toluene, ethylbenzene, xylenes; HMW: high molecular weight; LMW: low molecular weight. molecular low LMW: weight; molecular high HMW: xylenes; ethylbenzene, toluene, benzene, BTEX: Acronyms: FIGURE

96 SPILLS OF DILUTED BITUMEN FROM PIPELINES

response. For the weathered bitumen BTEX is much less of a concern. In contrast, the level of concern is reversed for high molecular weight compounds that may be transferred in food webs and present a concern in terms of chronic toxicity. The major concern is the distinct lack of chemical characterization, biodegradation, and targeted toxicity studies for diluted bitumen compared to other commonly transported crude oils.

RESPONSE As discussed in Chapter 4, a central goal of the response strategies for spills of crude oils is to protect human health and avoid or minimize long-term detrimental outcomes in the environment. To further address the statement of task, a comparative approach was used when considering the response to spills of diluted bitumen in relation to spills of commonly transported crude oil. In the comparisons presented in Figures 5-1 through 5-3, density and adhesiveness emerged as the most distinctive properties of diluted bitumen that could raise the level of concern for long-term environmental impairment, particularly due to the potential of some portions of a weathered bitumen to become submerged under the specific circumstances. Figure 5-4 illustrates that there are distinctive aspects of effective response techniques to spills of diluted bitumen that need to be addressed in the near term to avoid such an outcome. The behavior of the light components of any spill in the first few days of an environmental release directly influences two aspects of the response. One immediate aspect of a spill is the potential health hazards to the public and the spill responders posed by volatile organic compounds. The extent of this risk for diluted bitumen products depends upon the diluent used; if condensate is used as a diluent, the level of concern may be comparable to that associated with light crude oil. The need to take appropriate measures, such as evacuation of nearby areas and/or providing appropriate personal protective equipment to the spill responders tasked with containing the spill, provides an added layer of complexity to response. These concerns are represented as being most significant in the first 2 days of the spill response for spills of both commonly transported crude oils and diluted bitumen and become less of a concern for diluted bitumen after weathering has occurred. Another aspect of the diluent behavior that influences spill response during the first few days is the progressive loss of the diluent by evaporation, which decreases the tendency for diluted bitumen to float and spread on the surface of a receiving water body. This is noteworthy for diluted bitumen because the density of the oil being transported may be that of a medium crude oil, but this can change substantially as weathering occurs to yield a residual material with a density that approaches that of water.

97 diluted bitumen relative to commonly transported crude oils. oils. crude transported commonly to relative bitumen diluted

Response operations: Response

5-4 FIGURE

98 SPILLS OF DILUTED BITUMEN FROM PIPELINES

As a result, the initial period when diluted bitumen can be contained and recovered by established response protocols coincides with the period when the exposure risk due to volatiles influences the spill response activities. In particular, the assessment of where and how much diluted bitumen has been spilled may be held back if potentially dangerous levels of volatiles are encountered. As indicated in Figure 5-5, if containment, booming, and/or recovery of a large portion of the spilled diluted bitumen are not achieved during this initial period, a significant portion of the spilled oil may aggregate with particulate matter and become submerged. In contrast, in spills of commonly transported crude oils, the major fraction of the oil will likely continue to float for a longer period and often only a minor fraction may adhere to particulate matter and become submerged. Thus, the rapid loss of diluent inherently presents a distinct immediate challenge in responding to spills of diluted bitumen compared to spills of commonly transported crude oils. Beyond the challenges presented by the behavior of the diluent, there are other distinctive aspects of the initial spill response for diluted bitumen. One is that use of in situ burning can only be effective for diluted bitumen within the first 24 hours (Figure 5-5). Unlike the outcome for commonly transported crude oils, where up to 99% can be removed by in situ burning over a longer period of time, a much lower percentage is removed upon burning diluted bitumen with the resulting burned residue being highly recalcitrant. Similarly, relative to commonly transported crude oils, there is a much narrower window of opportunity in which chemical dispersants can be applied effectively to spills of diluted bitumen. Based on laboratory and mesocosm studies, the effectiveness of dispersants on diluted bitumen is negligible after 6-12 hours, compared to effective dispersions of 50% to 90% after up to 72 hours for light and medium crude oils. A second distinctive aspect of diluted bitumen is that more oil may sorb onto structures and vegetation due to the increased adhesion described in Chapters 2 and 3. This will present challenges for removal with conventional methods. If containment is not successful in the initial period of the response to a spill of diluted bitumen, the contrasts with spills of commonly transported crude oils are greatly amplified. As the diluted bitumen weathers, more of the oil may become submerged if the mixing energy is high and particulate matter is available. These products may either be carried downstream or be deposited in sediments when turbulence decreases. Under conditions where commonly transported crude oil may become submerged, in situ biodegradation can be considered when one is evaluating the impacts of the residual oil versus the impacts of intensive removal of the oil. However, biodegradation is less likely to be effective for a submerged and sunken weathered bitumen. The removal of this weathered

PROPERTIES AFFECTING TRANSPORT, FATE, EFFECTS, AND RESPONSE 99

FIGURE 5-5 Time scales of environmental processes affecting spills of diluted bitumen and the windows of time for various response options.

bitumen may involve extensive physical disruption and generate large quantities of waste material and contaminated water to be treated. The environmental and economic benefits of avoiding this outcome are potentially great, but these benefits can only be achieved in the initial few days of the spill response. Thus, recognizing the need for prompt action while the diluent is still present is key to an effective response to a diluted bitumen spill. Based on this comparison of timelines and potential outcomes, there are distinct aspects of effective responses to spills of diluted bitumen in comparison to effective responses to spills of commonly transported crude oils.

100 SPILLS OF DILUTED BITUMEN FROM PIPELINES

CONCLUSIONS The prospect of a release of crude oil into the environment through a pipeline failure inherently raises a number of concerns. These concerns include not only minimizing a number of possible long-term environmental impacts but also protecting the safety of responders and the public during and after the spill response. When all risks are considered systematically, there must be a greater level of concern associated with spills of diluted bitumen compared to spills of commonly transported crude oils. In the context of fate, transport, and effects, the properties of diluted bitumen and weathered diluted bitumen that consistently result in greater levels of concern involve the higher density of the bitumen. The environmental outcome that should be most vigorously avoided in a spill response is the weathering of spilled diluted bitumen into heavy, sticky, sediment-laden residue that cannot readily be recovered, which requires dredging and disposal of large quantities of contaminated sediment and water, and which will not degrade if left in the environment. This weathering process begins rapidly following a release and can change the behavior of diluted bitumen in a matter of days. At the same time, the level of concern for responders and public safety associated with toxic and potentially explosive volatiles in the diluent fraction is similar as for commonly transported crude oils as these concerns are associated with properties of the diluents used. The oil and pipeline industries and the response community have developed approaches for addressing releases of crude oil that are based on accumulated experience in responding to the diversity of spills that have occurred, as well as knowledge of the general properties of crude oil. This experience is predominantly based on spills of commonly transported crude oils that can be expected to float for some time. Given these greater levels of concern, spills of diluted bitumen should entail special immediate actions in response, for example, that the properties of diluted bitumen and weathered bitumen put such spills in a class by themselves.

Regulations Governing Spill Response Planning

This chapter addresses the fourth component of the committee’s statement of task to “make a determination as to whether the differences between the environmental properties of diluted bitumen and those of other crude oils warrant modifications to the regulations governing spill response plans, spill preparedness, or cleanup.” The first section of this chapter provides an overview of the federal framework for spill response planning, preparedness, and response. The second section examines how well this framework accounts for the unique characteristics of diluted bitumen and where improvements are warranted to improve its effectiveness. The chapter concludes that, in light of the committee’s findings regarding the differences between diluted bitumen and commonly transported crude oils, modifications to the current regulatory framework are needed to better account for the unique characteristics of diluted bitumen.

FEDERAL SPILL PLANNING AND RESPONSE FRAMEWORK In 1968, problems faced by officials responding to a large spill of oil from the tanker Torrey Canyon off the coast of England heightened awareness of the importance of effective spill planning.103 That incident and later spills spurred recognition that the U.S. needed a coordinated approach to cope with potential spills in U.S. waters. Ultimately, a comprehensive system of spill reporting, containment, and cleanup was developed and codified in the National Oil and Hazardous Substances Pollu- tion Contingency Plan, more commonly called the National Contingency

101

102 SPILLS OF DILUTED BITUMEN FROM PIPELINES

Plan (NCP). The NCP,103 issued under the authority of the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA),104 provides a multiagency federal blueprint for responding to oil spills and to releases of other hazardous substances. Congress has broadened the scope of the NCP since its inception. A major milestone was the Oil Pollution Act of 1990 (OPA 90).105 Enacted in the wake of the Exxon Valdez spill, OPA 90 expanded reporting, planning, and response requirements for oil spills, prompting extensive revisions of the NCP that were finalized in 1994. Under the NCP, the planning and response process has several elements that are implemented at the pipeline or facility level and through national, regional, and area planning mechanisms. Figure 6-1 provides an overview of these elements and how they interact. Under ideal circumstances, planning at the pipeline or facility level should not occur in isolation, but as part of an integrated system that involves multiple governmental entities and the public.

FIGURE 6-1 The relationship among the various levels of oil and hazardous materials response plans under and related to the U.S. National Contingency Plan. SOURCE: Adapted from U.S. Environmental Protection Agency106

REGULATIONS GOVERNING SPILL RESPONSE PLANNING 103

National Response Team The National Response Team (NRT)107 is responsible for broad coordination and oversight of preparedness, planning, and response across the federal government. The U.S. Environmental Protection Agency (USEPA) and the U.S. Coast Guard (USCG) serve as chair and vice chair respectively, of the NRT. All federal agencies with some involvement in planning and response are members.

Regional Response Teams There are 13 Regional Response Teams (RRTs)108 in the United States, each representing a particular geographic region (including the Carib- bean and the Pacific Basin). RRTs are composed of representatives from field offices of the federal agencies that make up the NRT, tribal nations, and representatives of state government. The major responsibilities of RRTs include response, planning, training, and coordination. Regional Contingency Plans are established by RRTs to ensure that the roles of federal and state agencies during an actual incident are clear, and also to identify resources that are available from each federal agency and state within their regions, including equipment, guidance, training, and technical expertise, for dealing with chemical releases or oil spills.

Area Contingency Planning An Area Contingency Plan (ACP)109 is a reference document prepared for the use of all agencies engaged in responding to environmental emergencies in a defined geographic area. ACPs are generally initiated by RRTs and can be developed based on geographic features and jurisdictional boundaries. Within the boundary of an ACP, subareas with unique circumstances that warrant tailored response strategies can also be defined. ACPs are designed to ensure that all responders have access to essential area-specific information and to promote interagency coordination as a means of improving the effectiveness of responses. Among other things, ACPs are a potential vehicle for identifying in advance spill scenarios that may damage areas that are environmentally sensitive or of special economic or cultural importance. ACPs can also pinpoint high-risk locations such as fixed facilities or pipelines. This can lead to developing response strategies that are effective in mitigating or preventing a substantial threat of discharge and ensuring that adequate resources (personnel, equipment, and supplies) are available for response to spills and releases with potential for serious environmental or other consequences.

104 SPILLS OF DILUTED BITUMEN FROM PIPELINES

Facility Response Plans A Facility Response Plan (FRP) demonstrates a facility’s preparedness to respond to a worst case oil discharge. Under the Clean Water Act,110 as amended by OPA 90, preparation and submission of FRPs are generally required for land-based facilities that store and use oil, and for pipelines and vessels transporting oil if spills from these entities can reasonably be expected to cause “substantial harm”111 to the environment, for example, by discharging oil into or on U.S. navigable waters. Under the original terms of Presidential Executive Order 12777 issued in 1991,112 and subsequent updates, responsibility for spill response planning is divided among four agencies—the Pipeline and Hazardous Mate- rials Safety Administration (PHMSA), the USEPA, the USCG, and the Bureau of Safety and Environmental Enforcement (BSEE):

a. PHMSA has responsibility for overseeing preparation and approval of response plans for spills from onshore pipelines. b. USEPA reviews and approves response plans for spills from non- transportation-related onshore facilities. c. USCG performs these functions for vessels and onshore marine facilities. d. BSEE in the U.S. Department of Interior oversees spill response planning for offshore facilities.

Some facilities fall under the jurisdiction of two or more federal agencies and must meet multiple requirements. For example, a complex may have a transportation-related transfer area regulated by USCG, a pipeline regulated by PHMSA, and a non-transportation-related oil storage area regulated by USEPA. Significantly, the allocation of responsibility for responding to oil spills is different from the allocation of responsibility to oversee response planning. Authority for cleanup and response actions for all onshore oil spills is shared by USEPA and USCG. USCG leads responses to spills in the coastal zones and the Great Lakes, whereas USEPA has the lead for inland oil spills. However, while overseeing planning for spills from pipelines, PHMSA plays no role in the response to such spills, for which USEPA and USCG are responsible.

