Introduction
Climate change impacts farmed salmon in open net-pens through stressors like increased temperature, pH and storms. To adapt, the industry needs robust tools that can help to better understand complex interactions between the farm environment and production factors. This study developed a dynamic simulation tool (Virtual Farm) which can serve as a digital representation of salmon production in open net-pen system, allowing for experimentation and analysis of management and climate adaptation strategies. The criteria for comparing between different strategies are set by users based on the chosen performance indicators such as dissolved oxygen (DO) level in the net-pen, standard feeding rate, disease risk period and average daily weight gain by the fish.
Methods
The Virtual Farm model was constructed using Vensim DSS 10.2.1 simulation software. The main components and feedback structure of the model were first captured in a causal loop diagram (CLD), based on information gathered from both stakeholder engagements and the literature. Then following the CLD, the Virtual Farm model was developed using stocks, flows, feedback loops and time delays that are connected using mathematical relationships derived from knowledge and data about real farm system interactions. The differential equations in the model are solved using Euler integration method, with a one-day simulation time step to reflect daily and seasonal dynamics of the model parameters. Data loggers were deployed at three salmon farm locations on the west coast of Scotland and used to collect data on sea temperature and dissolved oxygen (DO) for model calibration and validation.
The model requires the user to input time-series data for seawater temperature, salinity, wind speed, chlorophyll-a, and water current speed as well as initial values for DO and production related factors like stocking weight, to simulate DO dynamics in the farm environment. The farm environment determines how an individual fish in a net-pen would respond in terms of feed intake, metabolism, growth, and then scaled up to the population level to simulate overall implications for farm production (total biomass and potential risk periods).
Results and Discussion
Results show that the modelled DO compare well to measured data at the three farm locations. Further model testing and refinement is underway through an iterative process and consultation with industry stakeholders.
The Virtual Farm model is designed to simulate the effects of multiple stressors (environmental and operational) and can help farmers and researchers to better understand and respond to climate change impacts on salmon production.
Funding: This work was funded by UK Research and Innovation through a UKRI Future Leaders Fellowship (MR/V021613/1).