ICES C.M. 1995 CM 1995/F:6 (revised)This report may not be quoted without permission of the ICES General Secretariat, and its contents do not necessarily represent the views of the Council.
Bedford Institute of Oceanography, Dartmouth, NS, Canada
6-8 September 1995
Report on International Workshop on Modelling Environmental Interactions of Mariculture, submitted to the ICES Working Group on Environmental Interactions of Mariculture by William Silvert and Barry Hargrave, Chairmen. (This version of the report has been revised to include abstracts of the papers presented.)
At the 1994 Annual ICES Meeting, the Council approved a recommendation from the Mariculture Committee to hold a Workshop on Modelling Environmental Interactions of Mariculture at the Bedford Institute of Oceanography (BIO) from 6-8 September 1995 to be chaired by Drs. William Silvert and Barry Hargrave of the Habitat Ecology Division at BIO.
The terms of reference for the Workshop were as follows:
Identify and compare different approaches to modelling the environmental interactions of mariculture.
Evaluate performance, application and cost-effectiveness of these models.
Prepare input on the modelling component of the Joint Theme Session entitled Mariculture: Understanding Environmental Interactions for the ICES Annual Meeting in Denmark in Sept. 1995.
Plan contributions to the modelling workshop which will be organised by the Shellfish Committee in 1996 (the workshop is being organised by B. Bayne and M. Héral and will be probably be held in France in the spring).
Submit a progress report in time for the ICES Annual Meeting later in the same month.
The objective of the Workshop was to bring together researchers actively involved in modelling projects in order to identify and compare different approaches to modelling the environmental interactions of mariculture, and to evaluate performance, application and costeffectiveness of these models.
Because the emphasis was on current developments that represented the state of the art, and not only on established methodology, a decision was made not to publish the proceedings as a refereed document. It was decided instead to ask the participants to publish all or part of their contributions as documents available over the Internet. A World Wide Web (WWW) server operated by the Habitat Ecology Section was designated for this purpose, and as of the time of the Workshop about a quarter of the presentations had already been published in this fashion.
The Workshop was opened by James Elliott, Regional Director of Science for the Maritimes Region of the Department of Fisheries and Oceans. Paul Keizer, Head of the Habitat Ecology Section and an active researcher in the field of mariculture impacts, welcomed the participants and described the present situation pertaining to aquaculture development in Nova Scotia and New Brunswick.
The Workshop program was divided into four categories, as follows:
The program for the Workshop was as follows, with abstracts and in some cases additional comments by the Chairs.
Jouni Vielma (Finland) and Santosh P. Lall (Canada)
In order to develop successful models to calculate environmental impacts of aquaculture, data on estimated nitrogen and phosphorus excretion by fish are essential. The composition and digestibility of diets utilized by fish culture significantly affects the excretion of suspended solids, ammonia and phosphorus. Diets containing a high level of protein or poor quality protein with a low level of energy supplied from non-protein sources results in higher nitrogen load. The phosphorus excretion could be controlled by limiting the use of feed ingredients supplying excessive level of phosphorus and by increasing the feed efficiency.
Chemical and physical characteristics of diets used for different fish species and for different stages of production vary. Instead of collecting empirical data under husbandry conditions, a more general physiological model is preferred. By formulating a simple excretion model, we studied model sensitivity for variations in diet composition, and energy and nutrient digestibility in 15 commercial diets. The paper emphasizes nitrogen and phosphorus excretion, since distinctive absorption, metabolism and excretion mechanisms may exist for compounds containing nitrogen and phosphorus. Carbon load reflects the overall feed efficiency, and is not discussed here.
Nitrogen is found in protein fraction of the diet, and nitrogen excretion is closely linked to the efficiency of protein utilization. The main source of protein in salmonid diets is fish meal, but potential alternate plant protein sources like soybean meals are now being evaluated. The quality of protein, mainly its digestibility and amino acid composition, and the ratio of digestible protein to digestible energy are the two significant variables affecting both nitrogen excretion and overall feed efficiency. Protein digestibility is around 90 % in the most commonly used protein sources. Energy digestibility reflects protein, fat and carbohydrate digestibilities, and varies more due to the variation in their contents in diets. Due to the development in feed manufacturing, content and digestibility of energy in diets has increased. Fish meal is also the main source of dietary phosphorus. There is large variation in phosphorus content (15-40 g/kg) and digestibility coefficient (45-70 %) of fish meals. Plant products contain less phosphorus (5-15 g/kg) of which most part is unavailable to monogastric animals like fish, since it is bound as insoluble salts (phytates). Despite the different functions in animal body, a model of similar structure can be used for simulating both nitrogen and phosphorus load in fish farming.
