MARINE ECOLOGY PROGRESS SERIES Mar Ecol Prog Ser Vol. 324: 37?55, 2006 Published October 23 INTRODUCTION Worldwide transfer and introduction of non-indige- nous species (NIS) by human activities has significant ecological, economic and human-health impacts (Wil- cove et al. 1998, Pimentel et al. 2000). Most attention has focused on invasions in terrestrial and freshwater habitats, but NIS invasions have also become a potent force changing coastal marine ecosystems. At least 400 marine and estuarine NIS are established in North America and over 200 of these species can occur in one estuary (Cohen et al. 1995, Ruiz et al. 1997, 2000). Some ? Inter-Research 2006 ? www.int-res.com*Email: herwig@u.washington.edu Ozone treatment of ballast water on the oil tanker S/T Tonsina: chemistry, biology and toxicity Russell P. Herwig1,*, Jeffery R. Cordell1, Jake C. Perrins1, Paul A. Dinnel2, Robert W. Gensemer3, 7, William A. Stubblefield3, 7, Gregory M. Ruiz4, Joel A. Kopp5, 8, Marcia L. House1, 9, William J. Cooper6 1School of Aquatic and Fishery Sciences, Box 355020, University of Washington, Seattle, Washington 98195-5020, USA 2Shannon Point Marine Center, Western Washington University, 1900 Shannon Point Road, Anacortes, Washington 98221, USA 3ENSR International, 4303 West LaPorte Avenue, Fort Collins, Colorado 80521, USA 4Smithsonian Environmental Research Center, 647 Contees Wharf Road, PO Box 28, Edgewater, Maryland 21307-0028, USA 5Petrotechnical Resources Alaska, 310 K Street, Suite 407, Anchorage, Alaska 99510, USA 6Department of Chemistry and Center for Marine Science, University of North Carolina at Wilmington, 5600 Marvin K. Moss Lane, Wilmington, North Carolina 28409, USA Present addresses: 7Parametrix Inc., 33972 Texas Street Southwest, Albany, Oregon 97321, USA 8Consulate General Monterrey, US State Department, PO Box 9002, Brownsville, Texas 78520, USA 9Northwest Indian Fisheries Commission, 6730 Martin Way East, Olympia, Washington 98516, USA ABSTRACT: Worldwide transfer and introduction of non-indigenous species in ballast water causes significant environmental and economic impact. One way to address this problem is to remove or inactivate organisms that are found in ballast water. In this study, 3 experiments were conducted in Puget Sound, Washington, USA, using a prototype ozone treatment system installed on a commercial oil tanker, the S/T Tonsina. Treatment consisted of ozone gas diffused into a ballast tank for 5 and 10 h. Treatment and control tanks were sampled during the ozonation period for chemistry, cultur- able bacteria, phytoplankton and zooplankton. Selected fish and invertebrates were placed in cages deployed in the treatment and control tanks. Ozone introduced into seawater rapidly converts bro- mide (Br?) to bromines (HOBr/OBr?), compounds that are disinfectants. These were measured as total residual oxidant (TRO). Ozone treatment inactivated large portions of culturable bacteria, phyto- plankton and zooplankton. The highest reductions observed were 99.99% for the culturable bacteria, >99% for dinoflagellates and 96% for zooplankton. Caged animal results varied among taxa and locations in the ballast tank. Sheepshead minnows and mysid shrimp were most susceptible, shore crabs and amphipods the least. Distribution of ozone in the treatment tank was not homogenous during experiments, as suggested by the observed TRO concentrations and lower efficacies for inactivating the different taxa in selected ballast tank locations. Low concentrations of bromoform, a disinfection byproduct, were found in treated ballast water. KEY WORDS: Aquatic nuisance species ? Non-indigenous species ? Ballast water ? Ozone treatment ? Bromine ? Total residual oxidant Resale or republication not permitted without written consent of the publisher Mar Ecol Prog Ser 324: 37?55, 2006 of these species have become numerically or func- tionally dominant, and have significant impacts on population, community and ecosystem-level processes (Cloern 1996, Ruiz et al. 1999, Grosholz et al. 2000). The National Invasive Species Act of 1996 (NISA) created a program whereby vessels arriving from out- side of the Exclusive Economic Zone (EEZ) voluntarily conduct open-ocean ballast water exchange (BWE), or use an approved alternate treatment of ballast water permitting ballast tanks to be discharged in US ports. Recently, individual states (e.g. California, Maryland, Oregon, Washington and Virginia) passed and imple- mented similar laws, some making BWE mandatory. BWE can usually be implemented and does not require retrofitting or installing new technology, but it is often viewed as a ?stop-gap? measure. BWE has some signif- icant limitations, including ship safety issues, costs of compliance and variable effectiveness (Woodward et al. 1992, National Research Council 1996, Waite 2002, Matheickal & Raaymakers 2004). Following rule mak- ing in 2004, ballast water management is now manda- tory for all areas of the United States (Code of Federal Regulations, Title 33, Part 151, Subparts C and D). Individual states (e.g. California, Maryland, Oregon, Washington and Virginia) passed and implemented similar laws, sometimes making BWE mandatory even for coastal traffic that would not arrive from beyond the EEZ. The United States does not require vessels to deviate from its voyage or delay its voyage to conduct a BWE. Individual US states may impose diversion and delay by requiring ships to perform BWE 50 nautical miles from shore. A number of ballast treatment methods are being explored as alternatives to BWE, but their evaluation is at an early stage (National Research Council 1996, Hallegraeff 1998). Presently, the US Coast Guard (directed by NISA) requires alternative treatments to be at least as effective as BWE. More recently, the International Maritime Organization (IMO) (2004) adopted treatment standards and the state of Washing- ton also established interim treatment standards (Washington Department of Fish and Wildlife 2002). Ozone has been used as a disinfectant since the late 1800s, is used widely in Europe and, to a lesser extent, in the U.S in drinking water treatment (Hoign? 1998). It is biocidal oxidant that is unstable in water (Langlais et al. 1991). Ozone chemistry in seawater differs from that in freshwater because of the presence of bromide ion (Oemcke & van Leeuwen 1998). Bromide ion cat- alytically decomposes ozone (Fig. 1) (von Gunten et al. 1996, von Gunten & Oliveras 1998, Salhi & von Gunten 1999, von Gunten & Pinkernell 2000, Pinkernell & von Gunten 2001, Gallard et al. 2003, Gujer & von Gunten 2003, von Gunten 2003a, b). In seawater, the primary brominated compounds formed by ozone are hypobro- mous acid (HOBr), which is in equilibrium with hypo- bromite (OBr?). These compounds have disinfection properties. Bromoform, a disinfection byproduct, is formed by a reaction with natural organic matter in the water. The chemistry of ozone in seawater is complex (Fig. 1). In the presence of ammonia, HOBr/OBr? will react rapidly to form monobromamine (Johnson & Overby 1971, Haag & Hoign? 1984, Yang et al. 1999, Lei et al. 2004, Perrins et al. in press). Monobromamine can disproportionate to NHBr2 and NH3 (e.g. Lei et al. 2004) or with excess HOBr/OBr? it can react further to form N2 and bromide (e.g. Brunetto et al. 1989, Hof- man & Andrews 2001). Monobromamine is unstable and will decompose to ammonia and bromide ion (Hof- man & Andrews 2001). In this study, we examined the chemistry, biology and toxicity of a prototype treatment system that dif- fused ozone into a ship?s ballast tank containing sea- water. We conducted 3 experiments on a commercial oil tanker during September and November 2001 in Puget Sound, Washington, USA. MATERIALS AND METHODS S/T Tonsina, and prototype ozone system. The S/T Tonsina (since sold and re-named) was a 265 m (869ft) American-flagged oil tanker operated by Alaska Tanker Company (Portland, Oregon) transporting crude oil mainly between Valdez, Alaska, and refiner- ies on the west coast of the United States. The ship had a capacity of 4.16 ? 107 l or 4.16 ? 104 m3 in 12 ballast 38 NH2Br HOBr CHBr3 O3 OH Br BrO BrO H+ OBr? BrO2? BrO3? NH3 NOM Br? O3 OH OH, CO 3 ? O3O3 O3 O3 ? ? ? ? ? ? Fig. 1. Reaction pathways for decomposition of ozone in seawater showing formation of hypobromite (OBr?), hypo- bromous acid (HOBr), and disinfection byproducts bromate ion (BrO3?) and bromoform (CHBr3). NOM = natural organic matter. (Adapted from Driedger et al. [2001] with permission from Elsevier). Thick lines: preferred steps in seawater suggested by shipboard experiments performed onboard S/T Tonsina; thin lines: pathways found in bromide-contain- ing water ; dotted lines: reaction steps involving free radicals Herwig et al.: Ozone treatment of ballast water tanks, and 807 000 barrels (1.28 ? 105 m3) of crude oil in 12 cargo tanks. The ship was double-hulled, with space between the hulls divided transversely for carry- ing ballast water when the ship was empty or partially loaded. These ballast tanks were along the outer hull and double bottom area. In fall 2000, a prototype Nutech-O3 (McLean, Vir- ginia) ozonation system was installed on the S/T Ton- sina. This prototype, known as the SCX 2000, fit in a standard ISO 20 foot (6.1 m) container, which was in- stalled on the stack deck, an exterior location on the ship?s stern. Ozone was produced by injecting oxygen- enriched compressed air through a series of water- cooled electrodes. In each electrode, a high voltage corona discharge (electric arc) was created, using a standard ship?s 480 V power transformed to more than 10 000 V. A fraction of the oxygen-rich air passing through each corona gap was converted into ozone, which was collected and piped into one of the ballast tanks, through a system of flow meters and stainless steel pipe. Ozone was distributed into the tank through custom designed ceramic coated stone diffusers, arranged to maximize the distribution and contact time of the ozone in the ballast water. Ballast tank sampling. The No. 3 port (3P) and number 3 starboard (3S) ballast tanks were used for ozonation and controls, respectively (Fig. 2). The 3P tank was divided into A (fore) and B (aft) sections, and the 3S tank into C (fore) and D (aft) sections for duplicate sampling. Sec- tions A, B, C, and D were sampled in Expt 1 and Sections A, B, and C were sampled in Expts 2 and 3. For all sam- ples except zooplankton and caged animals (see later subsection), water from each tank section (A B, C, or D) was sampled at 3, 9 and 15 m below the ship?s deck using a 5 l Niskin water sampler (General Oceanics). Both the experimental and control bal- last tank sections were sampled before ozonation began (0 h) and at 2.5 and 5.0 h during ozonation for Expt 1, and at 2.5, 5.0, 7.5, and 10.0 h during ozonation for Expts 2 and 3. Chemistry. General water chem- istry: Subsamples from the Niskin water samplers were placed in clean Nalgene containers, and analyzed (fol- lowing the instructions) with the Hach DREL/2010 Water Quality Laboratory kit (Hach Company). pH was deter- mined using a Hach Portable pH Meter. Dissolved oxygen (DO) was measured with a Model 21800-022 Traceable? DO meter that was air calibrated and adjusted to compensate for salinity. Salinity was mea- sured using a conductivity meter with a range of 0 to 80 PSU (Hach Company). Temperature was deter- mined using a field thermometer. Samples for inor- ganic nutrients (orthophosphate, nitrite, nitrate, am- monia, silicic acid) and dissolved organic carbon (DOC) were frozen on board ship and stored frozen until analyzed. Inorganic nutrients and dissolved DOC were analyzed at the Marine Chemistry Laboratory in the School of Oceanography, University of Washing- ton, using a Technicon Model AAII and a Shimadzu TOC5000, respectively (Parsons et al. 1984). Ozone chemistry: Total residual oxidant (TRO). TRO was determined using a standard DPD colorimetric analysis for total chlorine (APHA 1998). Hach Accu- Vac? Ampules were submerged and filled with water immediately after this was collected from the ballast tank, and then analyzed on a Hach DREL/2010 water quality laboratory spectrometer on the ship. The ampoules had a range of 0 to 4.5 mg l?1 as Br2 with a sensitivity of 0.1 mg l?1 as Br2. TRO is a measure of halogen-containing oxidants. As described in the ?Introduction?, ozone quickly reacts with bromide ion in seawater, forming hypobromous acid that is in equi- librium with hypobromite. Together, these compounds are referred to as bromines and they constitute TRO measured in the ozonation of seawater. 