Onshore Pipeline Spill Response Plans The response planning requirements applicable to operators of onshore pipelines are set forth in PHMSA’s Part 194 regulations,113 adopted in 1993 following passage of OPA 90. Under these regulations, each response plan must include procedures and a list of resources for

REGULATIONS GOVERNING SPILL RESPONSE PLANNING 105

responding to a Worst Case Discharge and to a substantial threat of such a discharge. To comply with this requirement, an operator is permitted to incorporate by reference into the response plan the appropriate procedures from its manual for operations, maintenance, and emergencies. Plans must be consistent with the NCP and the applicable ACP and must (i) determine the Worst Case Discharge for each of the operator’s response zones; (ii) provide procedures and a list of resources for responding to a Worst Case Discharge and a substantial threat of such a discharge; (iii) ensure, by contract or otherwise, the availability of equipment and other response resources sufficient to address a Worst Case Discharge within the specified time limits; (iv) identify environmentally and economically sensitive areas; (v) designate the qualified individuals responsible for implementing the plan within each response zone; (vi) describe responsibilities of the operator and of relevant agencies in the event of a discharge and in mitigating or preventing a substantial threat of a discharge; and (vii) establish procedures for testing of equipment, training, drills, and other measures to ensure that the plan will be effectively implemented in the event of a spill. Under Part 194, response plans must be updated immediately to address new or different operating conditions or information and resubmitted to PHMSA within 30 days. In addition, most plans must be revised and resubmitted every 5 years. As part of their integrity management programs under PHMSA’s Part 195 regulations,114 pipeline operators must determine if a release from a pipeline segment could impact high consequence areas (HCAs). Identification of HCAs for hazardous liquid pipelines focuses on populated areas, drinking water sources, and unusually sensitive ecological resources, defined as follows:

• Populated areas include both high-population areas (called “urbanized areas” by the U.S. Census Bureau) and other populated areas (referred to by the Census Bureau as a “designated place”). • Drinking water sources include surface water or wells where a secondary source of water supply is not available. The land area in which spilled hazardous liquid could affect a water supply is also treated as an HCA. • Unusually sensitive ecological areas include locations where criti- cally imperiled species can be found, areas where multiple examples of federally listed threatened and endangered species are found, and areas where migratory waterbirds concentrate.

The purpose of identifying HCAs under Part 195 is to focus attention on pipeline maintenance, corrosion prevention, and other safeguards of

106 SPILLS OF DILUTED BITUMEN FROM PIPELINES

pipeline integrity in areas where spills could have high impacts. Identifi- cation of HCAs is not intended to strengthen spill response plans under Part 194 although it clearly has value in designating environmentally and economically sensitive areas during preparation of response plans.

WEAKNESSES OF THE CURRENT PLANNING AND RESPONSE FRAMEWORK IN ADDRESSING SPILLS OF DILUTED BITUMEN There are a number of areas where the current regulatory framework is not effectively addressing the potential environmental impacts of spills of diluted bitumen. This section reviews these shortcomings and high- lights opportunities for the responsible agencies to improve their policies and procedures. Because spills have unpredictable consequences and each spill is unique, spill response plans are an important starting point for effective response actions but do not provide an exact step-by-step protocol for responding to any specific incident. Nonetheless, while midcourse corrections and adjustments are unavoidable after a spill occurs, a good response plan forces facilities to rigorously assess in advance where spills might occur and what response strategies and resources must be in place to maximize an effective response. A good response plan also provides a means for communication and joint decision making among all the relevant entities and affected members of the public in advance of and during response to a spill. This communication in turn enables informed and effective collaboration and timely public engagement when a spill occurs. For this reason, there is a focus not just on response plans in isolation but has examined the relationship between plans and other aspects of the planning and response process. Similarly, it has looked not only at the role of PHMSA but at the interactions of the key agencies within the federal response structure.

Adequacy Review versus Checklist Approach Roughly 400 pipeline response plans have been approved by PHMSA. The agency presently has approximately five members of its staff engaged in plan review, a more than doubling of its previous level. PHMSA’s review of response plans is conducted in its Washington, DC, headquarters, with little or no involvement by its field staff. USEPA and USCG, by contrast, generally review plans for facilities within their jurisdictions in their regional offices. This enables staff members with relevant, hands-on cleanup experience and familiarity with the environments of the region to advise their colleagues who are reviewing the plans. In some USEPA offices, On-Scene Coordinators (OSCs) actually conduct the reviews them-

REGULATIONS GOVERNING SPILL RESPONSE PLANNING 107

selves. When this occurs, the involvement of field staff familiar with the pipelines or industrial facilities and with HCAs within a region makes it more likely that the plan will identify and anticipate the greatest threats resulting from a spill. At PHMSA, the review of plans is focused on completeness, using the Part 194 regulations as a checklist to ensure that all necessary components are present. Assuming the plan is complete, PHMSA’s long-standing position is that it is legally obligated to approve the plan, and that it has no discretion to evaluate its likely adequacy and effectiveness or to recommend improvements. By contrast, USEPA and USCG review plans in two stages, the first focusing on completeness and the second on adequacy. As a result, reviewers from these agencies would be more likely to identify elements of a plan that may not be adequate when implemented for particular types of oils or in specific areas, and to request improvements. Response plans submitted to PHMSA vary considerably in length and level of detail, based on the philosophy and goals of the particular pipeline operator. PHMSA takes no position on the length and detail of plans, providing they satisfy the minimum requirements of Part 194. These aspects of the process at PHMSA make it unlikely that plans for pipelines that may transport diluted bitumen will be reviewed to assess whether they fully reflect planning needs relative to the properties of the diluted bitumen, the resources and expertise necessary to deal with spills of diluted bitumen, and the possible impacts of these spills on environmentally sensitive areas. A review process with a greater focus on adequacy and more interaction with regional response experts would create stronger incentives for plans to anticipate and include strategies for addressing the challenges posed by spills of diluted bitumen. As noted above, unlike USEPA and USCG, PHMSA plays no role in coordinating or implementing actions taken in response to spills from pipelines. Thus, while PHMSA oversees development of plans and has some ability to enforce requirements pertaining to those plans, the measures actually necessary to address spills when they occur are outside its jurisdiction. This means that PHMSA is not in a position to bring to bear the lessons and experience from spill mitigation and response in its oversight and strengthening of planning. For example, PHMSA’s minimal presence in the response to the spill in Marshall, MI (Box 3-1), limited its opportunity to learn key lessons from USEPA OSCs and other responders who encountered unanticipated cleanup challenges presented by diluted bitumen and its weathered residues. The Part 194 regulations allow, but do not require, PHMSA to consult with USEPA and USCG during the review of response plans.113 The normal practice is not to engage these other agencies during these reviews, although the Enbridge response plans were informally shared

108 SPILLS OF DILUTED BITUMEN FROM PIPELINES

with USEPA following the Marshall, MI spill. Routine involvement of USCG and USEPA personnel in reviews by PHMSA would increase the likelihood that plans realistically anticipate pipeline and area-specific challenges, including those presented uniquely by diluted bitumen.

Strengthening the Area Contingency Planning Process Area Contingency Plans augment facility response plans by enabling responsible officials from the relevant federal agencies to prioritize the greatest threats resulting from spills and releases within the area. Addi- tionally, they facilitate development of strategies and identification of resource needs for mitigating environmental impacts resulting from these events. While USEPA and USCG are actively involved in the ACP process, participation by PHMSA is more limited, reflecting its lack of a direct role in managing response activities as well as its constrained resources. When PHMSA does participate, it is typically represented by regional staff and not by the headquarters team responsible for review and approval of plans. The ACP process provides a vehicle for focusing attention on the challenges associated with particular types of crude oil such as diluted bitumen, the ecosystems and resources at risk from spills of these products, and the adjustments in response strategies necessary to mitigate environmental impacts. However, ACPs could do more to take advantage of this opportunity. ACPs often include Geographic Response Plans that identify and prioritize sensitive areas for protection, and some of them include site-specific protection strategies. However, nearly all of these plans were developed to respond to floating oil, and many have not been updated for years.115 Because of the increased risks to water-column and benthic resources, there is a need to revise these plans to address spills of oils that have the potential to submerge or sink. Under Part 194, plans must identify environmentally sensitive areas that could be affected by a pipeline spill.113 PHMSA does not examine the completeness of this portion of a plan. The identification of such areas could point to water bodies and related ecosystems where the potential for diluted bitumen or its residues to sink and attach to sediments would be greatest and could also describe the related ecological consequences and cleanup challenges. A second concern is whether pipeline operators fully integrate HCAs (declared under Part 195 integrity management programs) with the description of sensitive areas in Part 194 plans, and whether either set of regulations is focusing attention on the specific risks posed to HCAs by different types of crude oils. Fuller collaboration between PHMSA and other agencies through the ACP process would pro-

REGULATIONS GOVERNING SPILL RESPONSE PLANNING 109

vide the input to PHMSA and pipeline operators necessary to strengthen and focus these elements of response plans.

Identifying the Type of Crude Oil Being Transported and Its Properties The Part 194 regulations require response plans to specify the type of oil transported by the pipeline.113 In practice, however, operators generally meet this requirement by a generic description such as “crude oil” and PHMSA does not request greater specificity. Thus, the response plan will not indicate whether the pipeline is handling diluted bitumen and, if so, the source of the bitumen, the nature of the diluent, and physical properties such as density and viscosity. Such information is of great and immediate value to the response team addressing a spill. In the initial days of the diluted bitumen spill in Marshall, MI, confusion about the nature of the crude spilled and its relevant properties delayed an effective response because the team did not foresee the consequences of weathering and the resulting potential for residues of the diluted bitumen to submerge. In addition to the response plan itself, the Safety Data Sheet (SDS) submitted by the pipeline operator is potentially a vehicle for identifying the type of crude oil and its properties. In conjunction with the plan and other information sources, a detailed SDS containing the pertinent information would assist responders setting near-term priorities directly following a spill of diluted bitumen. It would also assist the public in understanding the nature and consequences of the spill. The Part 194 regulations recommend but do not require that response plans include SDSs for the crude oil being transported by the pipeline section.113 As a result, the adequacy of the SDS is not a required focus of the plan approval process by PHMSA and, in some cases, the agency may never receive the operator’s SDS. SDSs found in pipeline spill response plans are typically generic and do not identify and differentiate specific crude oils and their properties. At present, however, there is no mechanism at PHMSA to require more informative SDSs. Accurate and specific information identifying the product being transported and its properties must be readily available to responders and affected members of the public at the time of a spill. A more detailed response plan and SDS, along with additional steps to provide analytical data for the crude as soon as possible after the spill occurs, can help meet this need. Relative to the PHMSA Part 194 process, the USCG plan-review framework is more focused on differentiating between types of crude oil.114 USCG requires vessel operators to classify the crudes they are

110 SPILLS OF DILUTED BITUMEN FROM PIPELINES

transporting based on gradations defined by specific gravity. Under this scheme, crude oils are assigned to one of five groups. Group II-IV oils are generally positively buoyant; Group V covers oils that are generally negatively buoyant (specific gravity equal to or greater than 1.0). The status of diluted bitumen under this scheme is ambiguous because as transported its specific gravity is below 1.0 but, after weathering, its specific gravity could approach or increase above this level. USCG is considering whether adjustments should be made to its classification system to more precisely address diluted bitumen. The present system is imperfect but provides a better mechanism for identifying the type of crude being transported than the generic descriptor used in plans submitted to PHMSA. If improved, the Coast Guard system might be adopted by PHMSA and other agencies as a uniform approach for classifying crude oils for spill response and planning purposes.

Certification of Oil Spill Removal Organizations The USCG classification system plays an important role in its system for certifying Oil Spill Removal Organizations (OSROs), who contract with pipeline or facility operators to provide response resources and expertise on their behalf in the event of a spill. USCG certifies OSROs based on a policy document called the OSRO Classification Guidelines.116 Certifica- tion is voluntary under the OSRO Classification Guidelines but OSROs often seek certification because reliance on a certified OSRO streamlines the level of detail required in response plans. The intent of the OSRO Classification Guidelines is to assess the OSRO’s capabilities based on the type of oil it expects to address during a spill. Although USCG recognizes that floating and nonfloating oils (based on API gravity) require different equipment, the currently used (2013) OSRO Classification Guidelines do not specify the different response resources necessary to respond to nonfloating oils that may become nonfloating during weathering, nor do they address diluted bitumen specifically.i The Part 194 regulations allow pipeline operators to designate OSROs certified under the USCG OSRO Classification Guidelines to carry out response activities on their behalf. In practice, however, PHMSA does not review the OSRO’s qualifications and capabilities beyond confirming that it has been certified by USCG for “Worst Case Discharge-1, Rivers and Streams.” As a result, there is no independent inquiry into whether the OSRO has the expertise and resources to handle particular types of crudes such as diluted bitumen. If PHMSA instituted a qualitative plan

i As of the finalization of this report, the USCG is currently working on a classification for potentially nonfloating oils (Group IV) to include in the next release of its OSRO Guidelines.