A model similar to the generalized energy budget was constructed. 15 commercial diets used for feeding adult Atlantic salmon and rainbow trout, and supplied by four manufacturers operating in Scandinavia, were selected. Total energy, protein and phosphorus content of the diets was recorded as labeled by the manufacturers. The range for the values were 21.4-24.9 MJ/kg, 400-470 g/kg and 8-11 g/kg, respectively. Nutrient and energy contents of diets were fairly uniform. All had balanced protein to energy-ratio, i.e., there were moderate in protein and high in fat content. After surveying the literature for empirical values used in the model, the excretion of nutrients was computed by 1) estimating the energy cost of growth (assumed to be 19 MJ digestible energy per kg growth), 2) computing nutrient intake on the basis of the energy cost of growth, and digestible energy and total nutrient content of diet, 3) estimating the amount of absorbed nutrient on the basis of nutrient intake and nutrient digestibility coefficient, and 4) computing the amount of dissolved excretion by using the amount of absorbed nutrient and body content of the nutrient (assumed to be 4 g P/kg and 30 g N/kg). Digestibilities used in calculations were for energy, nitrogen and phosphorus 85 %, 85 % and 50 % respectively.
The calculated biological feed conversion ratio (no mortalities, no feed wastage) was 0.90-1.04. Due to relative uniform nutrient and energy contents, there were no large differences in calculated nutrient excretion rates. The average calculated phosphorus excretion was 5.1 g/kg (standard deviation 0.7 g/kg). 10 % of the phosphorus was excreted in dissolved form. For nitrogen, the average calculated excretion was 37 g/kg (standard deviation 3 g/kg) of which 73 % was excreted dissolved. In comparision, feeding Baltic herring, which is still practised to some extent in cage farming in Finland, results in four-fold phosphorus excretion due to its low energy and high nutrient contents. Decreasing assumed phosphorus digestibility to 40 % induced phosphorus deficiency in all but one diet. Clearly, in high-energy diets phosphorus content is very close being a limiting factor if digestibility can not be increased. Increasing energy cost of growth and decreasing energy digestibility by 10 % increased nutrient excretion by 38 %. To compute total nutrient load, feed wastage and biomass growth must be estimated. There are several growth models all of which need to be validated for specific husbandry conditions. To estimate feed wastage, a modeler must rely on empirical data.
It is concluded, that labeled nutrient contents of present commercial salmon diets reveal no large differences in their loading capacity. Ignoring the energetic estimation of nutrient intake may result in biased excretion values if diets different to standard commercial diets are used.
During discussion S. Cripps advocated monitoring effluents to check model, as is done in Norway. In response to a question about the confidence limits on effluent calculations, the ratio of low to high estimates was given as roughly 2:3.
Simon Cripps (Norway)
The particulate phase of the effluent from a salmonid hatchery was serially divided into seven size fractions using 200, 100, 85, 65, 47, 25, and 5 µm pore size nylon membranes. The total, dissolved and particulate phosphorus and nitrogen, suspended solids, and various particle size parameters, such as mean particle diameter and total particle volume, of each fraction were determined. Concentrations in the fractions ranged between: total phosphorus, 131.7 - 67.9 µg/l (a 48 % reduction from the unfiltered effluent to the fraction filtered using a 5 µm membrane); total nitrogen, 0.70 - 0.47 mg/l (33 % reduction); suspended solids, 6.9 - 1.8 mg/l (74 %); total particle volume concentration, 26.17 - 2.15 µm3/l·109 (92 %); particle number concentration: 602 - 378 /l·103 (37%). No single separated particle size fraction contained a disproportionate concentration of total phosphorus, total nitrogen, total particle volume concentration, or particle number concentration. The phosphorous content, in both the suspended solids and the volume of the particles, increased significantly with decreased particle size. The concentrations of suspended solids and total particle volume in the small pore size fractions were however less than in larger fractions. Total phosphorous and total nitrogen concentrations were therefore unaffected by variations in the particulate phosphorous concentration associated with different size particles. The nutrients and solids removal efficiency data for each fraction could be used to estimate the minimum achievable removal efficiencies of a commercial effluent screening device with a specified pore size. Total particle volume concentration, was significantly correlated with suspended solids concentration and so could be used in place of the latter term. Total particle volume concentration, could be measured to lower concentrations than suspended solids concentration. Particle size analysis, used in conjunction with other techniques, such as fractionation and nutrient analysis, is a powerful tool for the characterisation of aquaculture wastes and for the improvement in wastewater treatment efficiency.
T. G. Milligan (Canada)
Grain size is the most basic of classification criteria for sediments. Sediments can be characterised by the way in which the material they are made up of was deposited on the bottom. The size differentiation that produces a given sediment size distribution is a function of physical transport processes and a record of those processes will be preserved. The erosion of terrestrial material tends to create straight size distributions with the volume of material in each size class varying regularly. After being delivered to a bay or estuary by run off or shore erosion, these straight size distributions are modified to reflect the transport energy of the water. The fine particulate material, which also makes up the suspended load, is unable to settle through the water column unless it flocculates with other particles to form large, fast settling aggregates or marine snow. Studies have shown that flocculation rate is controlled by concentration, size and particle adhesion forces and that the particles in suspension deposit in proportions equal to that in which they are found in the suspension. By using precise Coulter grain size analysis and the model developed by the late Kate Kranck, it is possible to break down a sediment into the three major components from which it was formed: material deposited as flocs, material deposited as single grains from suspension and material carried under higher energy conditions. Using this method, both the amount of material deposited in a flocculated state and the maximum size, or floc limit, of the particles composing the floc can be determined. The increased fine particulate load, and the possible increase in the amount of biological "glue" in the form of polysaccharides, which results from mariculture can increase the rate of flocculation in an inlet and change the deposition rate and distribution of fine sediments. By examining the particle size distribution of sediments in regions of aquaculture it is possible to determine zones of fine particle deposition and to assess changes in the particle dynamics of an area.
