In this thesis, we analysed the circadian rhythmicity of wolf, lynx and jackal
activity using the cosinor method in combination with generalized linear
mixed models. Data from the animals’ telemetry collars were processed into
hourly activity, which was then modelled with a generalized Poisson model
with fixed and random effects. The rhythms of individual animals’ activity
were estimated using fixed-effects models, and the rhythms of the groups
using mixed-effects models. We found that rhythmicity differs primarily between individuals rather than between the two species. In individual animals,
a daily rhythm was present in all seasons with only the amplitude of the oscillation changing, while the timing of the peak remained the same. The
group model detected a rhythm in wolves but not in lynx, even though all
individuals were rhythmic. The reason for this is not an absence of rhythm
in lynx, but rather their differing phases, which cancel each other out when
individuals are pooled into a group. A group average shows what individuals have in common, so with a small sample and large differences between
individuals, an individual-level analysis provides a more reliable picture of
rhythmicity.
|