REGULATIONS GOVERNING SPILL RESPONSE PLANNING 111

review process instead of its present checklist approach, it would be able to confirm that the OSRO is in fact qualified for cleanups of the specific crude being transported by the pipeline. This would be helpful to ensure that, through the OSRO or otherwise, the pipeline operator has put in place response resources that are adequate to address spills of nonfloating oils such as diluted bitumen. The combination of improved OSRO Classification Guidelines, greater specificity in identifying crudes oils in PHMSA-required plans, and more review of the adequacy of these plans could lead to a closer examination of the qualifications of OSROs identified in those plans to address unique characteristics of crude oils like diluted bitumen.

Updating Response Plans The Part 194 regulations provide for updating of response plans at 5-year intervals and within 30 days where necessary to address new or different operating conditions or information. Following PHMSA’s renewed enforcement authority in their 2011 reauthorization bill and in the aftermath of the Marshall, MI spill, PHMSA reviewed all response plans between 2012 and 2014. Based on the currently used review protocol, only a handful of the more than 400 plans required substantial revisions. In hindsight, this would have been a useful opportunity to require pipelines transporting diluted bitumen to strengthen their plans. Present response plans dealing with aquatic environments focus on floating oil and do not address residues that submerge or sink. Updated plans could incorporate new technologies and methods, or faster response times, which are important for efficient response to spills of diluted bitumen.

Strengthening Drills and Exercising of Plans An important element of spill preparedness is announced and unannounced “exercising” of response plans to determine how well they work in practice and to identify vulnerabilities and opportunities for improve- ment. It does not appear that these exercises (which involve to varying degrees USEPA, USCG, and PHMSA) have systematically simulated the unique challenges presented by spills of diluted bitumen. Announced and unannounced plan exercises need to devote greater attention to scenarios involving spills of diluted bitumen in order to increase readiness and capability to respond to diluted bitumen spills.

112 SPILLS OF DILUTED BITUMEN FROM PIPELINES

CONCLUSIONS The current system for response planning, preparedness, and mitigation is geared to the properties, behavior in the environment, and response challenges of commonly transported light and medium crude oils. Thus, the focus is on preparing for spills of oil that float on the surface. Spills of diluted bitumen raise different issues because of the greater density that the product acquires as the diluent evaporates. This leads to the potential, depending on environmental conditions, for the diluted bitumen residues to aggregate with particles in the water column and submerge or sink to the bottom of the water body. Different strategies, expertise, and response capabilities are needed to effectively address this spill scenario. However, the current relevant U.S. regulations and agency practices do not capture the unique properties of diluted bitumen or encourage effective planning for spills of that product. The PHMSA Part 194 regulations are critical to the preparation of thoughtful and effective response plans but, in their current form, do not focus pipeline operators on the properties of diluted bitumen and the associated response challenges or provide the information necessary for responders to address spills of diluted bitumen. Shortcomings of the regulations include the absence of any requirement to identify the type of crude oil being transported and its properties, to describe how ecologi- cally sensitive areas might be impacted by spills of diluted bitumen, and to demonstrate that response strategies, resources, and capabilities are in place for effective responses to spills of diluted bitumen. These weaknesses are compounded by PHMSA’s reliance simply on a checklist approach to review plans, in contrast to focusing on the adequacy and effectiveness of plans, as is the case with reviews by USCG and USEPA. PHMSA also does not regularly consult with these agencies during plan reviews, despite their hands-on experience with spill response and knowledge of local conditions that could contribute to informed judgments about the adequacy of a plan. PHMSA has further been slow to require updates of plans, thus missing an opportunity for incorporating new information about the properties and environmental behavior of diluted bitumen in plans and enhancing their effectiveness. The response plan is only one element in the overall federal system of preparedness and response. Other aspects of this system also need to address for the challenges imposed by diluted bitumen and to improve readiness. For example, a greater focus on diluted bitumen in area contingency planning, with PHMSA playing a more active role in the process, may help address this challenge. In addition, a uniform nomenclature system used by all agencies would help in the differentiation of diluted bitumen from other crudes.

Recommendations

As this report demonstrates, diluted bitumen has unique properties that differentiate it from commonly transported crude oils. Because of these properties, diluted bitumen’s behavior in the environment following a spill is different from that of the light and medium crudes typically addressed in spill response planning, preparedness, and response. Of most significance are the physical and chemical changes that diluted bitumen undergoes as a result of weathering. When the diluent component volatizes, the remaining bitumen becomes denser and, depending on circumstances, may aggregate with particles in the water column and remain in suspension or sink to the bottom of a water body. The submergence of persistent residues of diluted bitumen in aquatic environments, as was seen in the Marshall, MI spill, and the potential for long-term deposition in sediments and banks and remobilization in the water column present environmental concerns and cleanup challenges not presented by commonly transported crude oils. These challenges necessitate different response strategies, including immediate efforts to recover spilled diluted bitumen before significant weathering occurs and effective methods to identify, contain, and recover suspended and sunken oil. The existing framework for pipeline spill planning, preparedness, and response is generally designed to address floating oil and not residues that mix throughout in the water column, aggregate with particles, and sink to the bottom of aquatic environments. As a result, the pipeline operators and the agencies responsible for spill planning and response may not be adequately prepared for diluted bitumen spills and may

113

114 SPILLS OF DILUTED BITUMEN FROM PIPELINES

lack the tools for effective cleanup. This is in part a shortcoming of the Pipeline and Hazardous Materials Safety Administration (PHMSA) Part 194 regulations and in part a shortcoming of the broader interagency contingency planning and response system. A more comprehensive and focused approach to diluted bitumen across the federal oil spill response family is necessary to improve preparedness for spills of diluted bitumen and to enable more effective cleanup and mitigation measures when these spills occur. The recommendations presented in this chapter are designed to achieve this goal.

a. Require the plan to identify all of the transported crude oils using industry-standard names, such as Cold Lake Blend, and to include Safety Data Sheets for each of the named crude oils. Both the plan and the associated Safety Data Sheets should include spill-relevant properties and considerations. b. Require that plans adequately describe the areas most sensitive to the effects of a diluted bitumen spill, including the water bodies potentially at risk. c. Require that plans describe in sufficient detail response activities and resources to mitigate the impacts of spills of diluted bitumen, including capabilities for detection, containment, and recovery of submerged and sunken oil. d. Require that PHMSA consult with the U.S Environmental Protec- tion Agency (USEPA) and/or the U.S. Coast Guard (USCG) to obtain their input on whether response plans are adequate for spills of diluted bitumen. e. Require that PHMSA conduct reviews of both the completeness and the adequacy of spill response plans for pipelines carrying diluted bitumen. f. Require operators to provide to PHMSA, and to make publicly available on their websites, annual reports that indicate the volumes of diluted bitumen, light, medium, heavy, and any other crude oils carried by individual pipelines and the pipeline sections transporting them.

RECOMMENDATIONS 115

g. Require that plans specify procedures by which the pipeline operator will (i) identify the source and industry-standard name of any spilled diluted bitumen to a designated Federal On-Scene Coordinator, or equivalent state official, within 6 hours after a spill has been detected and if requested, (ii) provide a 1-L sample drawn from the batch of oil spilled within 24 hours of the spill, together with specific compositional information on the diluent.

These recommended changes in the Part 194 regulations would take important steps to ensure that spill response plans recognize the differences between diluted bitumen and commonly transported oils and that pipeline operators and agency responders have the special expertise, capabilities, and cleanup tools necessary for effective responses to spills of diluted bitumen. Critical to effective planning and response is determining whether a pipeline segment is expected to transport diluted bitumen and, if so, identifying its cleanup-relevant properties (e.g., density, adhesion, viscosity, and biodegradability) before and after weathering. This should be accomplished by requiring spill response plans to provide industry-standard namesi for the crude being transported, accompanied by a relevant properties description. This information would also be required in the Safety Data Sheet (SDS), which would be submitted to PHMSA for review as part of the response plan. The SDS is an important vehicle for communicating response information to affected persons who may not receive the formal response plan, such as local communities, fire departments, and medical personnel. A clear and specific description of the crude being spilled is critical to perform this informational function. The response plans should also demonstrate that the operator under- stands the unique properties and potential environmental impacts of diluted bitumen and is prepared to implement response strategies that address its challenges. This should take the form of enhanced plan sections describing in detail (i) the areas most sensitive to the effects of a diluted bitumen spill, including the water bodies potentially at risk, and (ii) response strategies and resources necessary to mitigate the impacts of spills of diluted bitumen, including capabilities for detection, containment, and recovery of submerged and sunken oil. The regulations should provide that PHMSA will review these plan elements not simply to determine whether the plan is complete (the current “checklist” approach) but also to examine whether the plan is adequate and effective in anticipating and preparing for a diluted bitu-

i The industry-standard names should be agreed by both the industry and the relevant government regulators and responders.

116 SPILLS OF DILUTED BITUMEN FROM PIPELINES

men spill. This approach would be a significant change in how PHMSA interprets its responsibilities under OPA 90. Thus, PHMSA would need to reexamine its legal authority to determine whether it has discretion to conduct an adequacy review of plans submitted by pipeline operators. The PHMSA regulations should also provide that, in conducting such an adequacy review, PHMSA will routinely share the plan with USEPA and USCG and obtain their feedback. This consultation will take advantage of the on-the-ground spill response expertise and experience of these agencies (which PHMSA lacks because it has no direct role in the response process), thus bringing it to bear in examining the adequacy of plans in preparing for diluted bitumen spills. Requiring submission of annual reports by pipeline operators documenting the volumes of various crudes being transported and the pipeline routes and sections carrying them will fill a fundamental gap in publicly available information and enable responders and communities to be more knowledgeable about the types of spills that might impact their areas and the potential consequences. This will facilitate better planning at the area, regional, and national levels and encourage more informed public engagement. Finally, while the identification of diluted bitumen and other crudes by industry-standard names should be sufficient in the response plan and SDS, more detailed compositional information will likely be needed when a spill occurs, both to guide the emergency response as well as over the longer term to support forensic chemistry evaluations and site remediation. Given that compositions of oils carried in pipelines typically vary over time, and in the case of diluted bitumen the diluent may be particularly variable, there should be an expedited procedure for characterizing the specific composition of spilled crude oil after a spill. Crude oils transported by pipelines are routinely sampled and the samples are temporarily archived until the shipments are delivered to their final destinations in case questions about the quality of the delivered oil arise at refineries. The regulations should require that, in the event of a spill, sample(s) of crude oil from the archived set representing the spilled oil will be made available within 24 hours to the responsible government agency if requested. The 24-hour time period includes transit time from the origin of the line, which may be thousands of miles away. A minimum volume of 1-L should be sufficient for chemical analysis. Pipeline operators already maintain a custodial sample and the processes are already in place for the collection and storage. The chemical analysis from the crude oil sample will give the composition of the crude oil being transported, which will benefit critical early decisions responding to spills of crude oil. In addition, the operator should be required to inform the relevant authorities of the source and industry-standard name of any spilled diluted bitumen within 6 hours of

RECOMMENDATIONS 117

a spill being detected. The 6-hour time period was selected as a reasonable balance between the usefulness of the information to the response efforts and the ability of an operator to obtain the information. Because of the potential for submergence within a relatively short window, it is critical that this information be provided to the response community with enough time to act. On the other hand, because transmission pipelines operate in batch shipments, it will take some time for the operator to conclusively identify the specific batch that was spilled.

Spills of diluted bitumen products where the crude oil submerges in the water column or sinks to the bottom are particularly difficult for responders. Most of the effective response methods are based on the premise that the spilled oil floats. Proven methods are needed so that responders have effective means by which to determine where the crude oil is, track its movement over time, and effectively recover it. Detection of diluted bitumen spills on the bottom may pose different challenges than for conventional nonfloating crude oils because the diluted bitumen can occur as oil-particle aggregates that require different detection methods than those used to detect bulk crude oil on the bottom. In situations where water is moving, there are no proven techniques for containment of suspended or sunken crude oil to prevent remobilization and spreading prior to recovery. Various techniques have been proposed but few have been shown to be effective. Once the crude oil has sunk to the bottom, recovery methods are selected based on the environmental setting, amount and distribution of the crude oil, and cleanup endpoints. Better technologies are needed to minimize water and sediment removal and improve the separation and treatment of oil, water, and sediment.

Recommendation 3: USEPA, USCG, and state and local agencies should adopt the use of industry-standard names for crude oils, including diluted bitumen, in their oversight of oil spill response planning.