C. G. Hannah, J. W. Loder, and Y. Shen (Canada)
A sediment transport model is being developed for the dispersion and transport of suspended sediment in the benthic boundary layer on the continental shelf. The model is one component of an impacts zone assessment model for drilling discharges on benthic organisms such as the sea scallop on Georges Bank. A key feature of the model is the simplified representation of vertical mixing through the random shuffling of packets of material and the use of observational suspended sediment profiles as probability density functions for the overall vertical distribution. Horizontal transport is then represented through advection of the packets in vertically and timevarying flow fields. The dominant dispersion process acting in the model is horizontal shear dispersion - due to the interaction of vertical mixing and vertical shear in the horizontal currents. The effective horizontal dispersion rates in the model are consistent with the fundamental idea that shear dispersion is linearly proportional to the product of the vertical mixing time scale and the mean square velocity deviations experienced by the suspended sediment. This talk will concentrate on the model's representation of shear dispersion and comparison with analytical solutions. Applications to a drilling site on Sable Island Bank and to the scallop grounds on Georges Bank will also be discussed.
Gary Bugden, Paul Keizer and Dave Wildish (Canada)
Cage flushing times were measured by Rhodamine-WT dye injection in a heavily fouled, 12 m square salmon pen stocked at approximately 4.7 kg m-3. A five port manifold was used both to inject the dye and withdraw water samples for concentration analysis. Exponential decay curves were fitted to dye concentrations after an initial period of about 3 minutes required for the dye to reach a homogeneous distribution within the cage following injection. Eight runs were made under various conditions of incident current. Two of the eight runs had to be discarded as flushing was too rapid for the dye to reach homogeneity, a condition necessary for the exponential fits to be meaningful. Most rapid flushing occurred for currents across the open side of the pen vs. currents running along the length of the multiple pen array. Fitted exponential time constants ranged from about 1 to 7 minutes. Estimates were also made of the mixing within the cage due to various levels of fish activity. The results have application to both farm design and estimation of the environmental impact of parasite treatments.
Power spectra of dissolved oxygen time series observations from both within a mariculture site and at a control site were briefly presented. Diurnal and tidal signals were evident.
Michael Dowd, Keith R. Thompson, and Jinyu Sheng (Canada)
Assessing the near and far field effects of coastal mariculture requires detailed knowledge of particle dispersion processes. We present three case studies designed to illustrate the complexity of this problem on spatial scales ranging from an inlet to the continental shelf. The focus is on particle trajectories calculated from Eulerian flow fields obtained from numerical models. The first case deals with a tidally flushed coastal inlet (Upper South Cove, located near Lunenburg, N.S., Canada). The complex geometry and strong advective nonlinearities result in a marked flood/ebb tide asymmetry and complex particle paths. Second, we present an idealized study of periodic flow over isolated topography. In spite of the fact the flow field is governed by essentially two parameters (the tidal excursion and the forcing frequency), no simplified description of dispersion seems possible. In the final case, particle dispersion and retention indices are shown for the Scotian Shelf. In all three cases particle trajectories are complex and not amenable to simple parameterization schemes. We suggest instead that a statistical or probabilistic description of particle dispersion and flushing should be derived from existing circulation models. These reduced dimension models are then able to be effectively utilized in modelling the environmental effects of mariculture.
Carter Newell and V.G. Panchang (USA)
Considerable effort is being invested in making measurements of hydrodynamic, water quality and benthic parameters for monitoring and regulating the environmental impacts of marine finfish cage culture. The associated costs, generally borne by the industry, are large. Further, discrete measurements over limited time periods at specific sites do not give a truly representative overall picture; they cannot address spatial and temporal variations within lease sites or the cumulative effects of several operations within a coastal embayment. In a previous project funded by NOAA NMFS (technical report Maine Sea Grant TR-93-1), we have evaluated mathematical computer modelling methods and demonstrated their effectiveness in resolving these difficulties. We have recommended a step-wise comprehensive modelling strategy that involves a systematic investigation of tidal and storm-induced currents, wave effects, waste transport (using a particle tracking model), resuspension, and decay. These models, if accepted by regulatory agencies, can result in significant cost savings to the industry.
In our new project (November 1995 - April 1997) we wish to determine, via measurements using a field-deployed annular flume called the Sea Carousel, the critical shear velocity at which settled wastes are eroded and resuspended. At present, this information does not exist and this aspect was observed to be a critical parameter affecting lateral spread of particulate aquaculture waste.
We also propose to package all the models in a form usable by state and federal regulators, and transfer the technology (hardware and software) to regulators in Maine, who may apply the product to routine monitoring of existing lease sites and evaluation of potential new lease sites in Maine. In addition, we will examine model predictions relative to existing data on sediment granulometry and visual video transects at several study sites in Cobscook Bay, Washington County, Maine. It is anticipated that at least half of the finfish aquaculture sites currently monitored are highly erosional and costly monitoring of those sites may not be necessary.
