Research on the Dispersion of Sea Louse Pesticides in the Marine Environment

F. Page, B. Chang, W. Ernst, G. Julien, and R. Losier

Introduction
Commercial culture of the Atlantic salmon (Salmo salar) began in Atlantic Canada in 1979. Since then, the industry has grown exponentially so that it is now worth well over $100 million. The majority of the industry is located in the Quoddy Region of southwestern New Brunswick (see Fig. 1).

In recent years, the industry has experienced outbreaks of an external copepodid parasite, the salmon louse Lepeophtheirus salmonis. This parasite occurs naturally on wild salmon and, until recently, occurred in only low numbers on farmed salmon in southwestern New Brunswick (MacKinnon 1997). In the fall of 1994, an outbreak of L. salmonis occurred among New Brunswick farmed salmon (Hogans 1995) and in 1996, the New Brunswick salmon industry experienced for the first time outbreaks of Infectious Salmon Anemia (ISA), a viral disease believed to be transmitted, in part, by sea lice (Nylund et al. 1993). Together, these outbreaks have cost the industry tens of millions of dollars in lost revenue.

Chemical treatments are, and will likely continue to be, necessary to control sea lice for the foreseeable future. Currently, the only permitted chemical treatment in Canada is the organophosphate azamethiphos (Salmosan^®), which has had temporary registration in Canada since 1995. The industry has also requested registration of Excis^® (1% cypermethrin), which has an Investigational New Animal Drug (INAD) exemption for use in the State of Maine from 1995 through 1999, and has been used extensively in fish farms on the Maine side of Passamaquoddy Bay.

Unfortunately, these chemicals can be toxic to non-target marine organisms, including commercially valuable crustaceans such as lobster, and to other marine crustaceans that may be important to the coastal ecosystem. It was recognized that very little was known about the dispersal of sea louse pesticides; the exposure histories that non-target organisms might likely experience; and the toxicity consequences of experiencing particular exposures.

Given these concerns, a multi-disciplinary and multi-partner research program was initiated to help fill some of these knowledge gaps, in part to assist the Pest Management Regulatory Agency in the registration process for these chemicals. The program consisted of dispersal and toxicity studies conducted by researchers from Fisheries and Oceans Canada and Environment Canada, in collaboration with the New Brunswick Salmon Growers' Association. Aspects of the dispersal studies are described below.

The dispersal studies consisted of four main activities:
1. A series of dye and chemical releases;

2. A series of drifter releases made in the vicinity of selected fish farms;

3. A series of current meter moorings; and

4. the development of three dimensional circulation, advection-diffusion, and particle tracking models.

The purpose of the dye and chemical releases was to quantify empirically the rate of pesticide dispersal and dilution in areas typical of those occupied by fish farms, to use these data in conjunction with toxicity data to help assess the impact of the chemicals on non-target organisms, and to provide data for validation of the circulation and advection-diffusion models. The purpose of the drifter releases was to determine the direction and extent of the surface drift from half a dozen fish farms during the first few hours after a pesticide release and to provide drifter trajectories for validation of the circulation and particle tracking models. The current meter deployments provided information on current speed and direction throughout the tracking period at each study site and at 20 fish farms throughout the area. This enabled the dye release sites to be compared to actual fish farm sites. The models are being developed to allow extrapolation and prediction throughout the southwestern New Bruns wick area. This report deals only with data from the dye releases. The analyses reported are preliminary and are in the process of being refined.

Figure 1: Map showing sites of Hardwood Island and Basaltic Island dye dispersion trials in Passamaquoddy Bay, southwestern New Brunswick, September 1997. Salmon farms are also shown

Methods
In the fall of 1996 and 1997, six dye release trials were conducted in nearshore waters of the Quoddy Region of southwestern New Brunswick. In this report we provide information from the two most complete trials, conducted in Passamaquoddy Bay in 1997 (Table 1). During each trial a circular salmon sea pen (16 m diameter), with no net attached, was moored offshore of the intertidal zone. The locations were chosen to meet the criteria specified in the licence, issued by the Province of New Brunswick, permitting the release of the dye solutions. The criteria included the provisions that the dye release could not be conducted within 1 km of an existing fish farm and that the dye solution should not drift into the space occupied by an existing fish farm within a few hours of release.

Once the cage was moored, an InterOcean S4 current meter was moored several hundred meters away from the cage. A tarpaulin, typical of those used for sea louse pesticide treatments, was manually stretched across the inside of the cage. The tarpaulin was then allowed to fill with seawater to a depth of approximately 1 m and secured to the cage frame.

A solution of Rhodamine WT dye (10 kg active dye per trial), methanol, and cypermethrin was introduced into the enclosed water. A stream of air bubbles, generated by pumping air through a commercial air stone, was used to induce water movement and mix the dye solution throughout the enclosed volume of water.