39 Bow Port Fore Aft A B Treatment ballast tank Starboard Forepeak Number 1 port Number 2 port Number 3 port Number 1 starboard Number 3 starboard Number 2 starboard C D Control ballast tank Number 4 port Number 4 starboard Fig. 2. Diagram of S/T Tonsina showing locations of port and starboard ballast tanks. Samples were collected from fore and aft sections in treatment ballast tank (A and B) and control ballast tank (C and D). Circles: access hatches for treatment and control tanks; dashed lines: boundary between fore and aft sections. Figure is not to scale Mar Ecol Prog Ser 324: 37?55, 2006 Ozone. Ozone was measured using the indigo colori- metric technique (APHA 1998). Similar to the TRO measurement, AccuVac? Ampules were used with freshly collected samples and analyzed using a Hach DREL/2010 water quality laboratory spectrometer. The ampoules had a range of 0 to 1.5 mg?l, with a sensitiv- ity of 0.1 mg l?1 ozone. Oxidation reduction potential (ORP). ORP was mea- sured using an Orion 290A pH meter with a Cole- Palmer Combination ORP probe (Pt electrode, Ag/ AgCl reference cell). ORP was measured in mV. Bromate. Samples for bromate ion analysis were col- lected in 150 ml wide-mouth Nalgene HDPE bottles. They were stored on ice and shipped to analytical lab- oratories immediately after the end of each shipboard experiment. We used the US EPA Method 317.0 Revi- sion 2.0 (EPA 815-B-01-001), which measures bromate ion from 2 to 40 ?g l?1. For Expt 1, samples had a bromate concentration below the method detection limit of 2 ?g l?1. During the first set of analyses, we observed that control samples spiked with bromate were ?unrecoverable?. The sub- sequent evaluation of bromate standards prepared in distilled water showed good recovery. Experiments showed that at higher concentrations (i.e. at the mg l?1 level), spiked bromate could be recovered. Sub- sequently, all ballast water samples were diluted to 20% of their original concentration (1 part ballast water: 4 parts distilled water) in distilled water. With this dilution, we determined that adequate bromate ion recovery could be achieved at the 50 ppb level. Based on bromate recovery following a 1:4 dilution, all ballast water samples for Expts 2 and 3 were diluted. This dilution enabled detection of 10 ?g l?1 bromate, which is the maximum contaminant level (MCL) established for bromate in drinking water (EPA 816-F-01-010). Bromoform. Samples for bromoform analysis were collected in 40 ml volatile organic analysis (VOA) vials containing a sulfite fixative. They were stored on ice and shipped to the analytical laboratory immediately after completion of each experiment. Bromoform was analyzed using a purge and trap gas chromatograph following US EPA Method 524.2 (EPA 600-R-95-131), using a Tekmar Model LSC-2000 liquid sample con- centrator, interfaced with a Tekmar Model 2016 autosampler system, coupled to a Hewlett Packard 5890 Series II gas chromatograph. The chromatograph was equipped with a 30 m VOCOL capillary column, HP 3396A integrator/printer and flame ionization detector. Ultra pure carrier-grade helium gas was used for sparging samples. Bromoform standard was obtained from Ultra Scientific (North Kingstown, Rhode Island). The detection limit for bromoform was 5 ?g l?1 and standards were prepared to 200 ?g l?1. Biology. Culturable heterotrophic bacteria: Viable heterotrophic bacteria were quantified using a culture- based microbiological procedure. For enumeration, a 1 l sample from the Niskin sampler was placed in a sterile Nalgene plastic bottle and held on ice. Samples were transported on ice to the University of Washing- ton laboratory and maintained on ice until processed. Samples were processed within 24 h of collection. Numbers of culturable heterotrophic bacteria were determined on marine R2A agar. This medium is a modification of R2A agar (Difco), which is commonly recommended for freshwater samples (APHA 1998). We prepared the marine R2A agar using ONR sea- water salts. The formula for this marine salt solution at concentrations l?1 was: NaCl, 22.79 g; Na2SO4, 3.98 g; KCl, 0.72 g; NaBr, 0.083 g; NaHCO3, 0.031 g; H3BO3, 0.027 g; NaF, 0.0026 g; MgCl2 ? 6H2O, 1.12 g; CaCl2 ? 2H2O, 0.15 g; SrCl2 ? 6H2O, 0.0024 g; FeCl ? 4H2O, 0.0080 g. The salts were prepared in 3 separate solutions: a 10? solution of 7 compounds (NaCl, Na2SO4, KCl, NaBr, NaHCO3, H3BO3, NaF), a 50? solu- tion of the divalent compounds (MgCl2, CaCl2, SrCl2), and a 200? solution of FeCl. The 10? solution was mixed with R2A agar, pH was adjusted to 7.6, and the medium was sterilized by autoclaving at 121?C. The medium was cooled in a water bath to 50?C. Divalent cations solution (20.0 ml l?1 of a sterile 50? solution of MgCl2 ? 6H2O, CaCl2 ? 2H2O, SrCl2 ? 6H2O) and FeCl2 solution (5.0 ml l?1 of a sterile 200? solution of FeCl ? 4H2O) were added and mixed into the molten medium. The divalent cations and iron were added after autoclaving to minimize the formation of precipi- tate in the medium. Bacteria were enumerated using 2 methods. Aliquots of ballast water were inoculated onto the agar surface using the spread-plate method, or a larger volume of seawater was filtered through Pall Metricel? Black Membrane Disc Filters (47 mm diameter, 0.45 ?m pore size). Filters were placed on the surface of Marine R2A agar in a 50 mm diameter plas- tic petri plate. Filters were rolled onto the agar surface to prevent air bubbles from forming between the filter and agar surface. Larger 100 mm diameter petri dishes were used for the directly inoculated spread-plate method. Samples were inoculated in triplicate for each dilution, except for some filtered samples that were inoculated in duplicate. Inoculated media were incu- bated at room temperature (approximately 22?C) in the dark. Bacterial colonies were counted on the spread-plate agar surfaces and membrane filters after 4 d, when the colonies were large enough to see, but were not overlapping. Phytoplankton and microflagellates: Subsamples (1 l) from the Niskin sampler were preserved on board ship in Lugol?s iodine and shipped to the Smithsonian Environmental Research Center in Maryland for 40 Herwig et al.: Ozone treatment of ballast water analyses. In each subsample, the number of cells pre- sent for each phytoplankton and microflagellate spe- cies (or lowest taxonomic unit) was counted directly under a compound microscope. First, 200 individual cells were counted for each of 20 fields at 500? magni- fication; this provided data for the number of cells for small species (e.g. microflagellates and dinoflagel- lates). Second, 20 fields were also examined at 312? magnification, to estimate the number of larger and less numerous forms. To measure the effect of ozone treatment, changes in concentration (before and following 5 and 10 h of treat- ment) in the treatment and control tanks were com- pared. Counts were pooled across taxa for 3 major groups: dinoflagellates, microflagellates and diatoms. Species-level information was collected, but only effects on major taxonomic groups were compared, because there was high species composition variation both within replicates at a single collection time and among sampling periods. This higher taxa level approach was similar to the level of analysis for zoo- plankton and microbiology study components. Mesozooplankton: A 0.3 m diameter, 73 ?m mesh zooplankton net was used for zooplankton collections. The net was lowered through hatches into Treatment Columns A and B and Control Columns C and D, to within 0.25 m of the tank bottom and slowly retrieved to the surface. We took 3 replicate vertical hauls from each hatch before ozone treatment, after 5 h (all experiments), and after 10 h (Expts 2 and 3). Samples were gently washed with filtered seawater from the net collecting bucket into a plastic specimen jar and kept cool by placing the jar on ice. Samples were immediately examined on the ship under a dis- secting microscope. A field of view at 25? magni- fication was examined. Animal activity was scored as follows: animals moving or showing an escape re- sponse when probed with a fine needle (a 000 size insect pin mounted on a wooden stick), were scored as ?live?; those that were not mobile, but exhibited internal or external movement, were scored as ?mori- bund?; those with no internal or external movement were scored as ?dead?. Successive fields of view were examined until 100 organisms had been examined. In addition, qualitative observations were recorded about dominant taxa, and any taxa that appeared to be more or less affected by the treatment. Toxicology. Caged animals: Caged organism exper- iments were designed to evaluate the effect of ozone treatment on a range of aquatic organisms, some that are typically used in aquatic toxicology experiments. The organisms included: mysid shrimp Americamysis bahia, sheepshead minnows Cyprinodon variegatus, purple shore crabs Hemigrapsus nudus and amphi- pods Rhepoxynius abronius. They were chosen based on their known sensitivity or hardiness to a variety of aquatic toxicants and their use as ?standard? laboratory test organisms. Mysid shrimp and sheepshead min- nows were obtained from Aquatic Biosystems (Fort Collins, Colorado). Shore crabs and amphipods were collected from Puget Sound near Anacortes, Washing- ton. All organisms were acclimated to Puget Sound seawater and maintained under either static or flowing seawater conditions at Western Washington Univer- sity?s Shannon Point Marine Laboratory, Anacortes, Washington. Prior to testing, organisms were placed in individual exposure chambers and transported to the S/T Tonsina in ice chests containing aerated seawater. Animals and cages were not pre-selected for a treat- ment or control ballast tank. For the amphipods, 3 in situ chambers were put into a bucket containing sand at the bottom to act as an anchor, and were lowered to the bottom of the ballast tank (15 m). Each chamber contained 10 amphipods. Amphipod chambers were similar to those of Tucker & Burton (1999), consisting of 5 cm diameter clear plastic tubes approximately 12cm long enclosed at each end with polypropylene caps, with two 3 ? 