It was acknowledged that particulate losses from pens may be widely distributed in inlet, but was only modelled under cages. Formation of Beggiatoa mats was used as a threshold for unacceptable benthic impact. Licensing can be regulated on this basis, as is already the case in Maine.
Anders Stigebrandt (Sweden) and Jan Aure (Norway)
MOM (Monitoring-Ongrowing fish farms-Modelling) is a management system which can relate the environmental impact of fish farms to the holding capacity of a site. The holding capacity is defined as the sediment ability to receive organic effluents without causing the benthic impact to exceed predefined levels. The system consists of three integrated parts: 1) environmental quality standards, 2) a standardized monitoring program for checking the environmental conditions on a site according to environmental quality standards, and 3) a mathematical model for predicting the expected environmental impact in relation to farm specifications and information on the hydrography and topography of the site. MOM is described in Ervik and Kupka Hansen (1994: ICES C.M. 1994/F:26).
The present abstract describes a mathematical model that computes maximum fish production under the prerequisite that there has to be acceptable living conditions for benthic animals at the site. This model now constitutes the benthic part of the mathematical model for the MOM system as described above. The building blocks of the model are a new oxygen supply model, computing how bottom sediments and benthic fauna get oxygen from the overlying water, and a new dispersion model for particulate organic waste from the cages in a fish farm. The model will be published in Stigebrandt and Aure (1995: Fisken og Havet, no. 26).
The oxygen supply model computes the oxygen transfer through the turbulent benthic boundary layer. The model probably has general validity but is not yet verified, among other things the values of some constants in the model are not satisfactory determined. Further, the relationship between measured currents and the dimensioning current velocity to be used in the model is still uncertain. These are things that will be clarified during the further development and testing of MOM.
The dispersion model for particular organic waste from the fish cages has provided new and enlightening knowledge on which factors that are most important for the dispersion of the waste. In particular the model focus on dispersion effects of cage size and cage separation. From the current variability, expressed as the standard deviation s and estimated sinking times T for the organic waste, the model simulates how waste from a single cage is spread over the bottom. The loading with organic matter is generally maximal below the centre of the cage and decreases with the distance from the centre. The dispersion pattern is displaced by a possible non zero mean current past the cage. The loading of every point of the bottom beneath a marine fish farm is computed as the sum of the contributions from all cages. Simulations using the dispersion model show the following general results: 1) The natural dispersion ability at a site is determined by the 'dispersion length' sT. This may be increased using food with lower sinking speed. 2) The fraction of organic waste from a single cage deposited on the bottom outside the projected area of the cage increases, and the maximum loading beneath the cage decreases, with increasing dispersion length and decreasing area of the cage. 3) In a fish farm, the maximum loading of the sediment with organic waste decreases if the distance between cages is increased and the number of rows of cages is decreased.
The present version of the model for maximum fish production seems to give realistic results. It may also play an important role in the further development of MOM, not least because it provides a structured model of the problem and by that an objective possibility to rank different influencing factors after their relative importance. Computations of maximum fish production are done by a computer program described in Stigebrandt and Aure (1995). Finally, it should be mentioned that in addition to the model for the benthic holding capacity of a fish farm site presented above, the complete MOM model will also contain a model for the environmental conditions in the fish cages and a model ('Fjordmiljø') that computes the far field environmental impact. The foundation of Fjordmiljø is described in Aure and Stigebrandt (1990: Aquaculture, 90:135-156).
Even if the amount of production will influence the environmental impact, the amount of fish itself is not necessarily the responsibility of the authorities. If you can't sell the fish it's your problem. If you pollute the environment it's everybody's problem, no matter what quantity of fish it was that caused the problem.
The amount of fish you can produce without causing problems will depend on the potential of the locality (current, depth etc.), technical solutions, and your own skills. If a great potential exists, why not realise it and attain a cost effective production? This will also stimulate the farmers to find the best localities, often resulting in selecting sites in the outer coastal areas.
In Norway we have a regulation based on volume. Most farmers have a 12,000 m3 permission combined with a limitation of 25 kg/m3. The volume is calculated as 85% of the cage area multiplied by 5 m depth. The maximum biomass therefore is 300 tonnes of fish, resulting in a production potential of around 500 to 600 tonnes, no matter what the carrying capacity of the locality.
In real life however the average cage depth is around 12-13 m, resulting in better water quality for the fish when the biomass increases. Problems that may occur are seldom related to water quality because the general conditions are ideal for aquaculture. The current is typically 12-18 cm/s and the temperature varies from 4°C in the wintertime to around 15°C in the summertime. Even if the depth usually is 30 m or more, problems in relation to sedimentation may occur, at least in a longer perspective. The processes responsible for this are often predictable, assuming you know something about the physical conditions and have a some sort of clue of what's taking place at the fish farm. This is the point where computer simulations come into the picture.