A common nomenclature for identifying diluted bitumen and other crudes should be used across the federal response family and by state and local responders to improve communication and ensure that respond-

118 SPILLS OF DILUTED BITUMEN FROM PIPELINES

ers have accurate and detailed information about the composition of the crude oil being spilled. This is particularly important for diluted bitumen, which has unique and potentially variable properties not fully recognized by the federal response community that need to be clearly identified at the time of a spill. In addition, a common nomenclature system will have benefits for spills of other crude oils as well. This system of product names, which would be incorporated in response plans under Recommendation 1, should be developed through a collaborative process among PHMSA, USEPA, USCG, and state and local agencies and then be adopted by all agencies.

USCG categorizes oils into groups defined by their specific gravity, with Group IV oils defined as having a specific gravity equal to or greater than 0.95 and less than 1.0 and Group V oils as having a specific gravity equal to or greater than 1.0. These groups are important because they are used in Oil Spill Removal Organization classifications and setting of guidelines for response capabilities. Under the current approach, diluted bitumen oil products would be classified as Group IV because the fresh oil has a specific gravity less than 1.0. However, once spilled, weathering and loss of the diluent will result in a bitumen with a specific gravity that could approach or become greater than 1.0. The 2013 OSRO Guidelines lack guidance for nonfloating oils in terms of response and transportation and the USCG is in the process of creating a nonfloating oil classification system. Some oils, such as diluted bitumen, have unique characteristics that may cause an evolution from a Group IV oil to a potentially submerged oil (even if not surpassing a specific gravity of 1.0, as in the case of oil-particle aggregates formed in fresh water). A revised USCG classification system can better address these types of oils and should be incorporated in USCG, USEPA, and PHMSA regulations and strategies.

Advanced Predictive Modeling Recommendation 5: The National Oceanic and Atmospheric Administration (NOAA) should lead an effort to acquire all

RECOMMENDATIONS 119

data that are relevant to advanced predictive modeling for spills of diluted bitumen being transported by pipeline.

A combination of information from oil property and behavior testing, as well as from oil spill models, can improve the response to an oil spill. Accurate spill modeling is now a very important part of both contingency planning and actual spill response. Spill models combine the latest information on oil fate and behavior with hydrodynamic modeling to predict where the oil will go and how much it will change before it arrives. The movement of crude oils is predicted by using the water current and wind speeds along the predicted water path. In addition to predicting the movement, these models can estimate the amount of evaporation, the possibility of emulsification, the amount of dissolution and subsequent movement of the dissolved component, the amount and fate of the portion that is naturally dispersed, and the amount of oil deposited and remaining on shorelines. In the United States, much of the behavior and weathering information for spills in water is provided by NOAA via their Automated Data Inquiry for Oil Spills (ADIOS) oil weathering model. The NOAA Office of Response and Restoration, which developed and maintains ADIOS, has a mandate to support oil spill response, whereas other agencies do not have the base funding to do the same kind of work. This oil spill response tool models how different types of oil weather in the marine environment. The success of NOAA’s ADIOS is due to their extensive database of more than a thousand different crude oils and refined products. However, the NOAA databases would benefit from additional information on all diluted bitumen products. Data on diluted bitumen that are relevant to the NOAA databases are typically found in bulk form and are currently available through Environment Canada databases and other resources. NOAA should lead an effort to fill this gap.

Improved Coordination Recommendation 6: USEPA, USCG, PHMSA, and state and local agencies should increase coordination and share lessons learned to improve the area contingency planning process and to strengthen preparedness for spills of diluted bitumen. These agencies should jointly conduct announced and unannounced exercises for spills of diluted bitumen.

Improved coordination and communication among the many agencies with spill responsibilities would be valuable in creating stronger awareness of the response challenges posed by diluted bitumen, its

120 SPILLS OF DILUTED BITUMEN FROM PIPELINES

unique properties, and the most effective strategies and resources for addressing diluted bitumen spills. There are several vehicles for enhanced collaboration but the area contingency planning process is particularly important because it encourages a focus on the pipeline routes, types of crude being transported, and sensitive water bodies and ecosystems potentially impacted by spills in a defined area. Through the ACP process, plan reviewers and responders can share information gleaned from diluted bitumen response actions, better integrating these two aspects of the response effort. Strong PHMSA participation in area and regional planning, which generally has not occurred up to now, is essential for this collaboration to succeed. The committee’s understanding is that plan exercises, which are critical to evaluate plan adequacy and responder preparedness, have devoted little if any attention to diluted bitumen spills. USEPA, USCG, and PHMSA together with state and local partners should work together to ensure that announced and unannounced exercises include diluted bitumen spill scenarios so that agencies and pipeline operators can obtain feedback and experience regarding the adequacy of plans for these spills and improve response capabilities.

As highlighted by Table 2-6, diluted bitumen, and particularly its weathered residues, are highly adhesive. The amount of diluted bitumen residue that will adhere to a clean needle is more than 100-fold greater than the amount that adheres when a clean needle is immersed in the residue of a weathered light, crude oil. Reduction of uncertainties about the chemical cause of diluted bitumen’s avid adhesion, and development of a method that quantifies adhesion precisely, would be useful in tailor- ing optimal cleanup procedures. A limited study of the mechanism of adhesion, perhaps by analysis of the materials that adhere to surfaces (as opposed to analyses of the whole oils), should be used to inform the development of a method of adhesion measurement that will provide information not otherwise available to the spill response community and which can be well standardized.

RECOMMENDATIONS 121

PRIORITY RESEARCH AREAS As outlined throughout this report, many differences between diluted bitumen and commonly transported crudes are well established. While there are clearly enough data and information to support the findings and recommendations outlined herein, a more comprehensive understanding of diluted bitumen and its properties, environmental, and human health effects, would improve spill response in the future. There still remain areas of uncertainty that hamper effective spill planning and response. These uncertainties span a range of issues, including diluted bitumen’s behavior in the environment under different conditions, its detection when submerged, and the best response strategies for mitigating the impacts of submerged oil. Because of their importance to spill planners and responders, a concerted effort to fill these knowledge gaps with additional research is essential. Further research is needed to better understand the behavior of diluted bitumen in the environment, including consideration of the diversity of environmental settings in which spills could occur, the chemical constituents and their toxicological effects, as well as to develop more effective methods for detection and recovery of spilled diluted bitumen, particularly after it becomes submerged or sunken in water bodies. Some of these research needs have been articulated in the past, including a report from the National Coastal Research Council (on behalf of USCG) entitled “Spills of Nonfloating Oils: Risk and Response,” which included specific recommendations for detection, monitoring, modeling, and recovery of submerged oil; however this report focuses mainly on marine environments and did not consider the particular characteristics of diluted bitumen. The recent and projected future escalation of diluted bitumen transport in pipelines (and by other modes) has increased the possibility of release, and therefore research on the effects of diluted bitumen spills on the environment has become an ever more pressing need. Major questions targeted for research include the following: Transport and fate in the environment. How will the various combinations of bitumen and diluent change (weather) upon release, and can we predict when submergence is likely under a variety of conditions (turbulence, suspended matter, contact with benthic surfaces, plants, etc.)? How does biodegradation influence toxicity, for example, by “releasing” previously bound chemical constituents or by producing more toxic biodegradation products? Ecological and human health risks of weathered diluted bitumen. While much is known about the toxicity of hydrocarbon components in the commonly used diluents (e.g., benzene), there has been little study of the potential toxicity of bitumen to people or wildlife. For submerged and sunken oil in particular, there may be routes of exposure that have not

122 SPILLS OF DILUTED BITUMEN FROM PIPELINES

been considered sufficiently, such as sensitive egg and larval life stages of fishes, and exposures may occur over protracted periods given the apparent resistance of bitumen components to biodegradation. Other indirect effects of the oil may relate to enhanced bioavailability of co-occurring pollutants and altering properties of the impacted ecosystem (i.e. redox status, dissolved oxygen levels and pH). Detection and quantification of submerged and sunken oil. The Marshall, MI oil spill demonstrated that current options for sunken oil quantification are either unreliable (e.g., total petroleum hydrocarbons) or very expensive and time consuming (chemical fingerprinting), and consequently “poling” to disturb the sediments and observe the resultant appearance of floating globules and sheen became the primary means of mapping sunken oil (Box 3-1). Better measurement techniques should be a research priority. Techniques to intercept and recover submerged oil on the move. Submerged oil moving downstream in rivers or following wind- or tidally driven currents could be intercepted in theory, but in reality no techniques are known to be efficacious to capture oil beneath the water surface. Research should strive to develop options for the diversity of environmental settings in which oil can be spilled. Alternatives to dredging to recover sunken oil. Dredging is costly and environmentally destructive, producing voluminous waste that often must be landfilled, and therefore alternatives should be sought. Agitation and collection of resultant floating oil was conducted in the Kalamazoo River, but its efficacy in a particular spill needs investigation before being deployed again. Other alternatives should be studied as well. These research priorities are targeted broadly to the research community, but a specific mention is needed regarding the role of local and regional scientists in spill response.117 Improved access and collaboration with these scientists would help advance the scientific understanding of how oil behaves in the environment, particularly for emerging issues such as spills of diluted bitumen. Scientists from outside the formal response framework are typically not included in the formal oil spill response activities and, as a result, are often barred from site access by response officials, and their requests for source materials are denied. This situation hinders fundamental research on spill events—research that should ultimately benefit spill planning—and may also provide immediate benefit to response officials.

FINAL THOUGHTS Diluted bitumen has received extensive publicity in the past 5 years and will continue to be of interest due to production from Canadian oil

RECOMMENDATIONS 123

sands. As more diluted bitumen is transported, the need for efficient spill response planning, preparedness, and cleanup will be increasingly important. It is difficult to be completely prepared for a potential spill of diluted bitumen because our experience is limited to just a few significant spills, the products involved can vary in chemical composition, and the environmental settings where spills could occur are extremely diverse. Nonetheless, the recommendations put forward are designed to improve current oil spill planning and response to reduce negative impacts on human health and the environment.

References

1. Pipeline Safety, Regulatory Certainty, and Job Creation Act. Public Law 112-90, 2011. 2. National Research Council, Effects of Diluted Bitumen on Crude Oil Transmission Pipelines. The National Academies Press: Washington, DC, 2013. 3. Transportation and Housing and Urban Development, and Related Agencies Appro- priations Bill. Senate Report 113-45, 2014. 4. Levine, S.; Taylor, G.; Arthur, D.; Tolleth, M. Understanding Crude Oil and Product Mar- kets; American Petroleum Institute: Washington, DC, 2014. 5. (a) Canadian Association of Petroleum Producers Crude Oil Forecast, Markets & Transpor- tation; Publication Number 2015-0007; 2015; (b) Crosby, S.; Fay, R.; Groark, C.; Kani, A.; Smith, J. R.; Sullivan, T.; Pavia, R., Transporting Alberta Oil Sands Products: Defining the Issues and Assessing the Risks. U.S. Dept. of Commerce, NOAA Technical Memoran- dum NOS OR&R 43. Emergency Response Division, NOAA 2013, 153. 6. POLARIS Applied Sciences Inc. A Comparison of the Properties of Diluted Bitumen Crudes with other Oils. http://crrc.unh.edu/sites/crrc.unh.edu/files/comparison_ bitumen_other_oils_polaris_2014.pdf. 7. (a) Dupuis, A.; Ucan-Marin, F., A Literature Review on the Aquatic Toxicology of Pe- troleum Oil: An Overview of Oil Properties and Effects to Aquatic Biota. DFO Can. Sci. Advis. Sec. Res. Doc. 2015, 007; (b) Gosselin, P.; Hrudey, S. E.; Naeth, M. A.; Plourde, A.; Therrien, R.; Van Der Kraak, G.; Xu, Z., Environmental and Health Impacts of Canada’s Oil Sands Industry. The Royal Society of Canada: Ottawa, Canada, 2010. 8. Environment Canada; Fisheries and Oceans Canada; Natural Resources Canada, Prop- erties, Composition and Marine Spill Behavior, Fate and Transport of Two Diluted Bitumen Products from the Canadian Oil Sands. Environment Canada: Ottawa, Canada, 2013.

