Besides making use of the sophisticated fish labs in the Department of Collective Behaviour, I have set up my own experimental lab at the Limnological Institute at the University of Konstanz. The key feature of my new lab is the large experimental setup that enables the automatic testing of up to 8 fish simultaneously as well that of large schools of fish to look at both group-level and among-group level processes. The setup consists of a 1.5 m x 2.5 m tank that can be divided into 8 smaller tanks. The top part of the setup contains both visible and IR lighting, thus enabling us to test fish under a large range of light levels, including complete darkness. A total of eight Raspberry Pi computers with carefully calibrated cameras enable the automated recording of the fish. We have large social housing tanks in the basement of the institute and have custom-build individual housing tanks that enable the holding of up to 128 individual fish.
Automated recording and tracking
My research uses a strong mechanistic approach and often requires the testing of large number of individuals to look at the effects of individual heterogeneity in animal groups. Therefore I have developed my own Python packages to easily record experimental trials in a controlled and automated way (pirecorder) and track both individuals and groups of animals (ATracker). pirecorder is publicly available and comes with detailed documentation to help other researchers to set-up and work with a controlled raspberry pi recording system. It consists of a main recorder module and a number of helper methods that facilitate setting-up the the Raspberry Pi, configuring the the camera, scheduling recordings, and converting recorded media.
ATracker is a Python package I have developed to enables the automatic tracking of both individual animals as well as groups of tagged and untagged animals in a range of environments. It provides highly detailed spatio-temporal output of individual and group movement dynamics as well as behavioural state changes. It contains a number of helper methods, including camera calibration, dynamic tracking parameter configuration, interactive error checking, custom video output, and a manual tracker function. ATracker is currently still in beta and not yet publicly available. Both pirecorder and ATracker make use of a large number of utility image, mathematical and general functions from the associated AnimLab package, publicly available here.
All fish we study in the lab are from the wild, caught with a number of different fishing techniques. I have also started to set up a project in the Spanish Pyrenees to explore how fish populations deal with the extreme effects of environmental change using an individual-based approach. Hopefully more about this soon!