What if we could give the farmer an environmental forecast and prevent a breakdown of the bottom life? Adjustments could be made to adjust the production to the carrying capacity of the locality. We could do this before fish production starts, based on general information and measured parameters. Later on we could measure the real effects and relate them to standard official quality demands for aquaculture activity. If the development is bad we could give a warning, and if necessary we could demand a full stop in production. We would gain experience and be able to make adjustments to the simulator, bringing it closer to the real life step by step. Detailed information about production, amount of feed used, feed waste, etc. is already available to the Norwegian authorities.
A suggestion for a national environmental aquaculture standard has been made. A appurtenant standard monitoring program is being tested, and we have developed a computer model for determination of the carrying capacity at different sites. Further testing will tell us into what degree these tools are working together. Then we will consider the possibility of an integration with existing regulations.
In a nearby future maybe the farmer will have a permission either in volume, number of fishes, production in tonnes, or most likely an area in m2 combined with environmental standards. He will have permission to utilise the production potential of salmon so long as he doesn't harm nature. If such a danger exists, we will probably know in time to prevent a breakdown.
A simulator combined with environmental standards could represent a new way of regulation. First of all we could focus on the effects of what's been done instead of talking about how much feed is given. Next we could use a simulator for coastal planning and to make calculations about relationships between selected characteristics of the localities and their carrying capacity. Example: What is the optimum depth of a cage at a locality with depth = 30 m and with a current = average? Maybe we should not allow cages more than 1/3 of the total depth. What would be a reasonable density in a cage that is 10 m deep? The simulator could give us the answer, leading to a table of the relationships between total depth, cage depth and the maximum density permitted. Are there other connections that could be useful in an attempt to develop a more scientific based set of regulations?
The knowledge about biological processes taking place in fish farms has increased. They are complicated for sure, but still it seems possible to describe a lot of them mathematically. Giving these equations a user-friendly interface could represent a possibility of using them in public administration, often by people who don't have a detailed knowledge of the processes behind.
May be this is believing in Father Christmas, but then again its Christmas time the moment this is written.
Cédric Bacher (France)
In bays and estuaries, the interactions between cultivated species and the environment is most often driven by the hydrodynamics. For extensive systems for instance, the food supply to the cultivated species and the impact of the excretion from these species depend on the water mixing and on the residence time of the water, e.g. the time needed for the natural renewal of the water within the cultivated area.
A new algorithm was proposed to compute water residence time in bays or estuaries. Based on a probabilistic approach of the tidally averaged circulation of the water, the method allowed to generalize other temporal scales definitions as transit time and local residence time in sub-areas of the system. The method required the knowledge of local flows, for instance derived from an hydrodynamical model. In this case, the area concerned with the water circulation was divided in a discrete number of spatial cells. The connection between cells yielded a connection matrix describing existing exchanges. From this, a transition matrix was built to assess the probability for elementary water particles to leave a cell and arrive to another cell connected to the former. Iterating transitions over time provided a way to describe the transport of water particles under the assumption of steady state flows. At each time step the probability of exiting the system was computed, which established the basis for defining different temporal scales. Considering several water particles distributions among the whole area, the residence, transit, or local residence time were represented as the expected value of a random variable which distribution was derived from the probability of exiting the system. These concepts and algorithms were validated in the one dimension case for systems with constant flows and water volumes, for which the transit and residence times are known.
It was also used to compute the residence time of the water in the Marennes-Oléron Bay (France), where the importance of the tidal excursion resulted in the use of a spatial box model. The bay was therefore divided in 15 spatial boxes and the flows between the boxes were estimated from residual flows computed with an hydrodynamical model. An ecosystem model was conceived to assess the phytoplankton consumption by cultivated molluscs and the growth rate of these molluscs due to the phytoplankton abundance. Combining the spatial box and the ecosystem models yielded to assess the impact of the cultivated molluscs on the phytoplankton abundance and production. It was shown that this impact is linked to the water residence time which was computed as described previously.
Pedro Monteiro (South Africa)
Quantitative assessment and modelling of the impact of fish farms on the organic load and composition of marine sediments is often a problem because, especially in the context of a second input, it is difficult to establish a direct link between the source and the sediment data. Stable isotopes (Carbon 13C, Nitrogen 15N and Sulphur 34S) may offer a solution to this problem as a result of the link between the variability of the natural abundance of light and heavy isotopes and trophic position in the food web. The benefits of using stable isotopes are particularly evident when two and if possible three are simultaneously measured introducing sufficient independent parameters to solve enough mass balance equations for the number of inputs.
The intensive use of stable isotopes to study the trophic dynamics of marine food webs is a relatively recent phenomenon but it has proven very useful to support or question existing views on trophic dynamics in often complex systems. Its use has been extended to the study of the far-field impacts of land based organic discharges (sewage) on sediment biogeochemistry and benthic food webs. The objectives of this presentation are:
The use of these techniques were illustrated with a case study where the appearance and development of an Ulva sp. bloom in Saldanha Bay, South Africa was linked to a fish meal factory discharge using stable isotopes. Stable isotopes are not only useful tracers but also help to understand the ecological interactions which occur as a result of enhanced organic loading of marine systems which are the cause of eutrophication. It is therefore a "tool box" of techniques which though still under utilised in marine aquaculture, offers clear benefits to constrain environmental impacts and assess near and far field model predictions.