125

126 SPILLS OF DILUTED BITUMEN FROM PIPELINES

9. (a) Fitzpatrick, F. A.; Boufadel, M. C.; Johnson, R.; Lee, K.; Graan, T. P.; Bejarano, A. C.; Zhu, Z.; Waterman, D.; Capone, D. M.; Hayter, E.; Hamilton, S. K.; Dekker, T.; Garcia, M. H.; Hassan, J. S. Oil-Particle Interactions and Submergence from Crude Oil Spills in Marine and Freshwater Environments - Review of the Science and Future Science Needs; Open-File Report 2015-1076; U.S. Geological Survey: Reston, VA, 2015; (b) King, T. L.; Robinson, B.; Boufadel, M.; Lee, K., Flume Tank Studies to Elucidate the Fate and Be- havior of Diluted Bitumen Spilled at Sea. Mar. Pollut. Bull. 2014, 83 (1), 32-37; (c) Witt O’Brien’s; Polaris Applied Sciences; Western Canada Marine Response Corporation A Study of Fate and Behavior of Diluted Bitumen Oils on Marine Waters: Dilbit Experiments - Gainford, Alberta; Trans Mountain Pipeline ULC: 2013; p 163. 10. (a) King, T. L.; Robinson, B.; McIntyre, C.; Toole, P.; Ryan, S.; Saleh, F.; Boufadel, M.; Lee, K., Fate of Surface Spills of Cold Lake Blend Diluted Bitumen Treated with Dispersant and Mineral Fines in a Wave Tank. Environ. Eng. Sci. 2015, 32 (3), 250-261; (b) SL Ross Environmental Research Limited Meso-scale Weathering of Cold Lake Bitumen/Condensate Blend; Ottawa, Canada, 2012. 11. Enbridge Energy Partners LP Form 10-Q for the Period Ending June 30, 2014. http:// www.sec.gov/Archives/edgar/data/880285/000119312514290178/d765165d10q1.pdf (accessed 10/7/2015). 12. Dollhopf, R. J.; Fitzpatrick, F. A.; Kimble, J. W.; Capone, D. M.; Graan, T. P.; Zelt, R. B.; Johnson, R., Response to Heavy, Non-Floating Oil Spilled in a Great Lakes River Envi- ronment: A Multiple-Lines-Of-Evidence Approach for Submerged Oil Assessment and Recovery. In Proceedings of the International Oil Spill Conference, Savannah, GA, 2014; pp 434-448. 13. U.S. Energy Information Administration U.S. Imports by Country of Origin - All Coun- tries. http://www.eia.gov/dnav/pet/pet_move_impcus_d_nus_Z00_mbbl_a.htm (accessed 06/11/2015). 14. U.S. Energy Information Administration Refinery Receipts of Crude Oil by Method of Transportation. http://www.eia.gov/dnav/pet/pet_pnp_caprec_dcu_nus_a.htm (accessed 06/11/2015). 15. U.S. Energy Information Administration U.S. Crude Oil Production to 2025: Updated Projection of Crude Types. http://www.eia.gov/analysis/petroleum/crudetypes/ pdf/crudetypes.pdf (accessed 10/7/2015). 16. U.S. Energy Information Administration California Field Production of Crude Oil. http://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=MCRFPCA1&f=M (accessed 06/11/2015). 17. Sheridan, M. California Crude Oil Production and Imports; CEC-600-2006-006; California Energy Commission: Sacramento, CA, 2006. 18. Seelke, C. R.; Villarreal, M. A.; Ratner, M.; Brown, P. Mexico’s Oil and Gas Sector: Back- ground, Reform Efforts, and Implications for the United States; R43313; Congressional Research Service: 2015. 19. National Energy Board Estimated Canadian Crude Oil Exports by Type and Destina- tion. http://www.neb-one.gc.ca/nrg/sttstc/crdlndptrlmprdct/stt/stmtdcndncrdlx- prttpdstn-eng.html (accessed 06/11/2015). 20. National Energy Board Canadian Crude Oil Exports - By Export Transportation System Summary - 5 year trend. http://www.neb-one.gc.ca/nrg/sttstc/crdlndptrlmprdct/ stt/cndncrdlxprttrnsprttnsstm5yr/2013/cndncrdlxprttrnsprttnsstm5yr2013-eng.html (accessed 06/11/2015).

REFERENCES 127

21. Adams, J.; Larter, S.; Bennett, B.; Huang, H.; Westrich, J.; C. van Kruisdijk, The Dynamic Interplay of Oil Mixing, Charge Timing, and Biodegradation in Forming The Alberta Oil Sands: Insights from Geologic Modeling and Biogeochemistry. In Heavy-Oil and Oil- Sand Petroleum Systems in Alberta and Beyond, Hein, F. J.; Leckie, D.; Larter, S.; Suter, J. R., Eds. American Association of Petroleum Geologists, Canadian Heavy Oil Association, and American Association of Petroleum Geologists Energy Minerals Division: Tulsa, OK, 2013; pp 23-102. 22. Hollebone, B., The Oil Properties Data Appendix. In Handbook of Oil Spill Science and Technology, Fingas, M., Ed. John Wiley and Sons Inc.: NY, 2015; pp 577-681. 23. Swarthout, R. F.; Nelson, R. K.; Reddy, C. M.; Hall, C. G.; Boufadel, M.; Valentine, D.; Arey, J. S.; Gros, J., Physical and Chemical Characterization of Canadian Dilbit and Related Products. 2015. 24. Mullins, O. C., The Asphaltenes. Annu. Rev. Anal. Chem. 2011, 4, 393-418. 25. McKenna, A. M.; Donald, L. J.; Fitzsimmons, J. E.; Juyal, P.; Spicer, V.; Standing, K. G.; Marshall, A. G.; Rodgers, R. P., Heavy Petroleum Composition. 3. Asphaltene Aggrega- tion. Energy Fuels 2013, 27 (3), 1246-1256. 26. Yang, C.; Wang, Z.; Hollebone, B. P.; Brown, C. E.; Yang, Z.; Landriault, M., Chromato- graphic Fingerprinting Analysis of Crude Oils and Petroleum Products. In Handbook of Oil Spill Science and Technology, John Wiley & Sons, Inc: 2014; pp 93-163. 27. Wang, Z.; Hollebone, B. P.; Fingas, M.; Fieldhouse, B.; Sigouin, L.; Landriault, M.; Smith, P.; Noonan, J.; Thouin, G. Characteristics of Spilled Oils, Fuels, and Petroleum Products: 1. Composition and Properties of Selected Oils; Environment Canada: 2003. 28. Hollebone, B; Brown, C., Cold Lake Bitumen PAH Analysis Results. Environment Canada ETC Spills Technology Databases, Oil Properties Database. 29. National Research Council, Oil in the Sea III: Inputs, Fates, and Effects. The National Academies Press: Washington, DC, 2003. 30. Jokuty, P.; Whiticar, S.; Fingas, M.; Meyer, E.; Knobel, C., Hydrocarbon Groups and Their Relationship to Oil Properties and Behavior. In Proceedings of the 18th Arctic and Marine Oilspill Program (AMOP) Technical Seminar, Environment Canada: Ottawa, Canada, 1995; pp 1-19. 31. Environment Canada ETC Spills Technology Databases, Oil Properties Database. http://www.etc-cte.ec.gc.ca/databases/oilproperties/ (accessed 10/7/2015). 32. Aeppli, C.; Carmichael, C. A.; Nelson, R. K.; Lemkau, K. L.; Graham, W. M.; Redmond, M. C.; Valentine, D. L.; Reddy, C. M., Oil Weathering after the Deepwater Horizon Disaster Led to the Formation of Oxygenated Residues. Environ. Sci. Technol. 2012, 46 (16), 8799-8807. 33. (a) Garrett, R. M.; Pickering, I. J.; Haith, C. E.; Prince, R. C., Photooxidation of Crude Oils. Environ. Sci. Technol. 1998, 32 (23), 3719–3723; (b) Maki, H.; Sasaki, T.; Harayama, S., Photo-oxidation of biodegraded crude oil and toxicity of the photo-oxidized products. Chemosphere 2001, 44 (5), 1145-1151; (c) Prince, R. C.; Garrett, R. M.; Bare, R. E.; Grossman, M. J.; Townsend, T.; Suflita, J. M.; Lee, K.; Owens, E. H.; Sergy, G. A.; Brad- dock, J. F., The roles of photooxidation and biodegradation in long-term weathering of crude and heavy fuel oils. Spill Sci. Technol. Bull. 2003, 8 (2), 145-156; (d) Radović, J. R.; Aeppli, C.; Nelson, R. K.; Jimenez, N.; Reddy, C. M.; Bayona, J. M.; Albaigés, J., Assess- ment of Photochemical Processes in Marine Oil Spill Fingerprinting. Mar. Pollut. Bull. 2014, 79 (1–2), 268-277. 34. D’Auria, M.; Emanuele, L.; Racioppi, R.; Velluzzi, V., Photochemical Degradation of Crude Oil: Comparison Between Direct Irradiation, Photocatalysis, and Photocatalysis on Zeolite. J. Hazard. Mater. 2009, 164 (1), 32-38. 35. Chapelle, F., Ground-water microbiology and geochemistry. John Wiley & Sons: 2001.

128 SPILLS OF DILUTED BITUMEN FROM PIPELINES

36. Barron, M. G.; Carls, M. G.; Short, J. W.; Rice, S. D., Photoenhanced Toxicity of Aqueous Phase and Chemically Dispersed Weathered Alaska North Slope Crude Oil to Pacific Herring Eggs and Larvae. Environ. Toxicol. Chem. 2003, 22 (3), 650-660. 37. (a) Prince, R. C., Petroleum Spill Bioremediation in Marine Environments. Crit. Rev. Microbiol. 1993, 19 (4), 217-242; (b) Transportation Safety Board of Canada Crude Oil Pipeline - Third-Party Damage; Pipeline Investigation Report P07H0040; Trans Mountain Pipeline L.P.: Burnaby, Canada, 2007. 38. (a) Boufadel, M. C.; Sharifi, Y.; Van Aken, B.; Wrenn, B. A.; Lee, K., Nutrient and Oxy- gen Concentrations within the Sediments of an Alaskan Beach Polluted with the Exxon Valdez Oil Spill. Environ. Sci. Technol. 2010, 44 (19), 7418-7424; (b) Li, H.; Boufadel, M. C., Long-Term Persistence of Oil from the Exxon Valdez Spill in Two-Layer Beaches. Nat. Geosci. 2010, 3 (2), 96-99. 39. (a) Geng, X.; Boufadel, M. C.; Personna, Y. R.; Lee, K.; Tsao, D.; Demicco, E. D., BIOB: a mathematical model for the biodegradation of low solubility hydrocarbons. Mar. Pollut. Bull. 2014, 83 (1), 138-147; (b) Torlapati, J.; Boufadel, M. C., Evaluation of the Biodegradation of Alaska North Slope Oil in Microcosms Using the Biodegradation Model BIOB. Front. Microbiol. 2014, 5, 212. 40. Wang, Z.; Fingas, M., Separation and Characterization of Petroleum Hydrocarbons and Surfactant in Orimulsion Dispersion Samples. Environ. Sci. Technol. 1996, 30 (11), 3351-3361. 41. U.S. Environmental Protection Agency Environmental Response Team’s Final Bench Scale/ Screening Level Oil Biodegradation Study; Report Number 1597; 2013. 42. (a) Hamoda, M. F.; Hamam, S. E. M.; Shaban, H. I., Volatilization of Crude Oil from Saline Water. Oil Chem. Pollut. 1989, 5 (5), 321-331; (b) Stiver, W.; Mackay, D., Evapora- tion Rate of Spills of Hydrocarbons and Petroleum Mixtures. Environ. Sci. Technol. 1984, 18 (11), 834-840. 43. Fingas, M. F., Modeling Oil and Petroleum Evaporation. J. Pet. Sci. Res. 2013, 2 (3), 104-115. 44. (a) Camilli, R.; Reddy, C. M.; Yoerger, D. R.; Van Mooy, B. A. S.; Jakuba, M. V.; Kinsey, J. C.; McIntyre, C. P.; Sylva, S. P.; Maloney, J. V., Tracking Hydrocarbon Plume Transport and Biodegradation at Deepwater Horizon. Science 2010, 330 (6001), 201-204; (b) Reddy, C. M.; Arey, J. S.; Seewald, J. S.; Sylva, S. P.; Lemkau, K. L.; Nelson, R. K.; Carmichael, C. A.; McIntyre, C. P.; Fenwick, J.; Ventura, G. T.; Van Mooy, B. A. S.; Camilli, R., Com- position and Fate of Gas and Oil Released to the Water Column During the Deepwater Horizon Oil Spill. Proc. Natl. Acad. Sci. U. S. A. 2012, 109 (50), 20229-20234; (c) Ryerson, T. B.; Camilli, R.; Kessler, J. D.; Kujawinski, E. B.; Reddy, C. M.; Valentine, D. L.; Atlas, E.; Blake, D. R.; de Gouw, J.; Meinardi, S.; Parrish, D. D.; Peischl, J.; Seewald, J. S.; Warneke, C., Chemical Data Quantify Deepwater Horizon Hydrocarbon Flow Rate and Environmental Distribution. Proc. Natl. Acad. Sci. U. S. A. 2012, 109 (50), 20246-20253. 45. (a) Fay, J. A., The Spread of Oil Slicks on a Calm Sea. 1971, 53-63; (b) Hoult, D. P., Oil Spreading on the Sea. Annu. Rev. Fluid Mech. 1972, 4, 341-368. 46. (a) National Research Council, Oil Spill Dispersants: Efficacy and Effects. The National Academies Press: Washington, DC, 2005; (b) Zhao, L.; Boufadel, M. C.; Adams, E. E.; Socolofsky, S. A.; Lee, K., A Numerical Model for Oil Droplet Evolution Emanating from Blowouts. In Proceedings of the International Oil Spill Conference, Savannah, GA, 2014; pp 561-571. 47. Johansen, Ø.; Brandvik, P. J.; Farooq, U., Droplet Breakup in Subsea Oil Releases – Part 2: Predictions of Droplet Size Distributions with and without Injection of Chemical Dispersants. Mar. Pollut. Bull. 2013, 73 (1), 327-335. 48. Boufadel, M. C.; Abdollahi-Nasab, A.; Geng, X.; Galt, J.; Torlapati, J., Simulation of the Landfall of the Deepwater Horizon Oil on the Shorelines of the Gulf of Mexico. Environ. Sci. Technol. 2014, 48 (16), 9496-9505.