5cm ports made of 1mm mesh. Chambers were soaked in both freshwater and seawater for 24 h to dissipate any construction-related toxicity. For mysid shrimp and sheepshead minnows, 10 individuals of each species were placed inside chambers (as de- scribed above) containing 2 rectangular windows (3 ? 5 cm) covered with 750 ?m mesh for mysid shrimp and 1 mm mesh for sheepshead minnows. For shore crabs, 10 individuals were placed into commercially available plastic crab bait buckets (11 cm high ? 9 cm diameter) that were drilled with numerous 8 mm holes. Groups of 3 chambers, 1 for each species, were placed in coarse- mesh polyethylene nets and attached to the tether rope with clamps, and deployed at specific depths in the ballast tank. Groups of caged organisms were placed into the control and treatment tanks. Each exposure group consisted of a plastic bucket containing sand and connected to a tether rope. Buckets with amphipod exposure chambers were at the bottom of the ballast tank, with chambers for the other 3 species being suspended from the tether rope at approximately 1, 6 and 12 m) from the ballast water surface. At the com- pletion of the 5 or 10 h ozone treatment, cages were removed and the number of live, moribund or dead organisms was recorded. Amphipods were classified as moribund if they failed to rebury in the sand. Whole effluent toxicity (WET) testing. Samples of ozone-treated water were collected at the end of each ozone treatment for laboratory toxicity testing of whole effluent toxicity (WET). We performed 2 standard acute toxicity tests: mysid shrimp Americamysis bahi 48 h static acute toxicity test, and topsmelt Atherinops 41 Mar Ecol Prog Ser 324: 37?55, 2006 affinis 48 h static acute toxicity test. These species are among the most sensitive to toxic chemicals in seawa- ter (Suter & Rosen 1988), and are commonly used to evaluate the toxicity of effluents discharged into marine waters. All toxicity tests were performed in accordance with standard regulatory procedures (US Environmental Protection Agency 1993, 1999). The seawater used as controls and for dilution of ballast water samples was prepared using laboratory water (1 ?m filtered) and commercially available seawater salts (Hawaiian Marine Mix). The seawater salinity was 30?2 PSU. Mysid shrimp were obtained from Aquatic Biosys- tems (Fort Collins, Colorado). We exposed 5-d-old mysids for 48 h in a static test to 5 dilutions of ozonated ballast water: 6.25, 12.5, 25, 50 and 100%, and to a dilution water control. A water temperature of 25 ? 1?C and a 16:8 h light:dark cycle were maintained. Test solutions were not aerated and mysid shrimp were not fed during the tests. We used 4 replicate test solutions containing 5 to 10 shrimp per chamber at each treat- ment level in all tests. Procedures for the topsmelt tests were similar to those for the mysid shrimp. We exposed 15-d-old topsmelt larvae obtained from Aquatic Biosystems for 48 h in a static test to 5 dilutions of ozonated ballast water samples: 6.25, 12.5, 25, 50 and 100% ballast water and to a dilution water control. RESULTS Chemistry Ozone delivery Table 1 summarizes the water volume capacity of both sections of the ozone treatment tank (No. 3 port ballast tank) and number of ozone diffusers in each section, as well as the calculated ozone-loading rate in each section for each of the 3 experiments. Note that the ?Port, vertical portion? row in Table 1 gives infor- mation pertaining to the vertical wing tank, that is the portion from which samples for our experiments were taken. The ozone-loading rate in this wing tank increased by 22% between Expts 1 and 2, and then by 87.5% between Expts 2 and 3. This increase in ozone loading is generally reflected in the chemical and bio- logical data presented below. Seawater chemistry The ballast water used in the experiments was col- lected by the S/T Tonsina in northern Puget Sound and in the Straits of Juan de Fuca, near the Pacific Ocean. Observed salinities varied by less than 1 PSU in each experiment. Salinities were 33.3 to 33.7 PSU in Expt 1, 35.0 to 35.9 PSU in Expt 2, and 33.9 to 34.4 PSU in Expt 3. Salinity did not change in any tank during the exper- iments. Water temperatures were slightly higher in Expt 1 than in the other 2 experiments. In Expt 1, water tem- peratures were 12.7 to 15.5?C, while in the November experiments the temperatures were 9.4 to 11.8?C. The temperature either remained the same or decreased slightly in all tanks. The pH for all samples was typical of seawater, rang- ing from 7.4 to 7.9. In Expt 1, the pH was not as pre- cisely measured as in the later experiments. There was no pH difference between the treatment and control tanks and it did not vary with time during the experi- ments. DO in the ballast water was relatively high at the beginning of the experiments, at ?8 mg l?1 in Expt 2 and >6 mg l?1 in Expt 3. DO was not measured in Expt 1. DO generally increased during ozonation and maxi- mum levels 2 to 3 times those of initial levels (approx. 20 mg l?1) were found at the end of the treatment. In the control tanks, the oxygen concentration did not increase during the experiment. DOC measured at the beginning of each experiment was similar among all the experiments, and for the 42 Experimental Volume No. of Diffuser density Ozone production Ozone distribution Ozone loading rate ballast tank (m3 ? 103) Diffusers (m3 diffuser?1) (g h?1) (%) (mg l?1 h?1) (No. 3) Expt Expt Expt 1 2 3 1 2 3 1 2 3 Port 3.11 72 4.32 ? 101 1460 1760 1660 0.47 0.56 0.53 Port, horizontal 1.88 56 3.36 ? 101 50 40 0 0.39 0.38 0.00 portion Port, vertical 1.23 16 7.69 ? 101 50 60 100 0.59 0.72 1.35 portion Table 1. Estimated ozone production, distribution and loading in the total, horizontal and vertical portions of treatment tank in Expts 1, 2, and 3 performed onboard S/T Tonsina Herwig et al.: Ozone treatment of ballast water experimental and control samples collected within each experiment. DOC concentrations ranged from 0.7 to 1.1 mg l?1. Phosphate ranged from 0.06 to 0.07 mg l?1, silicate from 1.3 to 1.5 mg l?1, nitrate from 0.2 to 0.4 mg l?1, and nitrite from 0.004 to 0.006 mg l?1; ammonium ranged from 0.03 mg l?1 in Expt 1 to about twice that concentration (0.07 mg l?1) in Expts 2 and 3. Ozone chemistry In each experiment, TRO and ORP increased during the period of ozonation in the treatment tank (Table 2). TRO and ORP increases were not as great in Expt 1 with 5 h of ozonation, as in Expts 2 and 3 with 10 h of ozonation. Some of the concentrations exceeded the capacity for the colorimetric assay, i.e. were greater than 5 mg l?1 measured as Br2. In Expt 1, the highest TRO level found (0.26 mg l?1) was in the A15 location (Column A, 15 m from surface) at 5 h. The highest TRO in Column B was approximately one-half this value. In Expts 2 and 3, the TRO levels exceeded 5 mg l?1 follow- ing 7.5 to 10 h of ozonation. The TRO levels increased more quickly in Column A than in Column B in the treatment tank. TRO levels were near 0.0 mg l?1 for all samples collected in the control ballast tank. The TRO achieved is a product of the ozone loading rate (Table 1) and the length of time that a column of seawater was treated. As described above, the ozone loading rate was highest in Expt 3 and lowest in Expt 1. Samples were not collected from Column D in the control bal- last tank during Expts 2 and 3. Seawater in the ballast tanks was well oxygenated at the start of the study and had very positive ORP values, measured as mV. Following ozonation, the ORP values rapidly increased from approximately 100 mV to over 600 mV. Maximum ORP values were between 780 and 799 mV. ORP values in the control tank fluctuated between 97 and 439 mV with a mean of 260 mV. 43 Location Sample time Total residual oxidant (mg Br2 l?1) Oxidation reduction potential (mV) (h) Expt Expt 1 2 3 1 2 3 A3 0.0 0.00, 0.00 0.06, 0.07 0.07, 0.01 129.5 077.1 71.6 2.5 0.21, 0.43 2.74, 2.80 4.02, 4.07 372.4 725.1 767.3 5.0 0.23, 0.26 2.39, 2.37 >5.00, >5.00 718.9 774.3 761.6 7.5 ns >5.00 >5.00, >5.00 ns 781.7 782.1 10.0 ns >5.00, >5.00 >5.00, >5.00 ns 789.5 794.9 A9 0.0 0.00, 0.00 0.06, 0.04 0.02, 0.02 140.2 069.4 75.6 2.5 0.08, 0.00 2.70, 2.78 3.62, 3.77 363.7 738.3 750.7 5.0 0.20, 0.03 2.84, 2.15 >5.00, >5.00 738.6 782.6 785.1 7.5 ns >5.00, >5.00 >5.00, >5.00 ns 793.2 791.7 10.0 ns >5.00, >5.00 >5.00, >5.00 ns 796.4 788.2 A15 0.0 0.00, 0.00 0.06, 0.05 0.00, 0.01 136.8 072.5 95.7 2.5 0.15, 0.14 0.37, 0.39 0.32, 0.31 289.7 629.3 574.8 5.0 0.24, 0.26 2.42, 2.39 2.68, 2.72 753.0 792.0 713.9 7.5 ns 4.70, 4.62 4.53, 4.80 ns 787.4 785.5 10.0 ns >5.00, >5.00 >5.00, >5.00 ns 797.5 793.2 B3 0.0 0.00, 0.01 0.02, 0.00 0.00, 0.00 115.7 074.3 89.3 2.5 0.00, 0.00 0.57, 0.56 0.70, 0.59 217.0 297.1 637.5 5.0 0.10, 0.01 >5.00, >5.00 2.90, 3.80 385.7 748.2 754.2 7.5 ns 3.89, 3.94 4.83, 4.72 ns 774.7 781.4 10.0 ns >5.00, >5.00 >5.00, >5.00 ns 784.7 793.2 B9 0.0 0.05, 0.02 0.01, 0.00 0.00, 0.01 144.6 077.0 92.6 2.5 0.03, 0.01 0.85, 0.84 1.00, 1.08 217.3 981.0 721.1 5.0 0.02, 0.00 >5.00, >5.00 3.98, 3.96 506.6 765.6 774.6 7.5 ns 4.40, 4.37 >5.00, >5.00 ns 776.2 786.3 10.0 ns >5.00, >5.00 >5.00, >5.00 ns 785.5 798.7 B15 0.0 0.01, 0.03 0.00, 0.03 0.00, 0.00 162.2 075.7 95.8 2.5 0.01, 0.00 0.63, 0.61 0.96, 1.04 339.9 672.3 716.9 5.0 0.09, 0.17 >5.00, >5.00 4.14, 4.12 495.6 762.6 772.9 7.5 ns 3.91, 3.96 >5.00, >5.00 ns 779.0 790.9 10.0 ns >5.00, >5.00 >5.00, >5.00 ns 793.9 799.0 Table 2. Total residual oxidant (TRO) and oxidation reduction potential (ORP) of treated and control ballast tanks in Expts 1, 2 and 3 performed onboard S/T Tonsina. Location: letter represents column in ballast tanks, number represents distance (m) from ballast tank surface. Where 2 values are given, these are for duplicate analyses. >5.00 (TRO > 5.0 mg Br2 l?1) = out of range for assay, ns = not sampled Mar Ecol Prog Ser 324: 37?55, 2006 Disinfection byproduct chemistry We analysed 2 disinfection byproducts of possible concern, bromate and bromoform. Bromate was always below the method detection limit in all samples. When bromate was spiked into the treated samples in the laboratory, the spike was never re- covered fully, indicating bromate demand in the water. The cause of this apparent demand was not under- stood; however, it may have been related to the high concentration of ?active? bromine (i.e. HOBr/OBr?) in the samples. In all 3 experiments, bromoform concentration increased over the ozonation period, with the maxi- mum found at the last sampling (Table 3). Where a direct comparison was possible (i.e. from one experi- ment to another at the same time point), it was clear that the concentration of bromoform increased more in Expt 1 than in either Expts 2 or 3, particularly in sam- ples collected from Column A. In Expt 1, the maximum concentration of bromoform found was 145 ?g l?1, in Expt 2 it was 98 ?g l?1, and in Expt 3 it was 107 ?g l?1. In Expt 3, the quantities of bromoform were very comparable between Columns A and B. Biology Culturable heterotrophic bacteria The number of culturable bacteria was determined using either the direct spread-plate method or the membrane filtration method for each sample. The numbers presented (Table 4) were selected from the 44 Colony forming units (CFU) l?1 Location Time Expt (h) 1 2 3 A3 0.0 4.70 ? 