After approximately 1 h, the duration of a typical commercial application of sea louse pesticide, the sides of the tarpaulin were released, and the dye began to advect out of the cage and disperse. Within about 30 minutes of the dye release, 5-10 GPS drifters were deployed into the dye patch and were tracked throughout the duration of the dye experiment. The horizontal and vertical extents of the dye patch were surveyed regularly using a submersible pump, hose and fluorometer system deployed from the CCGC Pandalus III, a 12.8 m research vessel. Water samples were taken at frequent intervals for laboratory analyses of dye and chemical concentration and for use in toxicity experiments. Vertical profiles of water temperature and salinity were also taken during each trial. A Canadian Coast Guard helicopter was used to take aerial photographs at several time periods during these two trials.

We estimated the horizontal dimensions of the patches from aerial photographs and from the fluorescence data. The dye patches began as circles and elongated into ellipses. For the aerial photographs, the lengths of the major and minor axes of the ellipses were estimated using the overall length of a vessel, or the diameter of a fish pen, appearing within each photograph as a length scale. For the fluorescence data, the vessel path and fluorescence readings were used to divide the total data record (for each trial) into spatially and temporally separate patches. Patch dimensions were then estimated as the maximum distances along the major and minor axes between points having dye readings of 1% or more of the maximum post-release readings (Elliott et al. 1997).

Table 1: Two dye release trials conducted in Passamaquoddy Bay, southwestern New Brunswick, 1997.

The area of the patch (A) was then calculated as
(1)

where 2·a and 2·b were the lengths of the major and minor axes of the patch. The radius (R_e) of a circle with equivalent area can then be calculated as
(2)

The rate of change in size of a dispersing patch of neutrally buoyant substance instantaneously released into the marine environment has been estimated to be within an order of magnitude of that predicted by the empirical relationship developed by Okubo (1971, 1974). Okubo's relationship relates s_rc², the horizontal variance of concentration in the elliptical dye patch, to the time (t) after release (eq. 3)
(3)

where c₁e = 2.5·10^-5 cm² sec^-3 (117 m²h^-3) for time scales of about 1-12 h and length scales of about 100-1000 m (Okubo 1974).

The Okubo relationship is based on dye releases conducted in offshore areas. Elliott et al. (1997) compared the Okubo relationship to dye release data compiled from releases conducted within a few hundred meters of the shoreline in areas off Ireland and showed that the Okubo relationship provided a reasonable order of magnitude estimate of dispersal in at least these nearshore areas. In this study, we compared the Okubo relationship with the dye dispersal at our nearshore sites.

Elliott et al. (1997) found that the length scales (diameters) of a patch (whose outline was defined by the locations at which the dye concentration had fallen to 1% of the peak value) represented 4s, where s was the standard deviation of the dye concentration profile in the selected direction. This means that we can estimate Okubo's s_rc as
(4)

When the initial patch size is non-zero, as in the case of dye released from a fish cage, equation (3) can be modified as follows
(5)

If the patch size vs. time data agreed well with the Okubo relationship, we also wished to compare the temporal decay in dye concentration, as measured by the fluorescence data, with values predicted by the Okubo relationship. We assumed that the dye was homogeneously mixed within a cylinder with area equal to that of the patch and a constant depth representative of the observed depth of dye mixing. To calculate the predicted dye concentrations, the initial amount of dye (10 kg in each trial) was divided by the patch volume, which was estimated using patch areas estimated by the Okubo relationship and a constant depth estimated from vertical dye profiles. These predicted values were compared to the observed values for mean dye concentrations for each patch, calculated as the mean of all fluorescence readings taken within each patch (mostly near surface readings).

Figure 2: Aerial photograph of the Hardwood Island dye release (10 September 1997) approximately 30 min after release

Results
Dye Patch Size vs. Time
In the Hardwood Island trial, the dye was released at the onset of flood tide. Initially, the dye moved and dispersed very slowly (see Fig. 2). The dye then began moving slowly in a north-easterly direction, and shortly after 2 h, the plume moved around the eastern tip of Hardwood Island. The plume then continued moving and dispersing in a northerly direction.

In the Basaltic Island trial, the dye was also released at the onset of flood tide. As in the Hardwood Island trial, the dye initially moved and dispersed very slowly. It then began moving and dispersing slowly northward past the western tip of Basaltic Island. Within 1.5h, the plume had extended to the mouth of the channel between the island and the mainland. The dye then began moving eastward into the channel.

The relationships between observed patch size and time for Hardwood and Basaltic Islands are shown in Fig. 3 along with values predicted by the Okubo (1974) relationship. As indicated by the plots, the temporal increase in the observed scale of the dye patches agrees reasonably well with the Okubo (1974)

relationship for time scales of about 1-12 h. For the Basaltic Island trial, observed patches beyond 2 h after dye release (to + t > 2.4 h) appear to fall below the predicted Okubo values, probably because the dye patch had entered a shallow, narrow channel, thus forcing the dye patch to be constricted.

