Department of Medical Pharmacology, Rudolf Magnus Institute for Neurosciences, Utrecht University, Utrecht, The Netherlands
The automated tracking of single animals in various experimental settings has recently become quite common in the fields of ethopharmacology and neurosciences. Automated tracking has several distinct advantages above direct observation: long observation durations, reliability and the saving of time. In addition, automated tracking offers a set of parameters representing the behavioural state of an animal which direct observation cannot. An example of such a parameter is an accurate measurement the distance an animal moved. The image-processing techniques to track multiple individually identified animals can be subdivided into two categories. One approach reduces the apparent size of the object which has to be tracked by painting half of its body in the background colour of the observation arena. Using a size criterion, the image-processing software can identify each individual. The number of animals that can be tracked using this technique is limited to 2 animals. An alternative approach, using colours to identify individual rats, is presented in this paper.
The spatial coordinates representing the position of each animal can be used to address a myriad of experimental questions. The different categories of data analysis can be arranged into three levels: parameters concerning a single animal, parameters pertaining to a single animal relative to other animals and parameters expressing the location of animals in relationship to their environment. The coordinates of dyads of animals allow the researcher to answer questions such as the amount of time animals spent in proximity of each other, the distance at which animals were located from each other, which animal followed another and conversely which animal was chased by another. Zones in the observation arena that are of experimental interest can be investigated, like a home-base location or access to a feeding bin or water bottle. A large number of parameters can be calculated for these zones, for instance time in zone and latency to zone. An alternative method to graphically present data is a plot in which the frequency a rat was located at a certain coordinate is represented with a colour-scale with blue for low frequencies to red for high frequencies. Using these techniques observations were performed on groups of animals housed in a colony situation.
In the first series of experiments either a female or male intruder was introduced during 45 minutes in a colony of four male rats. To validate the performance of the automated system, EthoVision, direct observations were performed concurrently with the automated observations. The amount of time each animal spent interacting with the intruder as obtained through directed observation was correlated with various parameters calculated by the automated system. The automated system could discriminate between the behavior of residents towards a male or female intruder, using parameters such as distance between subjects, movement to and movement from.
In a second series of experiments animals were housed during 21 days in a colony with a resident of the territorial and aggressive Tyron Maze Dull S3 strain while control animals were housed in a similar colony with a Wistar resident. The residents were released for either 20 minutes, 3 hours or 8 hours a day. The changes in the behavior of the experimental animals were assessed using direct and automated observation.
Automated observation systems could be used in the future for more refined detection and registration of the actual behavior an animal is performing. Stereotypic behaviors like circling can be detected using the position of the center of gravity of an animal. However, new parameters, such as shape of the animal under observation or orientation of its body axis, are required to recognize more complex behaviors.