Peter Strain (Canada)
Simple models have been developed to describe inlet scale water quality for two inlets in Atlantic Canada. Letang Inlet, N.B., is a macrotidal inlet adjacent to the Bay of Fundy; Ship Harbour, N. S., is an inlet on the Atlantic coast of Nova Scotia whose water exchange is controlled by a mixture of estuarine circulation and offshore forcing. These models use one or more simple descriptions of the physical process that dominates water exchange to predict the impact of nutrient inputs to the Letang and to determine nutrient regeneration rates in Ship Harbour. Simple models such as these are cost effective means of answering some questions about water quality, but the models must be inlet specific, must adequately describe the dominant physics, and must be used judiciously. The predictions of such models must always be viewed in the context of the quality of the input data and the assumptions underlying the models
Dror Angel (Israel), Peter Krost (Germany), and Bill Silvert (Canada)
The effects of a net pen fish farm, situated in the northern Gulf of Aqaba (Red Sea), on the underlying benthos were studied from 1991 to 1994. The flux of particulate organic matter released from the fish farm (measured by sediment traps) varied between 4.5 and 12.7 g-C m-2 d-1, and covered approximately 17,000 m2 under the fish farm. The organically enriched sediments (LOI = 4 - 12%) contained high concentrations of dissolved nitrogen, phosphorus and hydrogen sulfide and were covered by microbial mats, consisting mainly of the chemoautotroph Beggiatoa spp. and an assortment of photoautotrophic and heterotrophic micro-organisms. Rates of organic carbon (OC) decomposition, calculated from benthic flux chamber measurements and sediment porewater profiles, ranged between 1 and 5 gC m-2 d-1. Despite the fact that OC sedimentation rates exceeded decomposition rates, there was no sign of OC accumulation in the sediment. The missing organic material was probably removed by one or a combination of the following factors: resuspension and removal by occasional strong bottom currents, decomposition by bioturbation and uptake by demersal fish and invertebrates. A 3-year time series of 100 logs summarizing diver observations on the status of the benthic environment below the fish farm was established in the course of this study. The initial purpose of the dive logs was to monitor the benthic dynamics in descriptive terms, since many of the changes that occurred were not measurable by routine geochemical analyses, yet they provide valuable information on the status of the benthos. The observations in the diver logs were mostly qualitative and subjective and there were large inconsistencies in the type of data recorded making it difficult to analyse the data by conventional approaches. We decided to apply fuzzy logic (Zadeh 1965) to the data since this approach allows us to incorporate ambiguity and nonquantitative data into a classification scheme to describe the status of the benthos. To the best of our knowledge, this is the first time fuzzy logic is being used to analyse environmental data and the preliminary results appear very promising. We found a strong correspondence between fuzzy scores that match highly enriched sediments and high levels of sediment organic matter (loss-on-ignition), porewater nutrients and organic matter decay rates.
Barry Hargrave (Canada)
A survey of benthic variables at cage and reference sites was used to identify variables useful for cost effective monitoring of benthic conditions. Benthic gas fluxes, sulfide and Eh were the most significantly different variables between reference and cage sites. Surface sediment Eh and organic carbon were combined to calculate a benthic enrichment index (BEI) which was inversely correlated with organic carbon sedimentation. Sedimentation >1 g-C m-2 d-1 represents a threshold of sustainable benthic oxidation. Regression analysis was used to calculate carbon sedimentation from BEI which can be compared to sediment respiration. Imbalance indicates over or undersupply relative to potential loss through benthic respiration.
Barron Carswell (Canada)
The Salmon farming industry in British Columbia has been and continues to be affected by land use planning activities and by environmental assessments. In terms of Coastal Zone Planning the main result of these planning activities have been to define exclusion zones for aquaculture. Several individual agencies have relevant policy and statutes in place to regulate the fish farm industry however there is no overarching policy to guide and co-ordinate the Province's approach to salmon aquaculture. An important part of this process to development a provincial policy will be an environmental assessment in the Broughton Archipelago area. One expected result will be the definition of a threshold limit of feed input, above which individual salmon farm applications will be subject to a formal Environmental Assessment review. The present situation provides an excellent and perhaps the last significant opportunity to apply numerical salmon loading models to affect the Provincial policy on salmon aquaculture.
Peter Chandler (Canada)
The British Columbia Ministry of Agriculture, Fisheries and Food has undertaken the development of a numerical model to enhance its aquaculture management capabilities. The purpose of this computer based system is to provide a tool that can be used by Ministry staff to support the decision making process concerning the licensing of aquaculture sites and in assessing the local and regional impacts of aquaculture operations. Because the objective is to develop a means to make complex management decisions for natural resource systems that are themselves very complex a modular structure to the modelling system has been adopted. The modelling system is comprised of three sections; the preprocessor to establish the scenario to model, the processor to coordinate the execution of the submodules that simulate a range of biophysical processes, and a postprocessor to display the results. These three sections are linked by ancillary software; specifically a Windows based Graphical User Interface (Visual Basic) to facilitate the exchange of information between the operator and the model, an (ArcView 2) Graphical Information System to display the model output and incorporate existing environmental databases, and a Netscape hypertext markup language (HTML) support document to allow the operator to point and click for online reference material. An assessment of the modelling system has been carried out for a region of important aquaculture activity in British Columbia, the Broughton Archipelago. The tidally driven, two dimensional, depth integrated hydrodynamics of the area have been modelled using a 1 km resolution over the entire domain and at a 100 m scale over a subsection of particular interest. Spatial variations of dissolved oxygen concentration and the benthic footprint of carbon due to several fish farm operations are presented. Simulations are also made of the release of water borne toxins to determine the contours of concentration surrounding the release site. The advantages of the modelling system are shown to be the accessibility it provides to an operator having a limited background in computer systems but a requirement to use complex computer models, and the integration of several process models to provide a more comprehensive understanding of the chemical, physical and biological processes that govern the interactions between the fish farms and the environment.