Dye Concentration vs. Time
In Fig. 4 we compare the observed temporal decay in dye concentration with predicted values based on the Okubo (1974) relationship. The depths used to estimate the patch volumes (which were required to calculate the predicted values) were 3 m for Hardwood Island and 4 m for Basaltic Island, except 1 m depth was used for the initial patch volume in each trial (when the dye was still within the cage and tarpaulin).

The predicted and observed concentrations are in good qualitative agreement and suggest that the simple model is of some practical value. The dye, and hence pesticide, is diluted by approximately three orders of magnitude within about 3-5 hours.

Figure 3: Dye patch size estimates from fluorescence data (X) and aerial photographs (o) in Hardwood Island and Basaltic Island dye release trials. Also shown is the modified Okubo (1974) relationship, (solid line).

Discussion
The dispersal of pesticides, like that of most substances, is a complex process. Because of this, there is often a tendency to develop complex circulation and advection-diffusion models that represent many of the physical and chemical details of a specific area. Although such models are often desirable and warranted, their development takes considerable time and money. Hence, relatively quick and cheap approaches, that give reasonable, but perhaps less precise, indications of the relative dispersal and dilution of a pesticide are desirable.

The empirical relationships developed by Okubo (1974) fall into this category. Although, they were derived from dye releases conducted in offshore continental shelf regions where dispersal is not influenced by the close proximity of coastlines and complex circulation patterns, they appear to give a useful guide for dispersal rates in at least some inshore areas.

Although the data analyses and model development are not yet complete, the results to date have already demonstrated that:

· The distance travelled by a pesticide patch during the first 2 to 4 h after release ranges from a few hundred to a few thousand metres, and hence may be carried through an adjacent fish farm (currently, the minimum allowable separation distance between salmon farms in New Brunswick is 300 m

· Dispersal rates in at least some of the farmed areas of the Quoddy Region are consistent with those observed elsewhere in the world (Elliott et al. 1997) and with empirical relationships describing dispersal rates (Okubo 1974);

· dispersal of the pesticides is a three dimensional process that is likely dominated by vertical shear diffusion; and

· dispersal results in dilution of the pesticide by 3 to 4 orders of magnitude within 3-5 h.

These findings have countered the often held belief by some government officials and industry that pesticides released into the marine environment of the Quoddy Region are instantaneously diluted. They have also reinforced the reality that many of the fish farms are sharing the water from adjacent farms on a regular basis and that knowledge of the circulation and dispersal patterns is valuable and necessary information. These findings also made a significant contribution to the Pest Management Regulatory Agency's evaluation of azamethiphos and cypermethrin. Hopefully, as the data analyses, field data collection, model development, and information exchanges with industry and regulators continue, regulations and industry practices will be developed that will ensure the development of a sustainable and competitive salmon culture industry that produces a minimal environmental impact.

Figure 4: Comparisons between the observed mean concentration of dye with the predicted homogeneous concentration assuming an Okubo patch radius and a specified constant patch depth in Hardwood island and Basaltic Island dye release trials.

Acknowledgements
We thank the New Brunswick Salmon Growers' Association, as well as the individual salmon farmers who allowed us to conduct research in the vicinity of their farms. For assistance in the field research, we thank the following staff: from DFO St. Andrews Biological Station, Captain W. Miner and P. Leonard (CCGC Pandalus III), M. Ringuette, P. McCurdy, C. Kohler, and T. Johnston; from Environment Canada, C. Garron, K. Doe, P. Jackman, and A. MacDonald; from the the Huntsman Ma rine Science Centre, E. Carter (Skipper of the vessel W.B. Scott), W. Hogans, and T. Hurley; from DFO Coast Guard, C. Swannell (helicopter pilot) and C. Nisbet (logistical arrangements for the helicopter); from Martec Ltd., K. MacKay; and from the New Brunswick Department of Environment, M. Bolden. Funding was provided in part by the DFO Strategic Research Program and Environment Canada.

References

Elliott, A.J., A.G. Barr, and D. Kennan. 1997. Diffusion in Irish coastal waters. Estuar. Coast. Shelf Sci. 44 (Suppl. A): 15-23.

Hogans, W.E. 1995. Infection dynamics of sea lice, Lepeophtheirus salmonis (Copepoda: Caligidae) parasitic on Atlantic salmon (Salmo salar) cultured in marine waters of the lower Bay of Fundy. Can. Tech. Rep. Fish. Aquat. Sci. 2067: iv + 10 p.

MacKinnon, B.M. 1997. Sea lice: a review. World Aquaculture 28(3): 5-10.

Nylund, A., C. Wallace, ad T. Hovland. 1993. The possible role of Lepeophtheirus salmonis (Krøyer) in the transmission of infectious salmon anemia. In G.A. Boxshall and D. Defaye (ed.) Pathogens of wild and farmed fish: sea lice. Ellis Horwood Ltd., New York. pp. 367-373.

Okubo, A. 1971. Oceanic diffusion diagrams. Deep-Sea Research 18: 789-802.

Okubo, A. 1974. Some speculations on oceanic diffusion diagrams. Rapp. P.-v. Réun. Cons. int. Explor. Mer 167: 77-85.