2.4. Ethical note
Since the study did not involve any invasive procedures, no
ethical approval was required by French national legislation. The
birds were checked at least once daily and any injured birds were
removed to a single pen or culled, depending on the severity of the
wound.
2.5. Statistical analyses
The two experiments were analysed separately using R 3.0.2 statistical
software (R Development Core Team, 2014). All data from
scan observations were analysed with the scan as the statistical
unit, with the repetition of scans as random effect at the cage level
(the cage was the experimental unit). DB bout durations were analysed
using DB bout as statistical unit with cage as the random
effect. Data fromcontinuous observations were summed atthe cage
level.
In the two experiments, we first analysed the effect of main factors
(group size and feed–litter in Experiment 1, wheat-bran litter
and PSA pad in Experiment 2) in the entire cage, then in each area
of the cage (PSA, nest and middle areas). Secondly, we compared
behaviours between each area of the cage. Third, hen distribution
was compared to a homogenous distribution. The last analyses concerned
the location of DB in relation to the trough position.