REFERENCES 129

49. Valentine, D. L.; Fisher, G. B.; Bagby, S. C.; Nelson, R. K.; Reddy, C. M.; Sylva, S. P.; Woo, M. A., Fallout Plume of Submerged Oil from Deepwater Horizon. Proc. Natl. Acad. Sci. U. S. A. 2014, 111 (45), 15906-15911. 50. (a) Boufadel, M. C.; Bechtel, R. D.; Weaver, J., The Movement of Oil Under Non- Breaking Waves. Mar. Pollut. Bull. 2006, 52 (9), 1056-1065; (b) Boufadel, M. C.; Du, K.; Kaku, V.; Weaver, J., Lagrangian Simulation of Oil Droplets Transport Due to Regular Waves. Environ. Modell. Softw. 2007, 22 (7), 978-986. 51. Fingas, M.; Fieldhouse, B., Studies on Crude Oil and Petroleum Product Emulsions: Water Resolution and Rheology. Colloids Surf., A 2009, 333 (1-3), 67-81. 52. National Transportation Safety Board Enbridge Incorporated Hazardous Liquid Pipeline Rupture and Release, Marshall, Michigan, July 25, 2010; NTSB/PAR-12/01; Washington, DC, 2012. 53. Zhao, L.; Torlapati, J.; Boufadel, M. C.; King, T.; Robinson, B.; Lee, K., VDROP: A Com- prehensive Model for Droplet Formation of Oils and Gases in Liquids - Incorporation of the Interfacial Tension and Droplet Viscosity. Chem. Eng. J. 2014, 253 (1), 93-106. 54. (a) Frelichowska, J.; Bolzinger, M. A.; Chevalier, Y., Effects of solid particle content on properties of o/w Pickering emulsions. J. Colloid Interface Sci. 2010, 351 (2), 348-356; (b) Le Floch, S.; Guyomarch, J.; Merlin, F. X.; Stoffyn-Egli, P.; Dixon, J.; Lee, K., The Influ- ence of Salinity on Oil-Mineral Aggregate Formation. Spill Sci. Technol. Bull. 2002, 8 (1), 65-71. 55. Stoffyn-Egli, P.; Lee, K., Formation and Characterization of Oil–Mineral Aggregates. Spill Sci. Technol. Bull. 2002, 8 (1), 31-44. 56. Fitzpatrick, F. A.; Boufadel, M. C.; Johnson, R.; Lee, K.; Graan, T. P.; Bejarano, A. C.; Zhu, Z.; Waterman, D.; Capone, D. M.; Hayter, E.; Hamilton, S. K.; Dekker, T.; Garcia, M. H.; Hassan, J. S. Oil-Particle Interactions and Submergence from Crude Oil Spills in Marine and Freshwater Environments—Review of the Science and Future Science Needs; Open-File Report 2015–1076; U.S. Geological Survey: 2015. 57. Lee, K.; Bugden, J.; Cobanli, S.; King, T.; McIntyre, C.; Robinson, B.; Ryan, S.; Wohlgeschaffen, G. UV-Epifluorescence Microscopy Analysis of Sediments Recovered from the Kalamazoo River; Centre for Offshore Oil, Gas and Energy Research (COOGER): Dartmouth, Nova Scotia, 2012. 58. (a) Khelifa, A.; Hill, P. S.; Lee, K., The Role of Oil-Sediment Aggregation in Dispersion and Biodegradation of Spilled Oil. In Oil Pollution and its Environmental Impact in the Arabian Gulf Region, Al-Azab, M.; El-Shorbagy, W.; Al-Ghais, S., Eds. Elsevier: Amster- dam, Netherlands, 2005; pp 131-145; (b) Lee, K., Oil-Particle Interactions in Aquatic Environments: Influence on The Transport, Fate, Effect and Remediation of Oil Spills. Spill Sci. Technol. Bull. 2002, 8 (1), 3-8; (c) Sun, J.; Khelifa, A.; Zheng, X.; Wang, Z.; So, L. L.; Wong, S.; Yang, C.; Fieldhouse, B., A Laboratory Study on the Kinetics of the Formation of Oil-Suspended Particulate Matter Aggregates Using the Nist-1941b Sedi- ment. Mar. Pollut. Bull. 2010, 60 (10), 1701-1707. 59. Wang, C. Y.; Calabrese, R. V., Drop Breakup in Turbulent Stirred-Tank Contactors. Part II: Relative Influence of Viscosity and Interfacial Tension. AIChE J. 1986, 32 (4), 667-676. 60. Perez, S.; Furlan, P.; Hussein, N.; Shinn, D.; Crook, R., Interaction Between Oil and Suspended Sediments in Class 1-2 Rivers (Poster). In Proceedings of the International Oil Spill Conference, Savannah, GA, 2014; p 299120. 61. Waterman, D. M.; Garcia, M. H. Laboratory Tests of Oil-Particle Interactions in a Freshwater Riverine Environment with Cold Lake Blend Weathered Bitumen; No. 106; University of Illinois: Urbana, Illinois, 2015. 62. Short, J. W. Susceptibility of Diluted Bitumen Products from the Alberta Tar Sands to Sinking in Water; A51148; JWS Consulting LLC: 2013. 63. Fetter, C. W., Contaminant Hydrogeology. 2nd ed.; Prentice Hall: Upper Saddle River, NJ, 1999.

130 SPILLS OF DILUTED BITUMEN FROM PIPELINES

64. Ng, G. H. C.; Bekins, B. A.; Cozzarelli, I. M.; Baedecker, M. J.; Bennett, P. C.; Amos, R. T., A Mass Balance Approach to Investigating Geochemical Controls on Secondary Water Quality Impacts at a Crude Oil Spill Site Near Bemidji, MN. J. Contam. Hydrol. 2014, 164, 1-15. 65. Wesley, J. K. Kalamazoo River Assessment; Special Report 35; Michigan Department of Natural Resources, Fisheries Division: Ann Arbor, MI, 2005. 66. (a) Fitzpatrick, F. A.; Johnson, R.; Zhu, Z.; Waterman, D.; McCulloch, R. D.; Hayer, E. J.; Garcia, M. H.; Boufadel, M.; Dekker, T.; Hassan, J. S.; Soong, D. T.; Hoard, C. J.; Lee, K., Integrated Modeling Approach for Fate and Transport of Submerged Oil and Oil-Partricle Aggregates in a Freshwater Riverine Environment. In Proceedings of the Joint Federal Interagency Conference on Sedimentation and Hydrologic Modeling, Reno, NV, 2015; (b) Soong, D. T.; Hoard, C. J.; Fitzpatrick, F. A.; Zelt, R. B., Preliminary Analysis of Suspended Sediment Rating Curves for the Kalamazoo River and its Tributaries from Marshall to Kalamazoo, Michigan. In Proceedings of the Joint Federal Interagency Confer- ence on Sedimentation and Hydrologic Modeling, Reno, NV, 2015. 67. Hult, M. F. Ground-Water Contamination by Crude Oil at the Bemidji, Minnesota, Research Site: U.S. Geological Survey Toxic Waste--Ground-Water Contamination Study; Report 84- 4188; 1984. 68. Essaid, H. I.; Bekins, B. A.; Herkelrath, W. N.; Delin, G., N., Crude Oil at the Bemidji Site - 25 Years of Monitoring, Modeling, and Understanding. Groundwater 2011, 49 (5), 706-726. 69. Delin, G. N.; Essaid, H. I.; Cozzarelli, I. M.; Lahvis, M. H.; Bekins, B. A. Ground Water Contamination by Crude Oil near Bemidji, Minnesota; USGS Fact Sheet 084-98; U.S. Geo- logical Survey: Mounds View, MN, 1998. 70. Michigan Petroleum Pipeline Task Force Michigan Petroleum Pipeline Task Force Report; Michigan Department of Attorney General Lansing, MI, 2015. 71. Michel, J.; Rutherford, N. Oil Spills in Marshes: Planning & Response Considerations; U.S. Department of Commerce, National Oceanic and Atmospheric Administration, and the American Petroleum Institute: Washington, DC, 2013. 72. Shen, H. W.; Julien, P. Y., Erosion and Sediment Transport. In The Handbook of Hydrology, Maidment, D. R., Ed. McGraw-Hill: New York, 1993. 73. Geng, X.; Boufadel, M. C., Impacts of evaporation on subsurface flow and salt accumulation in a tidally influenced beach. Water Resour. Res. 2015, 51 (7), 5547-5565. 74. Crude Quality Inc. CrudeMonitor.ca. http://www.crudemonitor.ca/ (accessed 10/6/2015). 75. 3 Companies Plead Guilty to Burnaby Oil Spill. CBC News October 3, 2011. 76. Shang, D.; Buday, C.; van Aggelen, G.; Colodey, A., Toxicity Evaluation of the Oil Surface Washing Agent Corexit® 9580 and its Shoreline Application in Burrard Inlet, British Columbia. In Proceedings of the 35th Arctic and Marine Oilspill Program (AMOP) Technical Seminar, Environment Canada: Ottawa, Canada, 2012. 77. Stantec Summary of Clean up and Effects of the 2007 Spill of Oil from Trans Mountain Pipeline to Burrard Inlet; Project No. 1231-10505; 2012. 78. Madison, B. N.; Hodson, P. V.; Langlois, V. S., Diluted Bitumen Causes Deformities and Molecular Responses Indicative of Oxidative Stress in Japanese Medaka Embryos. Aquat. Toxicol. 2015, 165, 222-230.

REFERENCES 131

79. (a) Colavecchia, M. V.; Backus, S. M.; Hodson, P. V.; Parrott, J. L., Toxicity of Oil Sands to Early Life Stages of Fathead Minnows (Pimephales Promelas). Environ. Toxicol. Chem. 2004, 23 (7), 1709-1718; (b) Colavecchia, M. V.; Hodson, P. V.; Parrott, J. L., CYP1A Induction and Blue Sac Disease in Early Life Stages of White Suckers (Catostomus commersoni) Exposed to Oil Sands. J. Toxicol. Environ. Health, Part A 2006, 69 (10), 967-994; (c) Colavecchia, M. V.; Hodson, P. V.; Parrott, J. L., The Relationships among CYP1A Induction, Toxicity, and Eye Pathology in Early Life Stages of Fish Exposed to Oil Sands. J. Toxicol. Environ. Health, Part A 2007, 70 (18), 1542-1555; (d) Tetreault, G. R.; McMaster, M. E.; Dixon, D. G.; Parrott, J. L., Using Reproductive Endpoints in Small Forage Fish Species to Evaluate the Effects of Athabasca Oil Sands Activities. Environ. Toxicol. Chem. 2003, 22 (11), 2775-2782. 80. Papoulias, D. M.; Velez, V.; Nicks, D. K.; Tillitt, D. E. Health Assessment and Histopatho- logic Analyses of Fish Collected from the Kalamazoo River, Michigan, Following Discharges of Diluted Bitumen Crude Oil from the Enbridge Line 6B; Administrative Report; U.S. Geological Survey: Reston, VA, 2014. 81. Agency for Toxic Substances and Disease Registry Draft Toxicological Profile for Hydrogen Sulfide and Carbonyl Sulfide. U.S. Department of Health and Human Services: Atlanta, GA, 2014. 82. Committee on Environment and Natural Resources Interagency Assessment of Oxygen- ated Fuels; National Science and Technology Council: Washington, DC, 1997. 83. U.S. Centers for Disease Control and Prevention Drinking Water, Water Sources. http://www.cdc.gov/healthywater/drinking/public/water_sources.html (accessed 11/13/2015). 84. Railroad Commission of Texas. Field Guide for the Assessment and Cleanup of Soil and Groundwater Contaminated with Condensate From a Spill Incident. http://www. rrc.state.tx.us/oil-gas/environmental-cleanup-programs/guidance-documents-and- helpful-links/condensate-spill-guidance/ (accessed 11/13/2015). 85. Michigan Department of Community Health Public Health Assessment: Kalamazoo River/ Enbridge Spill: Evaluation of crude oil release to Talmadge Creek and Kalamazoo River on residential drinking water wells in nearby communities (Calhoun and Kalamazoo Counties, Michigan); U.S. Department of Health and Human Services: Atlanta, GA, 2013. 86. Baker, M. E.; Wiley, M. J.; Seelbach, P. W., GIS-Based Hyirologic Modeling of Riparian Areas: Implications for Stream Water Quality. J. Am. Water Resour. Assoc. 2001, 37 (6), 1615-1628. 87. American Society of Civil Engineers Task Committee on Modeling of Oil Spills, State- of-the-Art Review of Modeling Transport and Fate of Oil Spills. J. Hydraul. Eng. 1996, 122 (11), 594-609. 88. Lambert, P.; Goldthorp, M.; Fieldhouse, B.; Jones, N.; Laforest, S.; Brown, C. E., Health and safety concerns at dilbit crude oil spills for Environment Canada’s responders. In Proceedings of the 38th Arctic and Marine Oilspill Program (AMOP) Technical Seminar, Environment Canada: Vancouver, Canada, 2015; pp 664-678. 89. Harrill, J. A.; Wnek, S. M.; Pandey, R. B.; Cawthon, D.; Nony, P.; Goad, P. T., Strategies for Assessing Human Health Impacts of Crude Oil Releases. In Proceedings of the Inter- national Oil Spill Conference, Savannah, GA, 2014; pp 1668-1685. 90. U.S. Environmental Protection Agency Bridger Pipeline Release. http://www2.epa. gov/region8/bridger-pipeline-release (accessed 11/13/2015). 91. National Oceanic and Atmospheric Administration Shoreline Assessment Manual; HAZMAT Report No. 2000-1; Seattle, WA, 2013. 92. Whelan, A.; Clark, J.; Andrew, G.; Michel, J.; Benggio, B., Developing Cleanup End- points for Inland Oil Spills. In Proceedings of the International Oil Spill Conference, Savan- nah, GA, 2014; pp 1267-1280.