106 1.30 ? 106 4.10 ? 105 2.5 1.00 ? 104 1.00 ? 101 1.00 ? 101 5.0 <3.00 ? 103 4.00 ? 101 5.00 ? 100 * 7.5 ns <3.00 ? 100 5.00 ? 100 * 10.0 ns <3.00 ? 100 <5.00 ? 100 * A9 0.0 2.70 ? 106 9.20 ? 105 2.40 ? 105 2.5 3.00 ? 103 3.00 ? 101 7.00 ? 100 5.0 <3.00 ? 103 3.00 ? 100 <5.00 ? 100 * 7.5 ns 3.00 ? 100 <5.00 ? 100 * 10.0 ns <3.00 ? 100 5.00 ? 100 * A15 0.0 2.30 ? 106 9.30 ? 105 3.20 ? 105 2.5 <3.00 ? 103 5.80 ? 102 6.00 ? 102 5.0 <3.00 ? 103 <3.00 ? 100 2.00 ? 101 * 7.5 ns 1.00 ? 101 <5.00 ? 100 * 10.0 ns <3.00 ? 100 5.00 ? 100 * B3 0.0 1.64 ? 107 9.40 ? 105 3.60 ? 105 2.5 1.09 ? 106 9.00 ? 102 1.20 ? 103 5.0 3.00 ? 103 4.00 ? 101 5.00 ? 100 * 7.5 ns 1.00 ? 101 <5.00 ? 100 * 10.0 ns 1.00 ? 101 <5.00 ? 100 * B9 0.0 3.20 ? 106 8.70 ? 105 3.20 ? 105 2.5 6.40 ? 105 5.00 ? 102 1.30 ? 103 5.0 <3.00 ? 103 3.00 ? 101 7.00 ? 100 7.5 ns <3.00 ? 100 5.00 ? 100 * 10.0 ns <3.00 ? 100 <5.00 ? 100 * B15 0.0 1.10 ? 106 8.50 ? 105 5.20 ? 105 2.5 2.40 ? 105 3.00 ? 102 1.10 ? 103 5.0 3.00 ? 103 4.00 ? 101 7.00 ? 100 7.5 ns 1.00 ? 101 5.00 ? 100 * 10.0 ns <3.00 ? 100 <5.00 ? 100 * Table 4. Enumerations of culturable heterotrophic bacteria from treated and control ballast tanks in Expts 1, 2, and 3 performed onboard S/T Tonsina. Location: letter repre- sents column in ballast tanks, number represents distance (m) from ballast tank surface. *Sample enumerated in duplicate; other samples were enumerated in triplicate. ns = not sampled Location Sample Bromoform (?g l?1) Time Expt (h) 1 2 3 A3 0.0 <5.0 <5.0 <5.0 2.5 35.0 62.0 74.6 5.0 136.0 77.4 77.7 7.5 ns 91.2 93.0 10.0 ns 92.2 90.1 A9 0.0 <5.0 <5.0 <5.0 2.5 30.0 68.4 80.0 5.0 145.0 76.0 90.3 7.5 ns 94.0 94.7 10.0 ns 98.0 105.6 A15 0.0 <5.0 <5.0 <5.0 2.5 104.0 35.1 29.3 5.0 ns 75.2 75.2 7.5 ns 80.3 94.6 10.0 ns 82.4 96.1 B3 0.0 <5.0 <5.0 <5.0 2.5 <5.0 32.9 42.5 5.0 24.0 53.8 73.7 7.5 ns 73.6 96.5 10.0 ns 76.1 107.0 B9 0.0 <5.0 <5.0 <5.0 2.5 <5.0 44.6 55.5 5.0 47.2 70.4 70.6 7.5 ns 75.7 96.5 10.0 ns 83.0 103.0 B15 0.0 <5.0 <5.0 <5.0 2.5 <5.0 40.4 46.2 5.0 35.8 58.7 87.1 7.5 ns 74.8 79.0 10.0 ns 79.4 105.0 Table 3. Bromoform data in ballast tank treated with ozone (Columns A and B of ballast tanks). All samples collected from control ballast tank (Columns C and D) were below method detection limit. Location: letter represents column in ballast tanks, number represents distance (m) from ballast tank surface. <5.0 (bromoform <5.0 ?g l?1) = below method detection limit, ns = not sampled Herwig et al.: Ozone treatment of ballast water method that provided the best range of countable colonies for the sample. For example, for the ozonated seawater samples, the ozone treatment method was very effective in inactivating culturable heterotrophic bacteria. Therefore, if 100 ?l aliquots of treated seawa- ter were inoculated onto the surface of marine R2A agar by the spread-plate method, typically no colonies would be found. Therefore, the culturable microorgan- isms were concentrated by using a membrane filtration method so that the sensitivity of the enumeration assay could be increased. The numbers in Table 4 are an average of the plating performed in triplicate or dupli- cate for each aliquot. The number of culturable microorganisms was between 105 and 106 CFU l?1 before ozonation in treat- ment tanks and throughout the experiments in the con- trol tanks (Table 4). A few samples had higher levels. With ozonation, the number of viable bacteria had declined by the first 2.5 h sample. Except for the A15 sample collected in Expt 1, all of the 2.5 h samples col- lected from the treated tank in Expts 2 and 3 had lower numbers of culturable bacteria than those from Expt 1. After 10 h of treatment (Expts 2 and 3), the bac- teria population in the treated tank had decreased to ? 5.0 CFU l?1. One-third of the samples contained levels below the experimental detection limits (3.0 and 5.0 CFU l?1 for Expts 2 and 3, respectively). Samples collected at 7.5 and 10 h contained few if any viable bacterial cells. Therefore, ozonation using our methods reduced culturable microorganisms > 99.99%. One-third of the samples collected after 5.0 h of ozonation contained levels below the experimental detection limits (3.0 and 5.0 CFU l?1 for Expts 2 and 3, respectively. Samples collected at 7.5 and 10 h con- tained few, if any, viable bacterial cells. Therefore, ozonation using our methods reduced the culturable microorganisms by >99.99%. Phytoplankton and microflagellates Since cells were preserved on board ship, our method could not be used to determine viability. Instead, dinoflagellate, diatom and microflagellate densities were estimated, assuming that decreases were due to cell death. During both Expts 2 and 3 when they were collected, dinoflagellate populations exhibited sharp decreases in both columns (A and B) in the ozone treatment tank relative to Column C in the control tank (Table 5). In Expt 2, samples collected 10 h after ozone treatment were reduced by 82 to 100% in Column A (with concentration increasing with increas- ing depth) and by 100% in Column B. For Expt 3, dinoflagellates were not detected after the 10 h treat- ment, resulting in a >99% reduction. In contrast, dinoflagellates did not decline in any of the control tanks. Initial concentration of microflagellates ranged from 2 ? 105 to 3 ? 105 cells l?1 in the treatment tank. Similar to dinoflagellates, microflagellates declined between 70 and 99% in Column A and between 93 and 98% in Column B during Expt 2. No spatial variation was evi- dent in Expt 3, and microflagellates declined by 96 to 99%. The initial concentration of microflagellates ranged from 2 ? 105 to 4 ? 105 cells l?1 in the control tank. In Expts 2 and 3, microflagellates did not decline in the control tank. Diatom results were more variable. Concentrations varied from 17 to 135% of the initial concentrations in Expt 2 and from 20 to 120% of the initial concentra- tions in Expt 3 after 10 h ozonation. No decline was observed in the control tank. 45 Colony forming units (CFU) l?1 Location Time Expt (h) 1 2 3 C3 ? control 0.0 2.30 ? 106 1.10 ? 106 7.00 ? 105 2.5 1.10 ? 106 3.70 ? 107 6.40 ? 105 5.0 6.00 ? 105 8.40 ?105 7.20 ? 105 7.5 ns 7.90 ?105 6.70 ? 105 10.0 ns 7.60 ? 105 6.20 ? 105 C9 ? control 0.0 1.70 ? 106 7.70 ? 105 2.30 ? 105 2.5 9.00 ? 105 3.30 ? 107 6.60 ? 105 5.0 8.00 ? 105 7.90 ? 105 5.70 ? 105 7.5 ns 7.70 ?105 6.00 ? 105 10.0 ns 7.40 ? 105 6.30 ? 105 C15 ? control 0.0 9.00 ? 105 7.60 ? 105 3.20 ? 105 2.5 7.00 ? 105 8.70 ? 105 7.40 ? 105 5.0 5.00 ? 105 8.90 ? 105 6.60 ? 105 7.5 ns 7.80 ? 105 6.70 ? 105 10.0 ns 8.80 ? 105 7.60 ? 105 D3 ? control 0.0 9.00 ? 105 ns ns 2.5 7.00 ? 105 ns ns 5.0 8.00 ? 105 ns ns 7.5 ns ns ns 10.0 ns ns ns D9 ? control 0.0 8.00 ? 105 ns ns 2.5 5.00 ? 105 ns ns 5.0 6.00 ? 105 ns ns 7.5 ns ns ns 10.0 ns ns ns D15 ? control 0.0 5.00 ? 105 ns ns 2.5 5.00 ? 105 ns ns 5.0 4.00 ? 105 ns ns 7.5 ns ns ns 10.0 ns ns ns Table 4 (continued) Mar Ecol Prog Ser 324: 37?55, 2006 Mesozooplankton In the 5 h ozone exposure (Expt 1), the average percent of animals alive was uniformly high (range 94 to 97%) in pre-treatment samples (Table 6). Mor- tality after 5 h was 91% in Column A and 47% in Column B. The zooplankton assemblage in Expt 1 was domi- nated by the calanoid copepod Paracalanus sp., but also had several other relatively numerous copepods and larvae of barnacles, polychaetes and other ani- mals. In qualitative observations, 2 taxa ? the cyclo- poid copepod Corycaeus anglicus and large Cirripedia (barnacle) nauplii ? appeared relatively unaffected after 5 h ozone treatment. On the other hand, small calanoid copepod nauplii larvae were observed to have higher mortality than other mesozooplankton. Similarly, in Expt 2, mortality at 5 h was different between the 2 treatments (A and B) (Table 6). In con- trast to the Expt 1 conducted in September, survival was higher in Column A than in Column B. In addition, 5 h mortality was lower than in Expt 1 (20% in Column A, 66% in Column B). After 10 h treatment, mortality increased, but the difference between the treatment columns persisted. In Expt 3, mortality differences between the 2 treat- ment columns were much less and mortality was much higher at both sampling times (5 and 10 h) than in the other experiments (Table 6); >96% of the mesozoo- plankton were killed by 10 h. In Expts 2 and 3 conducted in November, diversity was much lower than in September (Expt 1). As in Expt 1, the zooplankton assemblage was dominated by the calanoid copepod Paracalanus sp. (mostly juveniles), but there were far fewer of the other taxa. Interestingly, in Expt 2 we observed a few Pseudodiaptomus marinus, an exotic Asian calanoid copepod, in most samples. As this species was not found in plankton tows from the ballast source water taken in Port Angeles harbor both day and night, it was assumed that they represented ballast water remnants from the ship?s last voyage to Long Beach harbor, where P. marinus is established. 46 Location Time Expt 2 (cells l?1) Expt 3 (cells l?1) (h) Dinoflagellates Microflagellates Diatoms Dinoflagellates Microflagellates Diatoms A3 0.0 1.79 ? 104 2.71 ? 105 1.11 ? 105 1.08 ? 104 2.96 ? 105 1.24 ? 105 5.0 9.62 ? 102 3.30 ? 104 1.76 ? 105 0 1.20 ? 104 1.28 ? 105 10.0 0 3.16 ? 103 3.71 ? 104 0 6.44 ? 103 6.95 ? 104 A9 0.0 5.92 ? 103 2.70 ? 105 8.34 ? 104 1.20 ? 104 2.57 ? 105 6.74 ? 104 5.0 1.02 ? 103 1.65 ? 104 5.66 ? 104 0 4.76 ? 103 6.77 ? 104 10.0 8.80 ? 102 8.10 ? 104 1.13 ? 105 0 9.06 ? 103 1.25 ? 105 A15 0.0 4.82 ? 103 1.85 ? 105 1.24 ? 105 1.13 ? 104 3.06 ? 105 9.15 ? 104 5.0 1.86 ? 103 2.44 ? 104 3.84 ? 104 1.03 ? 103 9.24 ? 103 6.88 ? 104 10.0 8.76 ? 102 3.54 ? 103 2.12 ? 104 0 1.83 ? 104 1.82 ? 104 B3 0.0 1.27 ? 104 2.84 ? 105 1.52 ? 105 1.10 ? 104 2.63 ? 105 5.34 ? 104 5.0 0 1.06 ? 104 5.82 ? 104 1.20 ? 103 1.58 ? 104 1.01 ? 105 10.0 0 1.97 ? 104 2.94 ? 104 0 5.89 ? 103 6.84 ? 104 B9 0.0 1.20 ? 104 1.90 ? 105 1.35 ? 105 2.86 ? 104 3.24 ? 105 8.84 ? 104 5.0 0 3.10 ? 105 2.50 ? 105 1.07 ? 103 3.21 ? 103 6.01 ? 104 10.0 nd nd nd 0 8.86 ? 103 5.86 ? 104 B15 0.0 1.02 ? 104 2.67 ? 105 3.27 ? 104 8.47 ? 103 2.36 ? 105 7.57 ? 104 5.0 0 2.05 ? 104 6.25 ? 104 2.32 ? 103 2.35 ? 104 5.04 ? 104 10.0 0 6.21 ? 103 1.62 ? 104 0 3.14 ? 103 4.49 ? 104 C3 ? control 0.0 4.28 ? 103 1.73 ? 105 1.51 ? 105 9.91 ? 103 3.30 ? 105 9.78 ? 104 5.0 4.92 ? 103 3.04 ? 105 8.00 ? 104 1.35 ? 104 2.82 ? 105 6.74 ? 104 10.0 8.35 ? 103 2.34 ? 105 5.98 ? 104 1.85 ? 104 2.61 ? 105 1.25 ? 105 C9 ? control 0.0 1.88 ? 104 3.10 ? 105 9.79 ? 104 1.19 ? 104 2.82 ? 105 8.67 ? 104 5.0 6.99 ? 103 2.16 ? 105 7.66 ? 104 9.18 ? 103 2.79 ? 105 1.33 ? 105 10.0 1.44 ? 104 2.47 ? 105 1.06 ? 105 3.76 ? 104 4.38 ? 105 8.87 ? 104 C15 ? control 0.0 5.74 ? 103 2.26 ? 105 1.20 ? 105 8.12 ? 103 2.98 ? 105 9.37 ? 104 5.0 5.52 ? 103 2.47 ? 105 7.88 ? 104 6.05 ? 104 4.14 ? 105 1.11 ? 105 10.0 9.95 ? 103 1.93 ? 105 1.69 ? 