Figure 1 Movement of individuals
Classical ecological modelling of aquatic ecosystems is based on a combination of hydrodynamic models and Euler type water quality models. The latter describe the biological and chemical variables as average concentrations within the water volume represented by each calculation node in the hydrodynamic model. The Euler type models are especially applicable for simulating changes in concentrations, e.g. dissolved oxygen, nutrients and micro-algae. An important constraint, however, is that the behaviour and interactions between individuals such as individual movement patterns partly or fully independent of the water current, flocking and emergent phenomena cannot be simulated. For this purpose Agent Based Modelling (ABM) can be applied.
DHI is presently implementing ABM functionalities in DHI’s ecological modelling module, ECO Lab. ECO Lab is an open equation solver applied as an add-on module on top of hydro-dynamic models for simulating biological and chemical phenomena in the aquatic environment. For predefined models all differential equations are fully accessible and editable, while new models can be easily implemented.
Movement
ABM makes it possible to define agents (e.g. pelagic larvae, fish or mammals) and their behaviour (~movement) by defining rules and algorithms describing the speed and direction. Each agent moves according to its own behaviour and the simulated hydraulic current.
The movement enforced by the agent itself can be calculated as a function of the surroundings of each individual agent, by e.g. evaluating the immediate neighbourhood and choosing the direction (and speed) of movement towards e.g. higher food abundances, towards other individuals (~flocking) or away from threats such as low oxygen conditions or predators.
States
For each type of agent (e.g. a fish) it is possible to define an unlimited number of state variables, such as body weight, length, age, lipid content or accumulated xenobiotics. Each state can be described as a function of the other state variables defined for the agent, (e.g. ideally as a physiological model) or as a function of any other state variables included in the hydrodynamic model and/or an Euler type water quality model.
Case study
The Houting definitely belongs amongst the most rare and endangered fish species in Denmark. www.snaebel.dk.
The Houting is a salmonide species living its adult life stages in the sea returning to spawn in the rivers from where it originally hatched.Juvenile Houtings cannot tolerate saline conditions the first few months after hatching, and it is important that they can stay in fresh water until a certain stage. Since the new hatched larvae have only a very restricted moving ability inundated areas with slow water movement such as floodplains are recognised as important habitats for juvenile Houtings.
Figure 2. The Houting
The only significant populations live in the Danish sector of the Warden Sea area and breed in a few Danish rivers. It has disappeared completely from Germany and the Netherlands as a result of the advance of industrialisation and dike building over the last centuries. To save the Houting from complete extinction the Danish Forest and Nature Agency in co-operation with local counties has initiated the EU financed Houting Project. As part of a specific wetland restoration project DHI was asked to evaluate 2 alternative designs of rivers and inundated areas to achieve the most efficient detainment (and thus assumable highest survival) of Houting larvae and fry. A combination of a 2D Hydrodynamic model (MIKE 21 FM) and ABM was used for the analysis.
A simple ABM was setup in ECO Lab based on the best available knowledge on the behaviour and preferences of juvenile Houtings, e.g.:
- The eggs hatch during a 14 days period.
- Growth increase exponential with age.
- Maximal swimming speed is a function of length.
- Death risk is 5% per day initially and reduces with age.
- The larvae seek areas with low current and vegetation.
- The larvae seek water depth of min.35 cm.
- At length above 5 cm the Houting migrates downstream.
In addition random walk was included a.o. to simulate dispersion processes.
MIKE21 FM was used for predicting the water depth and current conditions.
Results
The below figures show the movement of a single larvae (figure 3) and the distribution of larvae (figure 4) at a certain time during the simulation period.
Figure 3. Movement track of a single individual in the project area. Colours indicate water depth.
Figure 4. Distribution of fish larvae at a certain time during the simulation period. Colours indicate Manning’s number
Based on 1000 simulated individuals 2 different designs for the river restoration were compared. In Figure 5 is shown the length of Houting larvae when they leave the project area downstream. This is an indirect measure of the efficiency of the area for retaining the larvae in the critical growth phase.
Figure 5. The length distribution of Houting larvae when they leave towards downstream areas.
Conclusions
Although the calibrated hydraulic models showed that Alternative 1 significantly reduced average water velocities in the project area compared to alternative 2, it was not possible solely based on this observation to conclude that the 1st alternative would result in significantly longer residence time for Houting larvae in the constructed wetland. Only by combining the hydraulic simulation with ABM it was possible to evaluate the existing knowledge on behaviour of fish larvae and the implications in a highly fluctuating hydraulic environment.
(First published in the NZWWA journal September 2008)