132 SPILLS OF DILUTED BITUMEN FROM PIPELINES

93. Tsaprailis, H. Properties of Dilbit and Conventional Crude Oils; 2480002; Alberta Innovates: 2013. 94. Fingas, M., Diluted Bitumen (Dilbit): A Future High Risk Spilled Material. In Proceed- ings of Interspill, Amsterdam, Netherlands, 2015; p 24. 95. Fieldhouse, B.; Mihailov, A.; Moruz, V., Weathering of Diluted Bitumen and Implica- tions to the Effectiveness of Dispersants. In Proceedings of the 37th Arctic and Marine Oilspill Program (AMOP) Technical Seminar, Environment Canada: Ottawa, Canada, 2014; pp 338-352. 96. Guenette, C. C.; Sveum, P.; Buist, I.; Aunaas, T.; Godal, L. In Situ Burning of Water-in-Oil Emulsions; SINTEF Report STF21 A94053; 1994. 97. Michel, J.; Benggio, B.; Keane, P., Pre-Authorization for The Use of Solidifiers: Results and Lessons Learned. In Proceedings of the International Oil Spill Conference, Savannah, GA, 2008; pp 345-348. 98. Brown, H. M.; Goodman, R. H., The Recovery of Spilled Heavy Oil with Fish Netting. In Proceedings of the International Oil Spill Conference, Washington, DC, 1989; pp 123-126. 99. Brown, H. M.; Nicholson, P., The Containment of Heavy Oil in Flowing Water. In Proceedings of the 15th Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, Edmonton, Canada, 1992; pp 457-465. 100. Michel, J.; Galt, J. A., Conditions under which floating slicks can sink in marine settings. In Proceedings of the International Oil Spill Conference, Long Beach, CA, 1995; pp 573-576. 101. McLinn, E. L.; Stolzenberg, T. R., Ebullition-Facilitated Transport of Manufactured Gas Plant Tar from Contaminated Sediment. Environ. Toxicol. Chem. 2009, 28 (11), 2298-2306. 102. Dollhopf, R., Michel, J., Ed. U.S. Environmental Protection Agency: 2015. 103. National Oil and Hazardous Substances Pollution Contingency Plan. Code of Federal Regulations, Title 40, Part 300, 1994. 104. Comprehensive Environmental Response,Compensation, and Liability Act (CERCLA). 42 U.S. Code §9610. 105. Oil Pollution Act of 1990 (OPA). 33 U.S. Code §2701-2761. 106. U.S. Environmental Protection Agency. Facility Response Planning Compliance Assistance Guide; Oil Program Center: Washington, DC, 2002. 107. U.S. Environmental Protection Agency. National Response Team. http://www2.epa. gov/emergency-response/national-response-team (accessed 11/13/2015). 108. U.S. Environmental Protection Agency. Regional Response Teams. http://www2.epa. gov/emergency-response/regional-response-teams (accessed 11/13/2015). 109. U.S. Environmental Protection Agency. Area Contingency Planning. http://www2.epa. gov/oil-spills-prevention-and-preparedness-regulations/area-contingency-planning (accessed 11/13/2015). 110. Clean Water Act. 33 U.S. Code §1251 et seq. 111. Facility Response Plans. Code of Federal Regulations, Section 112.20, Title 40, 2012. 112. Executive Order 12777. Implementation of Section 311 of the Federal Water Pollution Control Act of October 18, 1972, as Amended, and the Oil Pollution Act of 1990. 1991. 113. Response Plans for Onshore Oil Pipelines. Code of Federal Regulations, Section 194, Title 49. 114. Transportation of Hazardous Liquids by Pipeline. Code of Federal Regulations, Section 195, Title 49. 115. Gilbride, P.; Barnes-Weaver, E.; Strasser, M. A.; Wake, S. EPA Could Improve Contingency Planning for Oil and Hazardous Substance Response; Report Number 13-P-0152; Office of Inspector General: Washington, DC, 2013.

REFERENCES 133

116. (a) Facilities Transferring Oil or Hazardous Material in Bulk. Code of Federal Regulations, Section 154, Title 33; (b) Oil or Hazardous Material Pollution Prevention Regulations for Vessels. Code of Federal Regulations, Section 155, Title 33; (c) Caplis, J. R. MER Policy Letter 03-13; Oil Spill Removal Organization (OSRO) Classification Program; U.S Coast Guard: Washington, DC, 2013. 117. McNutt, M., A Community for Disaster Science. Science 2015, 348 (6230), 11.

Appendix A

Glossary

Acute—An event occurring over a short time, usually a few minutes or hours. An acute effect happens within a short time after exposure. An acute exposure can result in short-term or long-term health effects. Acute toxicity to aquatic organisms is estimated from short exposures, usually 24, 48, or 96 hours and lethality (death) is the typical endpoint. Results from acute toxicity tests usually report the lethal concentration of the toxicant that causes death to 50% of the test organisms (LC50). The lower the LC50 value the greater the toxicity of the toxicant.

Bitumen—A mixture of hydrocarbons that is too viscous to flow under ambient conditions. Commercial quantities are recovered by thermal processes.

Chronic—An event occurring over a long period of time, generally weeks, months, or years. Chronic exposures occur over an extended period of time or over a significant fraction of an organism’s lifetime. Chronic toxicity to aquatic organisms can be estimated from partial life-cycle tests of relatively short duration depending on the organism (i.e., 7 – 21 days) and growth and reproduction are the typical endpoints. Results from chronic toxicity tests are reported as the toxicant concentration that causes a given effect.

Commonly transported crude oils—The oils carried by most transmission pipelines in the United States. Available data show that currently >70% of these are light and medium crude oils. 135

136 APPENDIX A

Conventional oil—Oil that is produced by drilling and pumping from a naturally permeable, subsurface reservoir.

Crude oil—Naturally occurring, unrefined petroleum hydrocarbons extracted from the earth to serve as feedstock for the petroleum industry.

Diluted bitumen—Bitumen diluted with lighter hydrocarbons or a combination of light hydrocarbons such as natural-gas condensate, lighter crude oil, or synthetic oil, such that its viscosity is reduced.

Diluted heavy oil—Heavy oil that is diluted with lighter hydrocarbons such as natural-gas condensate, lighter crude oil, or synthetic oil, such that its viscosity is reduced.

Gas condensate—A low-density mixture of hydrocarbon liquids obtained by condensing the less-volatile components of raw natural gas.

Heavy crude oil—A naturally occurring, unrefined petroleum product with a density greater than 0.93 g/cm3 or an API gravity less than 20.

High molecular weight petroleum compounds—Compounds with a molecular weight greater than about 250 Daltons; typically these compounds are viscous liquids or solids at ambient temperatures.

Light crude oil—A naturally occurring, unrefined petroleum product with a density ranging from 0.80 to 0.85 g/cm3 or an API gravity ranging from 35 to 45.

Low molecular weight petroleum compounds—Compounds with a molecular weight lower than about 250 Daltons; typically these compounds are liquids at ambient temperatures.

Medium crude oil—A naturally occurring, unrefined petroleum product with a density ranging from 0.85 to 0.92 g/cm3 or an API gravity ranging from 36 to 21.

Sublethal—Toxicant is below the concentration that directly causes death. Exposures to sublethal concentrations of a toxicant may produce less obvious and measureable effects on behavior, molecular, biochemical, cellular and/or physiological function (e.g., growth and reproduction) and histology of organisms.

APPENDIX A 137

Synthetic crude—Oil produced from bitumen by physical and chemical processes less elaborate than those used in full-scale refineries and implemented near the site of production.

Transmission pipeline—A continuous pipe used to transport oil and petroleum products from gathering points to storage or distribution points. It is distinct from smaller, shorter pipelines used to collect oil from individual wells or to distribute to points of consumption.

Unconventional oil—Oil that is produced by unconventional means, including thermal separation of non-liquid bitumen from a host rock and hydraulic fracturing of impermeable reservoirs or source rocks.

Undiluted heavy oil—Heavy oil that is transported without dilution. It may be heated to facilitate transport.

Upgraded bitumen—Bitumen that has been subject to some refinement to remove or convert some of the more recalcitrant components. It is also an intermediate in the production of synthetic crude oil.

Appendix B

Committee Member and Staff Biographies

Committee Diane McKnight (chair) is a professor of civil, environmental and archi- tectural engineering and a fellow of the Institute of Arctic and Alpine Research at the University of Colorado. Her research focuses on interactions between hydrologic, chemical, and biological processes in controlling the dynamics in aquatic ecosystems. This research is carried out through field-scale experiments, modeling, and laboratory characterization of natural substrates. In addition, Dr. McKnight conducts research focusing on interactions between freshwater biota, trace metals, and natural organic material in diverse freshwater environments, including lakes and streams in the Colorado Rocky Mountains and in the McMurdo Dry Valleys in Antarctica. She interacts with state and local groups involved in mine drainage and watershed issues in the Rocky Mountains. Dr. McKnight is a member of the National Research Council’s (NRC’s) Polar Research Board and is a former member of the Water Science and Technology Board. She is a past president of the American Society of Limnology and Oceanography and the Biogeosciences section of the American Geophysi- cal Union. She received her Ph.D. in environmental engineering from the Massachusetts Institute of Technology and is a member of the National Academy of Engineering.

Michel Boufadel is professor of environmental engineering and director of the Center for Natural Resources Development and Protection at the New Jersey Institute of Technology. He is a professional engineer in

139

140 APPENDIX B

Pennsylvania and New Jersey. Dr. Boufadel has conducted, since 2001, research projects funded by the U.S. Environmental Protection Agency and the National Oceanic and Atmospheric Administration (NOAA) on oil dispersion and transport offshore. He has adopted a multiscale approach where he conducts experiments in flasks and wave tanks of various sizes and models processes from the microscopic scale to the sea scale. Dr. Boufadel was involved in the response to the Deepwater Horizon blowout and assisted NOAA personnel conducting various tasks within the response. Dr. Boufadel has more than 80 refereed articles in publications such as Marine Pollution Bulletin, Environmental Science and Technology, and the Journal of Geophysical Research. He also has more than 30 publications in oil spill conference proceedings, such as those of the International Oil Spill Conference and Arctic and Marine Oil Spill. He is an associate editor of the Journal of Environmental Engineering, American Society of Civil Engineers.