105 5.70 ? 103 3.77 ? 105 1.46 ? 105 Table 5. Total numbers of dinoflagellates, diatoms, and microflagellates found in Expts 2 and 3. Samples were preserved onboard S/T Tonsina and later enumerated in laboratory. Location: letter represents column in ballast tank, number represents distance (m) from ballast tank surface. nd = no data Herwig et al.: Ozone treatment of ballast water This species, the harpacticoid copepod Microsetella sp. and nematode worms appeared to be more resistant to ozone treatment than the dominant Paracalanus sp., although overall mortality was very high. Toxicology Caged animals Different mortalities were observed for different species of caged animals. For organisms suspended in the water column, sheepshead minnows usually had greatest mortalities, shore crabs least, and mysid shrimp intermediate mortalities. There were also dif- ferences between experiments and within a given treatment tank. In Expt 1, the organisms were exposed in 2 ozonated columns in 1 ballast tank and in 2 control columns in the control tank for 5 h. Survival of control organisms was almost 100% (only 1 of 30 amphipods died, but 3 exposure chambers in Column C of the treatment tank were unfortunately lost) (Table 7); survival was also 100% for all species in Column B. In treatment Column A, dead and moribund mysid shrimp ranged from 80 to 100%; moribund and dead sheepshead minnows ranged from 80 to 100%. Sur- vival for both these species was directly related to depth: those closest to the bottom and nearest the ozone diffusers suffered highest mortality. In Column A, most of the mysid shrimp and sheepshead minnows died, except for mysid shrimp at the 1 m depth (Table7). In Expt 2, test organisms were exposed to a 10 h ozonation period in 3 treatment columns (A, AB ? mid- way between A and B ? and B) and in 1 control column (D). Control survival was almost 100% for all species (1 of 30 mysid shrimp died), and none showed adverse effects (Table 7). For mysid shrimp in Column A of the treatment tank, percent dead varied (10 to 100%). In contrast, 100% of the sheepshead minnows died and 100% of the shore crabs and amphipods lived. The pattern in the middle of the tank (Column AB) was somewhat similar, with 100% mortality for sheepshead minnows, 50 to 100% for mysid shrimp and 100% for shore crabs (except those nearest the ballast surface). In Expt 2, survival of all animals was greatest in Column B, with sheepshead minnows again having the greatest mortality. In Expt 2, all shore crabs survived and amphipods had only slight mortality (Table 7). In Expt 3, all test organisms were exposed to ozone for 10 h in Columns A, AB, and B and to control condi- tions in 1 column (D). Control survival for this experi- ment was 100% and none of the control animals showed signs of stress (Table 7). Highest mortalities were observed in the treatment tank in Expt 3. Sheepshead minnows had 100% mortality for all columns and depths; mysid shrimp, 100% for all sam- ples collected in Column AB, 100% for 2 of the depths in Column A and 1 depth in Column B; shore crabs, 0% (but the shore crabs were moribund); amphipods, 7%. All surviving shore crabs in the treatment tank were sluggish, and classified as moribund. In Expt 3, there was no obvious trend in survival rates with depth (although only mysid shrimp had partial kills, so data for this type of comparison was sparse). As in Expts 1 and 2, survival of mysid shrimp was highest in Column A, and amphipods had only slight mortality (Table 7). WET (acute toxicity) testing Tests conducted on mysid shrimp and topsmelt with control water samples (i.e. non-ozonated ballast water from the S/T Tonsina) exhibited no or minimal toxicity (i.e. <10% mortality) in all tests. For mysid shrimp, median lethal concentrations ranged from approxi- 47 Location Time % alive % moribund % dead (h) avg. SD avg. SD avg. SD Expt 1 A 0.0 93.7 0.6 5.7 1.5 0.7 1.2 B 0.0 95.3 1.2 3.7 2.1 1.0 1.0 C ? control 0.0 97.0 2.0 1.7 0.6 1.3 1.5 D ? control 0.0 95.7 1.5 3.0 1.0 0.3 0.6 A 5.0 1.7 0.6 7.3 3.1 91.0 3.0 B 5.0 25.0 4.0 27.7 0.6 47.3 3.5 C ? control 5.0 92.3 1.5 5.3 2.3 2.3 1.5 D ? control 5.0 92.7 2.9 6.0 2.6 1.3 0.6 Expt 2 A 0.0 96.3 1.2 3.0 0.0 0.7 1.2 B 0.0 93.7 1.5 4.0 1.7 0.3 0.6 C ? control 0.0 97.3 2.1 1.3 1.2 1.7 2.1 A 5.0 40.3 3.2 39.7 8.5 20.0 6.2 B 5.0 13.7 2.5 20.0 6.0 66.3 8.5 C ? control 5.0 97.7 1.5 2.3 1.5 0.0 0.0 A 10.0 13.7 1.5 19.3 8.7 67.0 9.6 B 10.0 1.7 1.2 1.0 1.0 97.3 2.1 C ? control 10.0 94.3 3.8 5.0 3.6 0.7 0.6 Expt 3 A 0.0 89.7 7.0 6.0 2.6 7.7 6.8 B 0.0 94.7 2.5 2.3 1.5 3.0 1.0 C ? control 0.0 93.3 4.0 3.7 0.6 3.0 3.6 A 5.0 7.7 5.7 8.3 4.2 84.0 7.0 B 5.0 1.7 1.2 6.0 2.0 92.3 3.1 C ? control 5.0 97.0 1.0 1.0 1.0 3.3 1.2 A 10.0 1.3 2.3 2.0 2.0 96.7 3.1 B 10.0 0.0 0.0 0.7 1.2 99.3 1.2 C ? control 10.0 93.3 1.5 2.3 0.6 4.3 1.5 Table 6. Effect of ozone treatment on mesozooplankton in Expts 1, 2 and 3 (n = 3). Location: letter represents column in ballast tank Mar Ecol Prog Ser 324: 37?55, 200648 Location Mysid shrimp Sheepshead minnows Shore crabs Amphipods (avg. of 3 cages) Live Moribund Dead Live Moribund Dead Live Moribund Dead Live Moribund Dead Expt 1 A1 10 50 40 20 10 70 100 0 0 A6 0 40 60 0 20 80 100 0 0 A12 20 0 80 0 0 100 100 0 0 A15 100 0 0 B1 100 0 0 100 0 0 100 0 0 B6 100 0 0 100 0 0 100 0 0 B12 100 0 0 100 0 0 100 0 0 B15 100 0 0 C1 ? control 100 0 0 100 0 0 100 0 0 C6 ? control 100 0 0 100 0 0 100 0 0 C12 ? control 100 0 0 100 0 0 100 0 0 C15 ? control nd nd nd D1 ? control 100 0 0 100 0 0 100 0 0 D6 ? control 100 0 0 100 0 0 100 0 0 D12 ? control 100 0 0 100 0 0 100 0 0 D15 ? control 97 0 3 Expt 2 A1 0 0 100 0 0 100 100 0 0 A6 27 0 73 0 0 100 100 0 0 A12 90 0 10 0 0 100 100 0 0 A15 100 0 0 AB1 40 0 60 0 0 100 10 0 90 AB6 0 0 100 0 0 100 100 0 0 AB12 50 0 50 0 0 100 100 0 0 AB15 80 0 20 B1 56 22 22 0 30 70 100 0 0 B6 63 25 12 0 0 100 100 0 0 B12 80 10 10 20 60 20 100 0 0 B15 97 0 3 D1 ? control 100 0 0 100 0 0 100 0 0 D6 ? control 100 0 0 100 0 0 100 0 0 D12 ? control 90 0 10 100 0 0 100 0 0 D15 ? control 100 0 0 Expt 3 A1 100 0 0 0 0 100 0 100 0 A6 0 0 100 0 0 100 0 100 0 A12 0 0 100 0 0 100 0 100 0 A15 93 0 7 AB1 0 0 100 0 0 100 0 100 0 AB6 0 0 100 0 0 100 0 100 0 AB12 0 0 100 0 0 100 0 100 0 AB15 93 0 7 B1 0 0 100 0 0 100 0 100 0 B6 0 80 20 0 0 100 0 100 0 B12 100 0 0 0 0 100 0 100 0 B15 93 0 7 D1 ? control 100 0 0 100 0 0 100 0 0 D6 ? control 100 0 0 100 0 0 100 0 0 D12 ? control 100 0 0 100 0 0 100 0 0 D15 ? control 100 0 0 Table 7. Mysid shrimp Americamysis bahia, sheepshead minnows Cyprinodon variegatus, shore crabs Hemigrapsus nudus, and amphipods Rhepoxynius abronius. Percentage live and moribund in Expt 1 following 5 h ozonation and in Expts 2 and 3 following 10 h ozonation. Location: letter represents column in ballast tanks (A, B, AB, C, D) number represents distance (m) from ballast tank surface. Column AB was located between Columns A and B in treatment tank. nd = no data Herwig et al.: Ozone treatment of ballast water mately 50 to 70% in ozone-treated water. Topsmelt were slightly more sensitive, with median lethal concentrations ranging from approximately 30 to 80% in ozone-treated water (Table 8). Treated water from Expt 3 appeared more toxic (i.e. had lower median lethal concentrations) than in the other experiments. DISCUSSION The impact of ballast water treatments on the chem- istry, biology and toxicity of the water must be under- stood before a potential treatment will be accepted by the shipping industry, regulatory agencies and other stakeholders. The expectations and regulatory environ- ment for ballast water treatment are still being devel- oped at the state, federal and international level. Re- sults of only a few ballast water treatment systems on ships are documented in peer-reviewed literature. A few additional publications describe results of full-scale treatment systems that were evaluated at test bed facil- ities, but these are generally not as comprehensive as our shipboard study. Since we conducted our shipboard sampling in 2001, international and national treatment standards have been proposed and will probably in- fluence future methods used in ballast water treatment experiments. Chemistry Water chemistry was homogeneous and similar between the treated and control tanks, and with depth in each water column before the ozone treat- ment began. Following ozonation, DO increased. This increase would probably not be detrimental to locations where ballast water is discharged, but it could accelerate corrosion of steel in ballast tanks where tank coating is deteriorated and steel is exposed. Deoxygenation is suggested as a method for reducing corrosion (Tamburri et al. 2002) and for eliminating organisms in ballast water (see later subsection). Initially it was thought that the ozone itself would be the primary biocidal agent. However, it became apparent that bromine (HOBr/OBr?) resulting from the rapid reaction of ozone with bromide ion was the effective oxidant. Bromine is known as an excellent biocide with residual properties, that is, it remains in solution for an extended time (Johnson & Overby 1971, Crecelius 1979). This attribute is important for ballast water treatment that is performed in ballast tanks during a voyage because bromine can pre- clude the rebound of organisms with high reproduc- tive potential. It will be important to monitor the fate of biocides in ballast water treatments. In the case of ozonation, both ORP and TRO were considered. Results showed that TRO increased with increasing ozonation time, but ORP increased initially and then approached a ma- ximum value that was nearly invariant with time of ozonation. These results coupled with the mainte- nance requirements of ORP electrodes due to tank intermittent dry/wet cycles, led to elimination of ORP as a monitoring tool. Measuring TRO is a simple and standard method, with well-developed field test proce- dures. Testing TRO is also fast, reliable, and relatively inexpensive. It is often used to control and monitor disinfection processes in wastewater treatment. Auto- mated flow-through analyzers could be used for feed- forward or feedback control of ozone dosage, and could be incorporated in the initial design of future ballast water treatment systems. Our TRO chemistry and biological results indicated that the diffusers did not homogeneously distribute ozone throughout the ballast tank on the ship, either vertically or horizontally. Heterogeneity of the ozone distribution was a significant problem with the proto- type treatment system. One reason is that ballast tanks within large ships have a significant amount of internal structure and platforms. These structures are designed to strengthen the hull of the ship and provide baffling so ballast water movement is minimized within the tank. When ballast water was treated by bubbling 49 Exposure conc. % Survival (% ballast water) Expt 1 Expt 2 Expt 3 Mysid shrimp Americamysis bahia 0 100 90 97.5 6.25 95 100 97.5 12.5 100 95 97.5 25 100 95 75 50 100 95 0 100 0 0 0 EC50 (95% CI) 70.4 70.7 49.5 (69.5?71.3) (50.0?100) (27.0?37.7) Topsmelt Atherinops affinis 0 76a 100 100 6.25 80 95 95 12.5 88 100 100 25 92 100 80 50 100 47.5 0 100 20 7.5 0 LC50 (95% CI) 78.4 55.4 30.8 (71.1?86.5) (47.8?63.1) (28.1?33.9) aSurvival below minimum criteria for acceptable control survival Table 8. Mysid shrimp Americamysis bahia and topsmelt Atherinops affinis. Survival and median lethal concentration (EC50/LC50 as % ballast water) in acute toxicity WET (whole effluent treatment) tests with samples from Expts 1,2 and 3. Samples collected following ozonation Mar Ecol Prog Ser 324: 37?55, 2006 ozone from the diffusers, treated water did not easily circulate and mix within the treatment tank. Alterna- tive methods of injecting ozone should be explored to provide a more homogeneous distribution of oxidant and biocide in ballast water tanks. In future research, some members of our research team will examine the efficacy of an ozone treatment system that injects ozone through a venturi installed in-line with the bal- last pump pipe. Disinfection byproduct chemistry In this study, we examined the formation of 2 disin- fection byproducts, bromate and bromoform (Fig. 1). Bromate was not detected in any samples. The pres- ence of bromoform provided evidence that the oxidant residual was bromine, HOBr/OBr?. Bromoform was formed in all 3 shipboard experiments, but was found in greatest concentration in Expt 1. Two major factors that may affect the creation of bromoform are DOC (part of NOM [natural organic matter] presented in Fig. 1) and temperature (Garcia-Villanova et al. 1997, Abd El-Shafy & Grunwald 2000, Nikolaou & Lek- kas 2001). For all 3 experiments, the DOC was near 1 mg l?1 and differences were therefore probably not related to DOC concentrations. However, the tempera- ture in Expt 1 was different than in Expts 2 and 3, and this may be the reason for the lower concentration of bromoform in the November experiments. When bro- mate was spiked into the treated samples in the labo- ratory, the spike was never recovered fully. The cause of this apparent demand is unknown, but may be related to the high concentration of ?active? bromine (i.e. HOBr/ OBr?) in the samples. If TRO remains in seawater, it is likely that ozonated ballast water will continue to increase in bromoform concentration. A literature review suggests that the levels of bromoform found in our study will not adversely affect marine organisms. Toxicity data are available for phytoplankton Skeletonema costatum, Thalassiosira pseudonana, Glenodinium halli and Isochrysis galbana (Erickson & Freeman 1978), mysid shrimp Americamysis bahia (US Environmental Pro- tection Agency 1978), brown shrimp Penaeus aztecus (Andersen et al. 1979), Atlantic menhaden Brevoortia tyrannus (Andersen et al. 1979), and sheepshead min- now Cyprinodon variegates (Heitmuller et al. 1981, Ward 1981). Data for these species suggest that the quantity of bromoform produced during our shipboard experiments was not acutely toxic with IC50 (50% inhi- bition concentration), LC50 (50% lethal concentration), or NOEC (no observed effect concentration) values 1 to 2 orders of magnitude higher than the quantities we observed. Biology Mechanisms for removal and inactivation of organisms Ozonation of seawater injures, kills, or lyses cells through the interaction of ozone or the residual oxidant (hypobromous acid and sodium hypobromite) with mol- ecules within and on the surface of cells. For microor- ganisms, interaction with a significant level of an oxi- dant may cause the lysis of the cell. In other studies (data not shown), we observed a decrease in total num- bers of microorganisms in seawater samples exposed to ozone that were stained with a nucleic acid stain and examined by epifluorescence microscopy or flow cy- tometry. Vissers et al. (1998) explained the mechanism of human red cell lysis by hypobromous acid by stating that it reacts with membrane lipids and proteins. Ozone has been used for many years to maintain water quality in seawater aquaculture settings. A low level of resid- ual ozone is beneficial, but slightly higher levels (>0.1 mg l?1) causes damage to gill membranes of fish. In general, fish may be more sensitive to residual ozone than invertebrates such as shrimp (Reid & Arnold 1994). Culturable heterotrophic bacteria From our results, ozone was capable of eliminating >99.99% of bacteria in ballast water. We attribute the toxicity of the treated seawater to the formation of bromines (measured in this study as TRO). If a signifi- cant amount of TRO remains in ballast water during a voyage, we conclude that heterotrophic microorgan- isms would continue to be inhibited. However, more recent laboratory studies have shown that when TRO disappears, marine heterotrophic microorganisms can rapidly rebound in number (Herwig et al. 2004). Whether microorganisms, such as the heterotrophic bacteria enumerated in our study, should be regulated is somewhat controversial (Dobbs & Rogerson 2005). For interim approval in the state of Washington, a treatment must reduce bacteria by 99% (Washington Department of Fish & Wildlife 2002). The IMO (Inter- national Maritime Organization) Convention ignores most bacteria, other than those of public health signifi- cance including Vibrio cholerae strains O1 and O139, fecal coliforms, and fecal enterococci (IMO 2004). We did not attempt to enumerate bacteria of public health significance in our shipboard study. Phytoplankton and microflagellates Our results suggested that ozone had a strong effect on vegetative cells of dinoflagellates and microflagel- 50 Herwig et al.: Ozone treatment of ballast water lates. The observed decline was probably due to lysis of vegetative cells caused by ozonation. Part of the observed decline could be from cells settling out, and we did not measure the accumulation of cells or resting stages at the bottom. However, because settling would also have occurred in the control ballast tank, mortality was still the most likely explanation for reduced densities of dinoflagellates and microflagellates in the treatment tank. Although our results suggested that ozone may be much less effective for diatoms compared to dinofla- gellates and microflagellates, this probably repre- sented a limitation of microscopic methods used. Diatoms were identified based on the shape and pat- terns of their silica cell walls (frustules) that do not quickly degrade and disappear after ozonation. Thus, although counted in relatively high numbers following treatment, our method could not distinguish between live and dead individuals. We recommend that another method for quantifying and determining phytoplank- ton (and particularly diatom) viability be developed. Overall, our results showed that ozone has promise for removing much of the phytoplankton from ballast water. In future tests, the measurement of chlorophyll a should be considered for assessing the impact of treatment on total phytoplankton biomass. This assay is relatively easy to perform using filtration and extrac- tion with a solvent (Holm-Hansen & Riemann 1978). Zooplankton Although mortality was variable and related to the ozone delivery efficiency, our results indicated that ozone treatment eliminated most zooplankton from ship?s ballast. When ozone delivery was greatest in Expt 3, >96% of the zooplankton was dead after 10 h. Even in experiments where ozone delivery was less efficient, large proportions of the zooplankton were classified as moribund and these probably would not have survived. As with microorganisms, the presence of residual TRO would be expected to continue sup- pression of any remaining individuals. The concentrations of all taxa were greatly decreased by ozone, but we qualitatively observed several that appeared more resistant to the treatment, including a known zooplankton invader, the calanoid copepod Pseudodiaptomus marinus. Laboratory meso- scale experiments with ozone using these taxa, includ- ing growing out treated water after dissipation of TRO, would be beneficial in further identifying resistant taxa and understanding how much ozone is required to eliminate them. Unlike the samples collected at discrete depths in the ballast tank for microbiology and chemistry ana- lyses, zooplankton samples were an integration of organisms present in columns of ballast water. The observational method used to examine mesozooplank- ton was very intensive in that samples collected had to be quickly processed and observed onboard ship. The intensity of this analysis limited the total number of organisms that could be observed in order that all col- lected samples could be processed within a reasonable time. In addition, observations were limited by the mesh size of the plankton net. For our study, a 73 ?m mesh was used. The diagonal measurement for this mesh is about 100 ?m, twice the length suggested by the IMO and pending legislation in the United States (see last subsection below). Based on our experience with this method, determinations of live, dead or mori- bund zooplankton and identification of taxa becomes more difficult when a smaller mesh is used. Toxicology Caged animals Our caged experiments represent a novel approach for evaluating ship-scale ballast water treatment effects. Mortality of caged organisms exposed to ozone was variable, with the least mortality experienced by shore crabs and amphipods. The mechanism for mor- tality was likely to have been related to damage to gill tissues and the animals? respiratory system. As noted earlier, fish tend to be most sensitive to inactivation by oxidizing biocides (Reid & Arnold 1994). Animals that are capable of minimizing their respiration or exchange with toxic water may avoid or delay the cellular damage caused by exposure to biocides. Shore crabs and amphipods demonstrated the great- est resistance to ozone treatment. This outcome may be related to their natural history and physiology. Pur- ple shore crabs Hemigrapsus nudus have a wide geo- graphic range, being found on the west coast from Alaska to Mexico. This crab lives in the intertidal region in and out of water and is capable of withstand- ing a wide range of temperature, salinities and desic- cation. H. nudus is an osmoregulatory organism and can tolerate both hypo- and hyper-osmotic conditions (Kozloff 1993, O?Clair & O?Clair 1998). The amphipod Rhepoxynius abronius is a marine benthic organism that is widely used in sediment bioassays (Swartz et al. 1988, ASTM 1998). Interestingly, this amphipod de- monstrated much greater resistance to ozone treat- ment than the other crustacean used in the shipboard tests, the mysid shrimp. Mortality for the caged animals varied with location in the ballast tank, corroborating evidence from other measured parameters that ozone was not homoge- 51 Mar Ecol Prog Ser 324: 37?55, 2006 nously distributed in the tank. The ship?s schedule only allowed for relatively short experiments, and caged organisms that did not show significant mortality after 10 h of exposure may have died after extended expo- sure to the treated water. Controlled toxicology labora- tory experiments (see next subsection) would help answer questions about extended exposure to oxidiz- ing biocides and delayed mortality, but our experience with the caged animal experiments suggest that this protocol should be considered in future shipboard experiments with biocide ballast water treatments. WET (acute toxicity) tests A limited number of WET tests were performed with water samples collected from ballast tanks of the S/T Tonsina. For experiments conducted in Port Angeles, Washington, we were generally unable to have the WET samples delivered to the toxicology laboratory within 24 h, so the amount of TRO present in the sam- pled water was reduced by the time of analysis. In a separate study (Jones et al. in press), the efficacy of ozone treatment was examined in the laboratory using adult mysid shrimp Americamysis bahia, juvenile topsmelt Atherinops affinis, sheepshead minnows Cyprinodon variegatus, and adults of 2 benthic amphi- pod species Leptocheirus plumulosus and Rhepoxinius abronius. Results from this well-controlled laboratory study showed a similar pattern of sensitivity to ozone- treated seawater as that seen in our shipboard caged animal experiments. Juvenile topsmelt and sheeps- head minnows were the most sensitive to oxidant exposure, while the mysid shrimp was the most sensi- tive invertebrate. In contrast, benthic amphipods were the least sensitive of all species tested. Mortality from ozone exposure occurred quickly with median lethal times ranging from 1 to 3 h for the most sensitive spe- cies, although additional mortality was observed 1 to 2 d following ozone exposure (Jones et al. in press). Shipboard and full-scale testing of ballast water treatment systems Few published studies describe the results of full- scale treatment systems evaluated on ships or at test bed facilities. These include sequential hydrocyclonic and ultraviolet light systems (Sutherland et al. 2003, Waite et al. 2003), deoxygenation treatment (Tamburri et al. 2002), and heat (Rigby et al. 1999). The biological efficacy of the prototype ozone treatment system installed on the S/T Tonsina generally compared well with previously described treatment technologies. Specific results are not directly comparable because the methods used for determining treatment efficacies were not similar. Some investigators took the strategy of adding a few representative organisms to ballast water and others examined only a few specific taxa that are present in seawater. In previous studies, bac- terial populations were usually not enumerated. For our shipboard tests, we attempted to perform a more comprehensive examination of the organisms present in ballast water. In recent experiments on deoxygenation, an inert gas generator was used to strip ballast water of oxygen and to introduce carbon dioxide and lower the pH (Tamburri et al. 2003). In the initial tests, 3 invasive invertebrates, Ficopomatus enigmaticus (serpulid polychaete), Carcinus maenas (European green shore crab) and Dreissena polymorpha (zebra mussel) were exposed to hypoxic conditions for 2 or 3 d in a ballast tank. Percent survival of all 3 species was reduced compared to controls, but the polychaete and zebra mussels demonstrated nearly 20% survival in the treated water. A comprehensive literature review (Tamburri et al. 2002) suggests that a variety of aquatic invertebrates and vertebrates can tolerate hypoxia or anoxia for a few days, and this treatment may therefore not be suitable for short voyages. Therefore, deoxy- genation may not be suitable for coastal voyages such as on the west coast of the United States, where travel between ports may only be a few days. Rather than waiting a few days to observe the lethality of ozone treatment, all our experiments were performed within 5 to 10 h, during which time the treatment rapidly killed a wide variety of organisms. Waite et al. (2003) described large-scale experiments using commercially available units: a hydrocyclone, a self-cleaning 50 ?m screen, and an ultraviolet (UV) unit. In experiments conducted on Biscayne Bay (Florida) zooplankton, phytoplankton, microbiology, ATP and proteins were analyzed. Results showed that hydrocyclonic separation was ineffective and that the 50 ?m screen removed most of the zooplankton. UV treatment initially reduced the viable counts of microorganisms, but bacterial regrowth was observed in samples held for 18 h. Unlike most biocides, UV treatment does not provide a residual toxicity in treated water. Waite et al. (2003) concluded that only the 50 ?m screen was effective in removing organisms, especially potential invaders such as larger zooplank- ton and invertebrate larvae. In our experiments, bacte- rial regrowth was not observed and zooplankton were largely eliminated, but as described earlier, a residual TRO must be maintained during a voyage to prevent the growth of bacteria. Sutherland et al. (2001) evaluated a similar system, a cyclonic first stage followed by a UV phase, in British Columbia, Canada. Samples were collected from dif- 52 Herwig et al.: Ozone treatment of ballast water ferent stages of the treatment. Invertebrates were assessed immediately after collection while phyto- plankton were incubated for ?grow out?. Following treatment, dead and moribund copepods were observed, but low densities and high variances pre- cluded statistical analyses of them. Phytoplankton analyses focused on 3 diatom species, Skeletonema costatum, Thalassiosira sp. and Chaetoceros gracile. Lowest concentrations and growth rates of these taxa were usually observed following UV treatment, with C. gracile being the most sensitive species. Our re- search team is interested in performing ?grow out? experiments for phytoplankton in future treatment tests, rather than enumerating phytoplankton in pre- served samples. While the grow out method can deter- mine viability of phytoplankton, the incubation period may last several weeks. In an ocean trial, Rigby et al. (1999) conducted a shipboard experiment using heated water from the ship?s main engine, flushed through a ballast tank, which resulted in complete elimination of zooplankton and limited survival of phytoplankton. The effect on bacteria was not reported. Small-scale laboratory work could be performed to determine the minimum tem- peratures and exposure times required to inactivate organisms found in ballast water. Ballast water treatment standards When we designed our sampling and analysis proto- cols for the shipboard experiments on the S/T Tonsina, few regulatory agencies or governments had devel- oped or promulgated standards for ballast water treat- ment. For example, the state of Washington treatment standards were released in 2001, and calls for ?inacti- vation or removal of ninety-five percent of zooplankton organisms and ninety-nine percent of phytoplankton and bacteria organism? (Washington Department of Fish & Wildlife 2002). The United States has no stan- dards for treatment, but legislation is pending before the US Congress. IMO (2004) adopted a convention that recommends member states adopt the following discharge standards for treated ballast water: ?Ships conducting ballast water management shall discharge less than 10 viable organisms per cubic metre greater than or equal to 50 micrometres in minimum dimen- sion and less than 10 viable organisms per milliliter less than 50 micrometres in minimum dimension and greater than or equal to 10 micrometres in minimum dimension; and discharge of the indicator microbes shall not exceed the specified concentrations.? IMO standards were also suggested for selected bacteria of public health significance. The convention will come into force 12 mo after 30 countries, representing 35% of the world?s shipping tonnage, ratify the convention. So far, only 1 country has ratified the convention and 7 others have stated that they intend to ratify it (Marine Environmental Protection Committee 2005). Obvi- ously, scientists and engineers who are evaluating potential treatments must adapt their sampling and analysis methods to standards that are currently avail- able or will be enforced in the future. The determina- tion of viability for all the diversity of taxa present in seawater that includes microorganisms, phytoplankton and zooplankton is not a simple task. The proposed IMO standards require a high level of sensitivity and do not differentiate between taxonomic or functional groups. Numbers are for all organisms present in the selected size fractions. We recommend that regulators, scientists and engi- neers engaged in ballast water research reach a con- sensus about suitable protocols for enumerating organ- isms and determining their viability; otherwise, the results from different research groups and technology vendors will be difficult to compare. Performing a com- prehensive evaluation of the biology, chemistry and toxicology of a potential treatment system onboard a ship is a challenging task. A commercial vessel may not be the ideal platform for performing experiments with treatment systems, particularly if the voyage pat- terns and routes are not known well in advance. Com- mercial vessels do not have space dedicated for per- forming research or for sophisticated analytical and biological analyses. The primary effort of the ship?s crew is to safely operate the vessel and transport cargo. Ship officers and crews are often very busy when their ship is in port and during a voyage, so it is difficult for them to lend a large amount of assistance to a science team. If ballast water treatment research is to be successfully conducted onboard commercial ves- sels, then regulatory agencies and governments must provide incentives to the shipping industry so this can participate and provide vessels for the development of ballast water treatment technology. Acknowledgements. The cooperation of Alaska Tanker Cor- poration was essential to the successful completion of this project. We thank the Captain and the crew of the S/T Ton- sina for their cooperation in every aspect of performing the shipboard research. R. Mueller and staff from the Northeast Technical Services Company operated the Nutech-O3 ozone generator. G. M. Detloff provided help with the collection of samples for toxicology analysis. S. Sulkin, N. Schwarck, G. McKeen and A. Olah provided assistance with the caged ani- mal studies. O. Kalata provided technical assistance in the identification of zooplankton from the S/T Tonsina. BP Trans- portation (Alaska) and Nutech O3, provided partial funding for this project. Researchers from the University of Washing- ton were supported in part from a US Fish and Wildlife research grant (98210-0-G738). The Regional Citizen?s Advi- sory Council of Prince William Sound also provided support 53 Mar Ecol Prog Ser 324: 37?55, 2006 for G.M.R. and his staff from the Smithsonian Environmental Research Center. We thank the NOAA Sea Grant Program (NA16RG2251, NA16RG1681, NA16RG1044) and the US Fish and Wildlife Service (98210-0-G738) for research support dur- ing the latter stages of this study. Washington Sea Grant Pro- gram provided partial salary support to R.P.H. LITERATURE CITED Abd El-Shafy M, Grunwald A (2000) THM formation in water supply in south Bohemia, Czech Republic. Water Res 34: 3453?3459 Andersen DR, Bean RM, Gibson CI (1979) Biocide by-prod- ucts in aquatic environments. Quarterly report covering period October 1 through December 31, 1978. Report No. US NTIS PNL-2941, Battelle Pacific Northwest Laboratory, Sequim, WA APHA (American