This project, like the one described by Jahnsen, used Microsoft Access.
Juhani Kettunen (Finland)
Dealt with bioeconomic decision modelling to balance industry goals (profit) with environmental concerns with no clear economic value. A major problem is discerning what society (clients) want.
The individual presentations were followed by a roundtable discussion of how the modelling work discussed can best be integrated with management and the decisionmaking process. A volunteer from each country commented from the perspective of that country's regulatory practices.
Several participants referred to the special issue of J. Appl. Ichthyology Vol. 10, No. 4, published in December 1994, which carried the Proceedings of the Workshop on Fish Farm Effluents in EC Countries, edited by H. Rosenthal. This is a valuable resource on policies and practices in the EC, and goes into some of the issues described below in greater depth.
Finland (Kettunen) - The program in Finland is ad hoc and includes extensive monitoring but little modelling.
US (Findlay) - The Environmental Protection Service exerts strong state control over siting of pens to minimise benthic impact. The sea bottom is considered to be common property so that site licences are a variant of common use held by lessee. Present regulations in Maine are based on Weston's report (1986) from studies in Washington. Annual monitoring at all sites by video transects is required by law.
Norway (Jahnsen) - since Jan 1994 to Jan 1995 biomass of cultured fish in Norway has increased from 121,00 tons to 170,000 tons. Presently 250,000 tons are held - far in excess of licensed capacity. Legal maximum fish density allowed at a site is 25 kg m-3 (water depth is usually >12 m). A new idea was presented to make cage depth a function of total depth. As water depth increases and deeper cages can be used, there are grounds for considering an increase in stocking biomass (e.g. increase from 125 kg m-2 at 1015 m to 240 kg m-2 for depths >100 m). A recommended protocol is being considered to monitor sediment Eh and pH. A new classification scheme was presented to characterise sedimentary organic loading based on these two variables. Monitoring of existing sites could be used to decide if expansion of stocking density should be allowed. Norway appears, relative to all other countries, to have the most advanced regulatory system based on limiting benthic impacts. There is also awareness of the need to achieve a compromise between environmental protection and the health of the industry. Sustainability is a priority issue.
Canada (Keizer) - several regulatory agencies are responsible for different aspects (Departments of Fisheries and Oceans, and of Environment), federal vs. provincial. New provincial regulations presently being implemented in N.B. based on video transects (visible benthic changes). Regulations are not based on the use of models, largely because there are so many different agencies involved. (The representatives from B. C. were unfortunately not able to participate in this discussion.)
Israel (Angel) - recent draft recommendations for monitoring environmental changes are based on observable benthic variables (diver logs). There is presently a small industry but there is growing recognition of the need for environmental observations of impacts (related to tourism).
France (Bacher) - models are not widely used for regulation in mariculture in France. Interest is primarily in predicting shellfish carrying capacity. Observations are usually in areas where bivalve culture has been established for many years. Variables of interest are usually related to water quality, not benthic impacts. The point was raised that tropical countries were not represented at this meeting where different environmental problems impact aquaculture development. Shrimp aquaculture production, for example, may be negatively affected by population-related increases in land erosion.
S. Africa (Monteiro) - three government departments (plus industry) are involved in licensing and development of a small but growing finfish aquaculture industry. New coastal zone management project to be initiated to understand interactions of relevant processes in an embayment to be used by the industry. Multiuser committees established to allocate use of environmental space. The concept of coastal zone management is well accepted in S. Africa, and sustainable development may actually become a constitutional principle.
This round-table discussion was followed by a discussion of topics that the participants felt should have been covered, but were not. These included:
Cripps: The concept of sustainability was not mentioned.
Bacher: Models should not assume that variables will always change continuously, and the possibility of catastrophic changes (sensu Thom) or destabilisation of the system must be considered.
Angel: There were no talks on the use of macrofauna or other species as indicators of impact.
Milligan: No papers were presented on far-field effects, although a cancelled paper by Robin Anderson would have addressed this topic.
Chandler, Cripps: The scope of topics was limited; for example, there were no papers on pathogens.
Silvert: Certain types of models were not presented at the Workshop even though they are widely used, such as Gowen-type models of deposition
It was also agreed that there was no mention of important interactions associated with mariculture in the tropics, such as those between mariculture and agriculture.
In addition, the following comments were received from scientists at the Stirling Laboratory in the UK, who were unable to attend the Workshop.