Merv Fingas is a scientist working on oil and chemical spills. He was Chief of the Emergencies Science Division of Environment Canada for over 30 years and is currently working on research in Western Canada. Dr. Fingas has a Ph.D. in environmental physics from McGill University, and three master’s degrees—chemistry, business, and mathematics—all from University of Ottawa. He also has a bachelor of science in chemistry from Alberta and a bachelor of arts from Indiana. He has more than 860 papers and publications in the field. Dr. Fingas has prepared seven books on spill topics and is working on two others. He has served on two committees on the National Academy of Sciences of the U.S. on oil spills including the recent “Oil in the Sea.” He is chairman of several ASTM and intergovernmental committees on spill matters. Importantly, he was the founding chairman of the ASTM subcommittee on in situ burning and chairman of oil spill treating agents and another on oil spill detection and remote sensing, positions he holds today. Dr. Fingas began his career in 1974 working for Environment Canada as a scientist working on oil and chemical spills. His first tasks were largely to work on the Beaufort Sea Studies, a multi-million-dollar joint industry-government program to develop oil spill readiness for the Canadian Beaufort Sea. His role in these studies was to coordinate chemical and physical studies and to prepare overview documents. With the completion of these studies a new study of large magnitude, the Arctic and Marine Oil Spill Program, was founded in 1977. Dr. Fingas, one of the founders of this program, worked in general coordination on the program and specifically managed a number of subprojects, including those on chemistry, oil behavior, remote sensing, spill tracking, and spill treating agents. Dr. Fingas continued in many of these research fields until

APPENDIX B 141

today. His specialties include Arctic oil spills, oil chemistry, spill dynamics and behavior, spill treating agents, remote sensing and detection, spill tracking, and in situ burning. He continues research and writing on these topics to this day.

Stephen Hamilton is currently a professor of ecosystem ecology and biogeochemistry at Michigan State University. He received his Ph.D. at the University of California, Santa Barbara, in 1994. His principal research interests involve ecosystem ecology and biogeochemistry, with particular attention to nutrients and biogeochemical processes in aquatic environments as well as agricultural ecosystems. His research integrates approaches from varied disciplines such as geology, chemistry, remote sensing, and hydrology as well as ecology. He has conducted research on various aspects of aquatic ecosystems in southern Michigan, including wetlands, streams, lakes, and watersheds. He also works on tropical ecosystems in South America and dryland river ecosystems in Australia. Since 2006 he has been President of the Kalamazoo River Watershed Council. He served as an independent, volunteer advisor to the U.S. Envi- ronmental Protection Agency for the 2010 Marshall, MI pipeline release of oil sands crude into the Kalamazoo River, and was a member of its Scientific Support Coordination Group.

Orville “O.B.” Harris is President of O.B. Harris, LLC, which is an independent consultancy specializing in the regulation, engineering, and planning of petroleum liquids pipelines. Currently, he is the Independent Monitoring Contractor for the Consent Decree between the U.S. and BP Alaska, Inc. From 1995 to 2009, he was Vice President of Longhorn Part- ners Pipeline, L.P., which operated a 700-mile pipeline that carried gasoline and diesel fuel from Gulf Coast refineries to El Paso, Texas. In this position, he was responsible for engineering, design, construction, and operation of the system. From 1991 to 1995, he was President of ARCO Transportation Alaska Inc., which owned four pipeline systems, including a portion of the Trans Alaska Pipeline System (TAPS) which transports crude oil from the North Slope of Alaska to the Port of Valdez. From 1977 to 1990, he held several supervisory and managerial positions at the ARCO Pipeline Company, including District Manager for Houston and Midland, Texas, Manager of the Northern Area, and Manager of Prod- ucts Business. While at ARCO Transportation, he directed the efforts of a team of corrosion experts guiding $400 million of repairs to the TAPS system. He is a past member of the Board of Directors of the Associa- tion of Oil Pipelines and the Pipeline and Hazardous Materials Safety Administration’s Technical Hazardous Liquids Pipeline Safety Standards Committee. Mr. Harris joins this committee having previously served on

142 APPENDIX B

the committee for the preceding NRC study, Effects of Diluted Bitumen on Crude Oil Transmission Pipelines. He holds a bachelor’s degree in civil engineering from the University of Texas and an M.B.A. from Texas Southern University.

John Hayes is Scientist Emeritus and retired Director of the National Ocean Sciences Accelerator Mass Spectrometry Facility (NOSAMS) at the Woods Hole Oceanographic Institution. His work has dealt with isotope effects in biochemical reactions and their significance and utility in studies of geochemical processes. Recent topics have included studies of carbon- and hydrogen-isotopic fractionations imposed by phyto- plankton and other microorganisms, paleoenvironmental studies based on sedimentary isotopic and organic-geochemical records, studies of the anaerobic oxidation of methane in marine sediments, the long-term record of 13C in sedimentary organic carbon, and developments in stable-isotopic analytical techniques. He has served as chair of the Organic Geochemical Division of the Geochemical Society and on the Executive Committee of the Integrated Ocean Drilling Program. He received his Ph.D. in chemistry from the Massachusetts Institute of Technology and is a member of the National Academy of Sciences and of the American Academy of Arts and Sciences.

Jacqueline Michel is a geochemist specializing in terrestrial and marine pollution studies, coastal geomorphology, and environmental risk assessments. She has specialized expertise in the behavior, tracking, recovery, and effects of submerged oil. Having worked in 32 countries, she has extensive international experience and has worked in many different coastal and marine environments. Dr. Michel is one of the founders of Research Planning, Inc. and has been President since 2000. She often leads multidisciplinary teams on projects where her problem-solving skills are essential to bringing solutions to complex issues. For example, her work during spill emergencies requires her to rapidly develop consensus and provide decision mak- ers with needed information. Because of her routine scientific support for spills, she has extensive knowledge of and practical experience in pollutant fate, transport, and effect issues. She has been a leader in the development of methods and the conduct of Natural Resource Damage Assessments following spills and groundings. She has taken a lead role in 29 damage assessments for federal and state trustees. Dr. Michel has been recognized for her achievements through appoint- ments to many respected committees and panels, including four National Academies committees: Spills of Nonfloating Oil (1999); Oil in the Sea (2002); Chair of Spills and Emulsified Fuels: Risk and Response (2001);

APPENDIX B 143

and Chair of the Committee on Understanding Oil Spill Dispersants: Effi- cacy and Effects (2005). She was on the Oceans Board for 2001-2005 and is a Lifetime Associate of the National Academies. She was on the Science Advisory Panel to the U.S. Commission on Ocean Policy. She is an adjunct professor in the School of the Environment, University of South Carolina. She has written over 225 technical publications.

Carys Mitchelmore earned her Ph.D. from the University of Birmingham (UK) in 1997 investigating toxicity processes and effects in aquatic organisms exposed to organic pollutants, including crude oil and its constituent polycyclic aromatic hydrocarbons (PAHs). Dr. Mitchelmore is an associate professor at the University of Maryland Center for Environmental Sci- ence, Chesapeake Biological Laboratory, in Solomons, MD. Her expertise lies in aquatic toxicology and her research experience includes understanding routes of exposure, bioaccumulation, metabolism, depuration, trophic transfer and the target sites of pollutants, including PAHs and emerging contaminants of concern. Investigations have used an array of organisms, from bacteria, algae, and invertebrate and vertebrate species, such as oysters, blue crabs, anemones, corals, fish, and reptiles. Cur- rent research projects are directed at understanding the uptake, routes of exposure (including chemical partitioning of dissolved and particulate fractions), fate and effects of oil, chemical dispersants (e.g., Corexit and alternatives) and dispersed oil. Focused areas of impact include DNA damage, oxidative stress and antioxidant responses, endocrine disruption, and immune function. Recent studies have investigated the use of oil rig- fouling organisms as biomonitoring tools to provide baseline datasets that can provide essential information regarding the recovery of organisms following a pollution event. Dr. Mitchelmore is also co-author of the 2005 NRC report Oil Spill Dispersants: Efficacy and Effects and also provided testimonies to various Senate and House committees following the Deep- water Horizon incident regarding dispersant use. Dr. Mitchelmore is also actively involved in determining the efficacy of various shipboard scale ballast water treatments and in investigating the occurrence and toxicity of chlorinated and brominated organic compounds.

Denise Reed, Ph.D., is a nationally and internationally recognized expert in coastal marsh sustainability and the role of human activities in modi- fying coastal systems. She has studied coastal issues in the United States and around the world for over 30 years. Dr. Reed has worked closely with Louisiana’s state government in developing coastal restoration plans. Her experience includes helping monitor natural resources in the Pontchartrain Basin following the Deep- water Horizon oil spill in 2010 and researching ecosystem restoration and

144 APPENDIX B

planning in the California Bay-Delta. She has served on numerous boards and panels addressing the effects of human alterations on coastal environments and the role of science in guiding restoration, including a number of National Research Council committees. Prior to joining The Water Institute of the Gulf, Dr. Reed served as Director of the Pontchartrain Institute for Environmental Sciences and as a professor in the University of New Orleans’ Department of Earth and Environmental Sciences. She is a member of the Chief of Engineers Environmental Advisory Board and the Ecosystems Sciences and Management Working Group of the NOAA Science Advisory Board. She earned a bachelor’s and doctoral degree in geography from the University of Cambridge.

Robert (Bob) Sussman is the principal in Sussman and Associates, a consulting firm that offers advice and support on energy and environmental policy issues to clients in the nonprofit and private sectors. He is on the adjunct faculty at Georgetown Law Center and Yale Law School. Sussman served for four and a half years in the Obama Administration, first as co-head of the transition team for the Environmental Protection Agency (USEPA) and then as Senior Policy Counsel to the USEPA Administra- tor. Mr. Sussman previously served in the Clinton Administration as the USEPA Deputy Administrator during 1993-1994. In this position, he was the Agency’s Chief Operating Officer and Regulatory Policy Officer. At the end of 2007, he retired as a partner at the law firm of Latham & Watkins, where he headed the firm’s environmental practice in Washing- ton, DC, for 10 years. Previously, he was a partner at Covington & Burling. For several years, Sussman was named one of the leading environmental lawyers in Washington, DC, by Chambers USA: America’s Leading Business Lawyers and The International Who’s Who of Environmental Lawyers. He was a Senior Fellow at the Center for American Progress in 2008, writing and speaking about climate change and energy. Sussman is a magna cum laude 1969 graduate of Yale College and a 1973 graduate of Yale Law School, where he was an editor of the Yale Law Journal. He clerked for Judge Walter K. Stapleton of the Third Circuit Court of Appeals.

David Valentine currently serves as a professor in the Department of Earth Science at the University of California, Santa Barbara. His research interests focus on the interface of geochemistry and microbiology. Valen- tine is an Aldo Leopold Leadership Fellow and the recipient of a CAREER award in chemical oceanography from the National Science Foundation. He is best known for his research on archaeal ecology, methane biogeochemistry, oil seeps, and the aftermath of the Deepwater Horizon event, as well as for engagement with popular media.“

APPENDIX B 145

Staff Douglas Friedman is a Senior Program Officer with the Board on Chemi- cal Sciences and Technology at the National Academy of Sciences, Engi- neering, and Medicine in Washington, DC. His primary scientific interests lie in the fields of organic chemistry, organic & bio-organic materials, chemical & biological sensing, and nanotechnology, particularly as they apply to national and homeland security. Dr. Friedman has supported a diverse array of activities since joining the Academies. He served as study director on Industrialization of Biology: A Roadmap to Accelerate the Advanced Manufacturing of Chemicals; Safe Science: Promoting a Culture of Safety in Academic Chemical Research; Transforming Glycoscience: A Roadmap for the Future; Determining Core Capabilities in Chemical and Biological Defense Sci- ence and Technology; Effects of Diluted Bitumen on Crude Oil Transmission Pipelines; and Responding to Capability Surprise: A Strategy for U.S. Naval Forces. Additionally, he has supported activities on Convergence: Safeguard- ing Technology in the Bioeconomy, The Role of the Chemical Sciences in Finding Alternatives to Critical Resources; Opportunities and Obstacles in Large-Scale Biomass Utilization; and Technological Challenges in Antibiotics Discovery and Development. Dr. Friedman is currently directing studies on the environmental effects of diluted bitumen oil spills, safeguarding technology in the bioeconomy, and the regulation of biotechnology. Prior to joining the NRC he performed research in physical organic chemistry and chemical biology at Northwestern University, the University of California, Los Angeles, the University of California, Berkeley, and Solulink Biosciences. He received a Ph.D. in Chemistry from Northwestern University and a B.S. in Chemical Biology from the University of California, Berkeley.

Camly Tran joined the Board on Chemical Sciences and Technology at the National Academy of Sciences, Engineering, and Medicine in 2014 as a postdoctoral fellow after receiving her Ph.D. in chemistry from the Department of Chemistry at Brown University and is currently an Asso- ciate Program Officer. During her time at Brown, she received various honors including the Elaine Chase Award for Leadership and Service, American Chemical Society Global Research Exchanges Education Train- ing Program, and the Rhode Island NASA grant. Dr. Tran completed the workshop summary Mesoscale Chemistry and is currently supporting activities on the environmental effects of diluted bitumen oil spills, the changing landscape of hydrocarbon feedstocks for chemical production, and the standard operating procedures for safe and secure handling, management, and storage of chemicals in chemical laboratories.