Public Health Association, American Water Works Association, Water Environmental Federation) (1998) Standard methods for the examination of water and wastewater, American Public Health Association, Wash- ington, DC ASTM (American Society for Testing and Materials) (1998) Standard guide for conducting 10-day static sediment tox- icity tests with marine and estuarine amphipods. E1367- 92, Vol 11.05. American Society for Testing and Materials, Philadelphia Brunetto M, Colin C, Rosset R (1989) Solution chemistry and stability of bromamines ? a high-performance reversed phase liquid-chromatography study. Analusis 17:112?118 Cloern JE (1996) Phytoplankton bloom dynamics in coastal ecosystems: a review with some general lessons from sus- tained investigation of San Francisco Bay, California. Rev Geophys 34:127?168 Cohen AN, Carlton JT, Fountain MC (1995) Introduction, dis- persal and potential impacts of the green crab Carcinus maenas in San Francisco Bay, California. Mar Biol 122: 225?237 Crecelius EA (1979) Measurements of oxidants in ozonized sea water and some biological reactions. J Fish Res Board Can 36:1006?1008 Dobbs FC, Rogerson A (2005) Ridding ships? ballast water of microorganisms. Environ Sci Technol 39:259A?264A Driedger A, Staub E, Pinkernell U, Marinas B, Koester W, von Gunten U (2001) Inactivation of Bacillus subtilis spores and formation of bromate during ozonation. Water Res 35: 2950?2960 Erickson SJ, Freeman AE (1978) Toxicity screening of fifteen chlorinated and brominated compounds using four spe- cies of marine phytoplankton. In: Jolley RL, Gorchev H, Hamilton DH Jr (eds) Water chlorination: environmental impact and health issues. Ann Arbor Science, Ann Arbor, MI, p 307?310 Gallard H, von Gunten U, Kaiser HP (2003) Prediction of the disinfection and oxidation efficiency of full-scale ozone reactors. J Water Supply Res Technol Aqua 52: 277?290 Garcia-Villanova RJ, Garcia C, Gomez JA, Garcia MP, Ardanuy R (1997) Formation, evolution and modeling of trihalomethanes in the drinking water of a town. 2. In the distribution system. Water Res 31:1405?1413 Grosholz ED, Ruiz GM, Dean CA, Shirley KA, Maron JL, Con- nors PG (2000) The impacts of a non-indigenous marine predator in a California bay. Ecology 81:1206?1224 Gujer W, von Gunten U (2003) A stochastic model of an ozonation reactor. Water Res 37:1667?1677 Haag WR, Hoign? J (1984) Kinetics and products of the reac- tions of ozone with various forms of chlorine and bromine in water. Ozone Sci Eng 6:103?114 Hallegraeff GM (1998) Transport of toxic dinoflagellates via ships? ballast water: bioeconomic risk assessment and effi- cacy of possible ballast water management strategies. Mar Ecol Prog Ser 168:297?309 Heitmuller PT, Hollister TA, Parrish PR (1981) Acute toxicity of 54 industrial chemicals to sheepshead minnows (Cyprinodon variegatus). Bull Environ Contam Toxicol 27: 596?604 Herwig RP, Cordell JR, Perrins JC, Ferm NC, Grocock JL, Blatchley ER III (2004) Mesocosm experiments for evaluat- ing potential ballast water treatment systems. Fourth SETAC World Congress, Society for Environmental Toxi- cology and Chemistry (SETAC), Portland, OR Hofman R, Andrews RC (2001) Ammoniacal bromamines: a review of their influence on bromate formation during ozonation. Water Res 35:599?604 Hoign? J (1998) Chemistry of aqueous ozone and transforma- tion of pollutants by ozonation and advanced oxidation processes. In: Hubrec J (ed) The handbook of environ- mental chemistry quality and treatment of drinking water. Springer-Verlag, Berlin, p 83?141 Holm-Hansen O, Riemann B (1978) Chlorophyll a determina- tion: improvements in methodology. Oikos 30:438?448 IMO (International Maritime Organization) (2004) Inter- national convention for the control and management of ships? ballast water and sediment. Available at: www.imo.org/Conventions/Mainframe.asp?topic_id=867 Johnson JD, Overby R (1971) Bromine and bromamine disin- fection chemistry. J Sanit Eng Div ASCE 97:617?628 Jones AC, Gensemer RW, Stubblefield WA, Van Genderen E, Dethloff GM, Cooper WJ (in press) Toxicity of ozonated seawater to marine organisms. Environ Toxicol Chem Kozloff E (1993) Seashore life of the northern Pacific coast. University of Washington Press, Seattle, WA Langlais B, Reckhow DA, Brink DR (1991) Ozone in water treatment: application and engineering. Lewis Publishers, Chelsea, MI Lei HX, Marinas BJ, Minear RA (2004) Bromamine decompo- sition kinetics in aqueous solutions. Environ Sci Technol 38:2111?2119 Marine Environmental Protection Committee (2005) Report of the Committee, paragraph 2.1. International Maritime Organization, London Matheickal JT, Raaymakers S (2004) Second Intenational Bal- last Water Treatment Symposium, 21?23 July 2003: Pro- ceedings. GloBallast Monograph Series 15, International Maritime Organization (IMO), London National Research Council (1996) Stemming the tide: control- ling introductions of non-indigenous species. National Academy Press, Washington, DC Nikolaou AD, Lekkas TD (2001) The role of natural organic matter during formation of chlorination by-products: a review. Acta Hydrochim Hydrobiol 29:63?77 O?Clair RM, O?Clair CE (1998) Southeast Alaska?s rocky shores: animals. Plant Press, Auke Bay, AK Oemcke D, van Leeuwen J (1998) Chemical and physical characteristics of ballast water: implications for treatment processes and sampling methods. Rep No. 23. CRC Reef Research Centre, Townsville, Queensland, Australia Parsons TR, Maita Y, Lalli CM (1984) A manual of chemical and biological methods for seawater analysis. Pergamon Press, New York Perrins JC, Cooper WJ, van Leeuwen H, Herwig RP (in press) Ozonation of seawater from different locations: formation 54 Herwig et al.: Ozone treatment of ballast water and decay of total residual oxidant ? implications for bal- last water treatment. Mar Pollut Bull Pimentel D, Lach L, Zuniga R, Morrison D (2000) Environ- mental and economic costs of non-indigenous species in the United States. Bioscience 50:53?65 Pinkernell U, von Gunten U (2001) Bromate minimization during ozonation: mechanistic considerations. Environ Sci Technol 35:2525?2531 Reid B, Arnold CR (1994) Use of ozone for water treatment in recirculating-water raceway systems. Prog Fish Cult 56: 47?50 Rigby GR, Hallegraeff GM, Sutton C (1999) Novel ballast water heating technique offers cost-effective treatment to reduce the risk of global transport of harmful marine organisms. Mar Ecol Prog Ser 191:289?293 Ruiz GM, Carlton JT, Grosholz ED, Hines AH (1997) Global invasions of marine and estuarine habitats by non-indige- nous species: mechanisms, extent, and consequences. Am Zool 37:621?632 Ruiz GM, Fofonoff P, Hines AH (1999) Non-indigenous spe- cies as stressors in estuarine and marine communities: assessing invasion impacts and interactions. Limnol Oceanogr 44:950?972 Ruiz GM, Fofonoff P, Carlton JT, Wonham MJ, Hines AH (2000) Invasions of coastal marine communities in North America: apparent patterns, processes, and biases. Annu Rev Ecol Syst 31:481?531 Salhi E, von Gunten U (1999) Simultaneous determination of bromide, bromate and nitrite in low ?g l?1 levels by ion chromatography without sample pretreatment. Water Res 33:3239?3244 Suter GW II, Rosen AE (1988) Comparative toxicology for risk assessment of marine fishes and crustaceans. Environ Sci Technol 22:548?556 Sutherland TF, Levings CD, Elliott CC, Hesse WW (2001) Effect of a ballast water treatment system on survivorship of natural populations of marine plankton. Mar Ecol Prog Ser 210:139?148 Sutherland TF, Levings CD, Petersen S, Hesse WW (2003) Mortality of zooplankton and invertebrate larvae exposed to cyclonic pre-treatment and ultraviolet radiation. Mar Technol Soc J 37:3?12 Swartz RC, Kemp PF, Schults DW, Lamberson JO (1988) Effects of mixtures of sediment contaminants on the marine infaunal amphipod, Rhepoxynius abronius. Envi- ron Toxicol Chem 7:1013?1020 Tamburri MN, Wasson K, Matsuda M (2002) Ballast water deoxygenation can prevent aquatic introductions while reducing ship corrosion. Biol Conserv 103:331?341 Tamburri MN, Little BJ, Ruiz GM, Lee JS, McNulty PD (2003) Evaluations of Venturi Oxygen Stripping? as a ballast water treatment to prevent aquatic invasions and ship cor- rosion. In: Second International Ballast Water Treatment R&D Symposium. International Maritime Organization, London, p 34?47 Tucker KA, Burton GA (1999) Assessment of nonpoint-source runoff in a stream using in situ and laboratory approaches. Environ Toxicol Chem 18:2797?2803 US Environmental Protection Agency (1978) In-depth studies on health and environmental impacts of selected water pollutants. Contract No. 68-01-4646, US Environmental Protection Agency, Duluth, MN US Environmental Protection Agency (1993) Methods for measuring the acute toxicity of effluents and receiving waters to freshwater and marine organisms. US Environ- mental Protection Agency, Washington, DC US Environmental Protection Agency (1999) Alternative dis- infectants and oxidants guidance manual. Rep No. EPA 815-R-99?014, Office of Water, Washington, DC Vissers MCM, Carr AC, Chapman ALP (1998) Comparison of human red cell lysis by hypochlorous and hypobromous acids: insights into the mechanism of lysis. Biochem J 330: 131?138 von Gunten U (2003a) Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Res 37: 1443?1467 von Gunten U (2003b) Ozonation of drinking water: Part II. Disinfection and by-product formation in presence of bro- mide, iodide or chlorine. Water Res 37:1469?1487 von Gunten U, Oliveras Y (1998) Advanced oxidation of bro- mide-containing waters: bromate formation mechanisms. Environ Sci Technol 32:63?70 von Gunten U, Pinkernell U (2000) Ozonation of bromide- containing drinking waters: a delicate balance between disinfection and bromate formation. Water Sci Technol 41: 53?59 von Gunten U, Bruchet A, Costentin E (1996) Bromate forma- tion in advanced oxidation processes. J Am Water Works Assoc 88:53?65 Waite TD (2002) Rationale for ballast water treatment stan- dards to minimize translocation of unwanted species. Mar Technol Soc J 36:29?37 Waite TD, Kazumi J, Lane PVZ, Farmer LL, Smith SG, Smith SL, Hitchcock G, Cap TR (2003) Removal of natural popu- lations of marine plankton by a large-scale ballast water treatment system. Mar Ecol Prog Ser 258:51?63 Ward GS (1981) Early life stage toxicity tests with a salt water fish: effects of 8 chemicals on survival growth and devel- opment of sheepshead minnows (Cyprinodon variegatus). J Toxicol Environ Health 8:225?240 Washington Department of Fish & Wildlife (2002) WAC 220- 77-095. Interim ballast water discharge standard approval process. Wilcove DS, Rothstein D, Dubow J, Phillips A, Losos E (1998) Quantifying threats to imperiled species in the United States. Bioscience 48:607?610 Woodward JB, Parsons MG, Troesch AW (1992) Ship opera- tional and safety aspects of ballast water exchange at sea. Mar Technol 31:315?326 Yang M, Uesugi K, Myoga H (1999) Ammonia removal in bubble column by ozonation in the presence of bromide. Water Res 33:1911?1917 55 Editorial responsibility: Otto Kinne (Editor-in-Chief), Oldendorf/Luhe, Germany Submitted: September 20, 2005; Accepted: February 13, 2006 Proofs received from author(s): September 26, 2006