Elsevier

Aquaculture

Volumes 388–391, 15 April 2013, Pages 137-146
Aquaculture

Modelling of Atlantic salmon (Salmo salar L.) behaviour in sea-cages: Using artificial light to control swimming depth

https://doi.org/10.1016/j.aquaculture.2013.01.027Get rights and content

Abstract

Submerged artificial light sources are commonly used to control sexual maturation in farmed Atlantic salmon, but may also be a tool to steer salmon to swim at depths which are optimal for production. In this study, we used an individual-based model of the behaviour of salmon toward environmental variability to simulate the swimming depths of salmon in different seasons, production environments and artificial light regimes. Model outputs agreed with direct observations of salmon swimming depths from literature, suggesting that the model accurately simulated the behavioural mechanisms behind responses toward artificial lights superimposed upon different environmental conditions. We used the model in a series of in silico experiments to predict the behavioural effects of submerged artificial lights placed at different depths in environmental conditions typical for coastal waters in winter, spring and summer. The model indicated that artificial lights controlled salmon swimming depths most efficiently in winter. Further, lights may be more efficient in sites with a more homogeneous environment throughout the water column (e.g. open coast) than sites that are thermally stratified (e.g. fjords). Placing submerged lights at the right depths could produce better culture conditions, ultimately resulting in increased growth. With standard measurements of temperature at several depths as a sole user input, the model could act as a tool to inform farmers of which depths to place their lights on any given day or season.

Highlights

► Mathematical model of salmon behaviour in response to submerged artificial lights. ► Verification of model against observation data. ► Estimates of efficiency in using lights to control farmed salmon swimming depths. ► Potential tool for intelligent placement of artificial lights in salmon cages.

Introduction

Modern salmon aquaculture has advanced from small, shallow cages containing thousands of fish placed at sheltered inshore locations to large, deep cages (157 m circumference, 50 m depth) holding 100,000–250,000 individual salmon in more exposed locations (Fredheim and Langan, 2009). Although larger production units offer considerable economies of scale, they introduce distinct challenges for production management that differ from smaller sea-cages, including feeding control (Talbot et al., 1999), photoperiod control (Hansen et al., 1992), sea-lice management (Costello, 2009) and maintaining optimal oxygen levels (Johansson et al., 2007). To counter such production challenges, cage management strategies customised to large-scale production conditions are required. A key element in developing such strategies is knowledge on how salmon behave in response to the culture conditions.

Within the confines of sea-cages, the behaviours of farmed Atlantic salmon are determined by a complex environment containing spatially and temporally varying factors, such as temperature, light, dissolved oxygen (DO) and salinity. Whereas the full effects of DO and salinity on adult salmon behaviour are still not known (Oppedal et al., 2011a), responses to temperature and light levels drive the vertical distribution of salmon within sea-cages outside feeding periods (Fernö et al., 1995, Johansson et al., 2006). Salmon have distinct ranges of preference for water temperature and light intensity, and regulate these factors behaviourally when they are outside their preference ranges (Dempster et al., 2008, Fernö et al., 1995, Johansson et al., 2007). As both temperature and light vary less horizontally than with depth in the water column, this often results in vertical movement. When the most preferable temperature and light levels occur at different depths, salmon may swim at depths that are a trade-off between these factors (Oppedal et al., 2011a). The distributions resulting from these trade-offs tend to be much denser than the initial stocking density of the cage (Oppedal et al., 2011a, Oppedal et al., 2011b).

The use of artificial light sources is a management strategy originally developed to inhibit fish maturation in cages (Hansen et al., 1992, Oppedal et al., 1997, Porter et al., 1999), but has also been shown to have positive effects on fish growth (Nordgarden et al., 2003, Oppedal et al., 1997, Oppedal et al., 1999, Oppedal et al., 2003). Submerged artificial lights also affect fish behaviour (Juell and Fosseidengen, 2004, Juell et al., 2003, Korsøen et al., 2012, Oppedal et al., 2001). Placing light sources at specific depths could be a strategy to shift fish vertically within the cage to distribute fish more evenly within the cage or to steer fish away from areas in the cage where conditions are sub-optimal for growth and welfare (Oppedal et al., 2007).

Individual-based (or Lagrangian) modelling (IBM) is a technique in which each individual animal in a population is modelled to produce a group-level outcome. This technique has been applied with great success to other livestock industries (e.g. Pomar and Pomar, 2005, Tedeschi et al., 2004) to improve production efficiency through precision livestock farming (Wathes et al., 2008). An IBM of salmon behaviour in sea-cages has been developed and verified against detailed observational data on salmon distributions (Føre et al., 2009). The model accurately simulates the behavioural trade-offs made by salmon between temperature and natural light levels when positioning themselves vertically in a sea-cage. However, the model does not currently include behavioural responses to artificial light sources which are in widespread use by the industry.

We aimed to model the behavioural effects of artificial lights in salmon aquaculture and verify the model using observational data on the swimming depths of salmon in full-scale sea-cages subject to artificial light sources. We expanded the model of Føre et al. (2009) by adding a new model module describing the responses of salmon to underwater artificial light sources, and verified model outputs against observational data provided by Oppedal et al. (2007). Through a series of simulation experiments using the model, we predicted the effects of artificial light placement depth, season, and water column thermal profiles (stratified versus well mixed), to determine which combinations were most likely to yield growth advantages for commercial production.

Section snippets

Materials and methods

The model described below is based on an individual-based model built and verified by Føre et al. (2009) which simulates the behaviours and swimming depths of Atlantic salmon in sea-cages in response to a range of environmental variables. Here, we only briefly explain the main features of the model and detail modifications to adapt the model to account for behavioural responses of salmon toward artificial light sources. The model explanation partly follows the ODD (Overview, Design concepts,

Winter period

The model results matched the results from Oppedal et al. (2007) well in that the predicted vertical distributions of fish were similar to those observed for all three winter scenarios (Fig. 1). In W1, temperatures varied from about 1.5 °C near the surface to approximately 7 °C at the cage bottom, with a sharp thermocline at about 2 m depth, while the maximal surface light levels exceeded 600 μEm 2s 1. At night-time (00:00–05:00 and 17:00–24:00), the majority of the fish were observed to reside at

Model verification: Oppedal et al. (2007)

The experiments performed by Oppedal et al. (2007) were conducted under a wide variety of environmental conditions. Temperature profiles ranged from having the lowest values near the surface (winter scenarios), through being almost invariant with depth (spring scenarios), to featuring clear maximums in the upper layers and distinct thermoclines (summer scenarios). Together with the randomised placement depths of the artificial lights, these variations produced nine different scenarios, all with

Acknowledgements

Funding for this work was provided by SINTEF Fisheries and Aquaculture through the Centre for Research-based Innovation in Aquaculture Technology (CREATE).

References (36)

Cited by (21)

View all citing articles on Scopus
View full text