From: Malcolm Beveridge <m.c.m.beveridge@stir.ac.uk>
I've been thinking quite a lot recently about sediments and aquaculture. I feel that the waste dispersion models, although not perfect, are sufficient for management purposes, although I'm not sure about the new, high digestibility diets which seem to induce a near diarrhoealike state in salmon. I also don't know what effect high dietary lipid content has on pellet stability, on sinking rates or on the sediment microbial community.
These points aside, the key question remains sediment capacity and I think the work coming out of Canada at present is at the forefront of this. There has been some comparable work carried out in Israel where Angel and colleagues have looked at carbon assimilation rates. As I remember, they suggest values of around 4.5 g-C/m2/day as being assimilable - and I think this is without anoxia being induced - which is around four times greater than has been suggested for temperate coastal areas.
From: James Muir <j.f.muir@stir.ac.uk>
I suppose the themes which would interest us in general would be:
the validation of simplified site index criteria
'natural' and stimulated recovery systems
scalar linking, and interactions between various subuses of larger scale systems
pathogen/stock/therapeutant/environment interactions
the development of more comprehensive environmental capacity indexes, and their linking with economic rent (e.g. in simple terms, the comparative value of different sites)
social and policy linkages to physical/environmental features.
As for the Recent Advances comment, what would be really valuable would be a good scientific/practical update on the subject of site assessment, monitoring and management, either from a local (though typical/generalisable) viewpoint, or a wider perspective covering a range of production systems and locations. I'd be pleased to help knock out an outline/contents list if useful, and would also be happy to consider a related alternative.
The final item on the agenda was discussion of recommendations to the ICES Working Group on Environmental Interactions of Mariculture. There was some overlap between items which were conclusions requiring further action on the part of modellers, and those which were seen as formal recommendations to the Working Group, so both categories are included under this general heading.
The following conclusions and recommendations are based on material presented at the Workshop and on discussions on the final day. However there has not been time to circulate this report to the participants for review, so the Chairmen assume final responsibility for any possible distortion of the opinions of the other participants.
Modelling results should be taken into account when regulatory guidelines are developed.
Model verification is important not only as part of the modelling process but also as an essential aspect of establishing the credibility of models
We have made and continue to make strides in our ability to understand and predict fishfarm impacts useful for management. Further work is required to implement new knowledge into the development of regulatory guidelines. Examples of the positive use of model results for regulation can be cited. We are now able to identify variables (Eh, pH, S=, gas fluxes) that can be used to characterise degrees of carbon loading and predict changes in benthic conditions (for example, presence/absence of sulfur bacteria mats).
The nations with the strongest approach to coastal zone management (Norway, S. Africa) have incorporated modelling approaches. Findlay countered that the B. C. experience, where there are strong regulations for coastal zone development, is that this has been detrimental to aquaculture development and that the issues are dealt with in a political forum rather than through modelling and other scientific approaches. It was pointed out that there is a history of extreme confrontations exist between environmental conservation groups, governments and industry in B. C., which may make the situation there exceptional.
New methods and technology development are needed to reduce waste effluent releases to the environment (capture wastes for treatment).
Modern database programs with graphical user interfaces (such as Microsoft Access) should be considered as an alternative to traditional programming tools for rapid development of user-friendly interfaces between modellers and clients.
Science used for management purposes must provide confidence limits around predictions. Problems arise when 'predictions' limit the siting of new farms or expansion of existing ones and their accuracy is called into question.
Comparisons with feedback are needed between what is predicted by a model and what is actually observed under different environmental conditions. There must be continued improvement in models that reflect increased understanding and knowledge.
We need to provide better delivery mechanisms for providing advice and incorporating model information into management practices.
The Workshop was restrictive in scope. It did not address many important shellfish culture issues, although this was deliberate, since a complementary meeting will be held in 1996 emphasising environmental interactions of shellfish mariculture. It also did not deal with environmental issues that are important in many developing countries, probably reflecting the fact that it was directed primarily at ICES countries.
Some important areas including the release of therapeutants, the spread of disease, and the genetic consequences of farm organisms escaping and mating with wild ones, were not discussed even though they clearly fall within the scope of the Workshop. Given the size of the Workshop, it is clear that a comprehensive meeting on the Environmental Interactions of Mariculture would have to be much larger, longer, and would probably require parallel sessions.
Modelling should be viewed as an essential component of understanding and controlling the environmental interactions of mariculture.
Modelling should be dealt with in the context of coastal zone management, and not as an isolated field.
There should be a focus on the elimination of effluents before they are released into the environment rather than on dealing with them once they are released, through increased emphasis on nutrition and treatment of waste water.
A Workshop should be held in two years with an emphasis on interfaces to management and on the delivery of model results to clients.
A symposium should be convened covering a broader range of the environmental interactions of mariculture. This should have a global orientation and should not focus on issues in the ICES countries.
There is a need for a continuing network for communication between scientists working in this field. The mailing list that was used for planning the Workshop was recognised as a prototype for such networking, and it has already been expanded into a permanent mailing list for modellers and other researchers. Thus this recommendation has already been informally implemented, but its endorsement by ICES is requested.