Introduction
A detailed focal sampling study of general activity in chickens is helpful to establish the organisation of the different activities performed during the time the chickens spend awake. However, such a recording is very time consuming and difficult to practice in commercial conditions where chickens can not be followed with one's eyes for a long time. The aim of the study was to find out the way several behavioural patterns are organised and to elaborate a simplified method of observation.

Methods
Twenty five broiler chickens were videotaped and observed for one hour per day at 1, 2, 3 and 5 weeks of age using The Observer 3.0. All behaviour patterns expressed when birds were awake were recorded as "activity bouts" (i.e. from the moment a bird stood up until it remained idle for at least 30 s). The 708 activity bouts recorded were analysed by Principal Component Analysis followed by an automatic procedure of partitioning, defining several classes of bouts.

Results

1. Four different classes of activity bouts were obtained by the classification (Figure 1) and differed highly in the mean duration of the bout: SHORT (78 s), M-EATING (M = Medium, 252 s), M-DRINKING (134 s) and LONG (559 s) classes. The bouts of the 4 classes also differed in their occurrence (46%, 16%, 13% and 25% of the bouts, respectively) and in their level of activity (duration of walking per bout: 4 s, 15 s, 14 s, 66 s, respectively). The SHORT class contained short bouts with low activity and mobility, but many preening per iods. The M-EATING and M-DRINKING classes contained bouts in which eating or drinking activi ties, respectively, contributed to the major part of the activity. The LONG class contained long-lasting bouts (almost 10 min), with high levels of activity, especially locomotion and exploration.
2. Age did not significantly affect the SHORT class. There was a large decrease in the occurrence and in the duration of the LONG class bouts throughout the life-span, whereas M-EATING and M-DRINKING classes bouts became more numerous as chicks grew up and represented the major part of the activity at the end of the growing period.
3. A simplified method using bout duration and the first 2 behavioural patterns occurring within the 3 first minutes allowed to predict the class into which each bout would fall with 76.8% accuracy.

Figure 1. Mean duration(s) of the behavioral patterns and total bouts duration in each class.

Conclusions

1. The activity bouts could be categorised into four classes, which differ highly in occurrence, duration and composition.
2. The LONG class contained bouts with high levels of locomotion and exploration. Their occurrence could be used as an indicator of general activity throughout the lifespan.
3. The relationship between the beginning of the activity bout and the entire activity suggests that the type of activity (composition and duration) performed by a chick is planned as soon as it stands up.
4. These results could be used in order to simplify the observation, and to decrease the time spent quantifying the behaviour of chickens.

© 2000 Noldus Information Technology b.v.