Work Package 4

Assessing the applicability of indicators and developing operational tools, e.g. models, and establishing the functional relationship between environment and aquaculture activities. 

Summary

Indicators of aquaculture-environment interactions determined in WP2 and for ecosystem change in WP3 will be tested with a range of pre-existing datasets, held by the partnership for a wide range of culture and environment types, to determine their utility and reliability in the context of the ecosystem approach. The WP comprises the following sequence of tasks:

1. Identification of the most appropriate indicators and models for testing with the available data;

2. The allocation of testing tasks to the most appropriate partners or groups of partners;

3. The testing of indicators and models against pre-agreed criteria of scientific robustness and practical utility; this will include comparisons of models with similar aims and scales;

4. The development of a consensus on which set of indicators and models should be proposed for field validation in WP5; this may include new models synthesised from the best features of models tested in step 4;

5. A re-iteration of the testing procedure using newly collected data from WP5;

6. Publication of a full report on the merits of the chosen indicator set including best methodologies for collection, analysis and interpretation, and on the recommended set of models, including criteria for choice of models depending on spatial scale and farm size, and guidance on the use of models to estimate site and water body assimilative capacity and sustainable production, and on the reliability of model predictions.

The finally evolved indicators and models will be presented as a “toolpack” of guidelines, indicators and tools for improved EIA and site selection for marine aquaculture. This will include a decision-support tool that will present the knowledge gained in the project to guide industry and regulators to the most useful indicators and tools for determining site suitability for aquaculture activities across varying environmental types.

Work Package Aims and Objectives in full.

When aquaculture develops in a new region, the key indicators are those of 'background' environmental conditions (e.g. dissolved oxygen concentrations in the case of salmonids, phytoplankton abundance in the case of shellfish) and the performance of the farmed organisms.  Subsequently, as farms increase in size, they begin to perturb environmental conditions, thus setting up a feedback loop which impacts on the farms themselves as well as other users of the coastal zone. At this stage, all significant components (farms, the physico-chemical environment, the biological communities, and other users) can be seen as comprising  a system described by a set of state variables and subject to significant internal dynamics as well as external forcing by boundary conditions. Most indicators can be dealt with as either state variables themselves (e.g. water transparency, decreased by direct and indirect effects of fish-farms and in turn controlling growth rates for phytoplankton and benthic macrophytes), as the first derivatives of state variables (e.g. primary production, oxygen demand), or as statistics describing these variables (e.g. upper 95%ile of nutrient concentrations). Dynamical mathematical models (and in some cases their equilbrium solutions) are ideal tools for describing, interpreting, predicting and managing such systems.

Such models can represent from few to many state variables, using spatial grids ranging from simple single boxes to a full 3D implementation.  Models with a few state variables take account mainly of biogeochemical interactions, whereas  multivariable models can address issues of biological diversity.  Simple models tend to be cheap and easy to apply, with minimum requirements for local parameterisation over a range of sites.  They can be used as tools for screening sites for initial suitability and approximate carrying capacity, identifying environmental impact 'hot-spots' requiring further study, and for low-cost management of low-impact sites.  Complex models are expensive to implement but may be needed for management of large sites with multiple inputs and which are deemed to be close to large-scale carrying capacity.  Biologically complex models are needed where interactions between numerous types of organisms are important, and physically complex models are needed where patchiness and variable dynamics are issues. They can also be used to improve the parameterisation of simple models.  Finally, most model implementations are designed and parameterised for particular spatio-temporal scales: scale A (local to farms); scale B (well-demarcated water bodies with residence time of days to weeks); and scale C (regional). If well-designed and properly validated, models add considerable value to environmental measurement, allowing interpolation and, in some cases, extrapolation.

A number of different operational tools for use in an ecosystem approach to management of the aquaculture sector have been developed across Europe, but these are at present only validated and implemented in few member states. The following models are known and available to partners in the project - who were in many cases key contributors to model creation and development.  Some were designed to assess particular aspects of environmental impact; others are capable of addressing issues such as food availability for cultivated shellfish, or the potential synergy between finfish farms (injecting nutrient into a water body) and shellfish farms (removing phytoplankton generated by these nutrients).

For particulate organic material and insoluble medicine residues, the DEPOMOD computer model (Cromey et al., 2002a;b) is used operationally by the Scottish Environment Protection Agency.  This model has also been validated for the Eastern Mediterranean through the MERAMED FP5 project. A hydrodynamic model TRIMODENA (Espino et al. 1997, González et al., 1998) has been used with success in several Environmental Impact Studies of Mediterranean Spanish cage culture, (González et al., 2002).   

For dissolved components, including nutrients, simple models are available which consider Equilibrium Concentration Enhancements.  In addition, for occasional discharges of waste dissolved materials, models have been developed that predict their maximum concentration relative to Environmental Quality Standards derived from ecotoxicity tests. The FP5 OAERRE project applied a number of simple and complex models to 'regions of restricted exchange' such as fjords and lagoons, and used these models to diagnose and investigate trophic status as a function of nutrient enrichment and ecosystem response (Tett et al., 2003).  Work has begun on coupling two of OAERRE's 'screening models' - the UK CSTT model and the University of Göteborg's FjordEnv model - into an improved simple model for estimation of assimilative capacity in relation to nutrient and organic loading on the scale of water bodies such as small fjords and bays.

For ecosystem-scale carrying capacity modelling, the EcoWin2000 ecological model has been used in systems in the EU and China, and is currently being applied in southern Africa. Several target organisms have been considered, including mussels, oysters, scallops and shrimp, under different culture conditions (monoculture and polyculture). This modelling approach considers key processes at the ecosystem scale that condition individual scope for growth, and simulates individual growth and population dynamics of cultivated species. Typical outputs include:

• Effects of overstocking on exploitation carrying capacity;
• Yield response to changes in culture practice;
• Impacts on target species of changes in anthropogenic inputs;
• Environmental modifications due to changes in aquaculture pressures (e.g. phytoplankton depletion).

Returning now to the tasks to be carried out in this workpackage, we remind the reader that a key aspect of our ecosystem-based philosophy is that indicators of sustainable aquacultural management are embedded in the <farm -- environment -- biological-community -- other users>  system, and are thus either implicitly or explicitly parts of dynamic models.  WP4 will thus focus on indicators in the context of models as operational tools. It will define and improve indicators and models through an iterative process that commences with the indicators identified in WP2 and WP3 and then interacts with existing and new (WP5) data.  For example, if DO is taken as an appropriate indicator, existing or improved models will be used to investigate change in this variable as a function of hydrodynamic conditions, meteorological and boundary forcing scenarios, as well as of farm and other inputs leading directly or indirectly to oxygen production or consumption.  Validated against observations, the model simulations will allow estimation of variability of DO levels with and without farm inputs, and thus provide guidance on EcoQSs, system assimilative capacity and farmed organism carrying capacity, and monitoring regime, as well as providing operational tools for farm, ecosystem and coastal zone management.

Another, more complex example, concerns the use of biological indicators (such as AMBI, see Borja et al., 2000) or the Infaunal Tropic Index (ITI, Word 1990), used as measures of impacts on the soft-bottom benthic community. Incorporation of such indicators into site-scale or water-body-scale models, could simulate the impacts on benthic diversity and the health of the community.  However, these indicators are in fact a type of multivariate statistic, and at first glance their simulation would require the use of ecosystem models with many biological variables.  Such a study is likely to be beyond the scope of this project, and a more practical method will be to develop a simple parameterisation of the functional relationship between AMBI or ITI and a state variable such as sediment organic content, or a time derivative such as organic sedimentation flux or sediment oxygen demand. Such a parameterisation could then be added to models such as DEPOMOD or FjordEnv.