Long Range Correlations in the Walking Pattern of Japanese Quails in Home Cage and Exposed to Mild Acute Stressors

KEMBRO, J. M.; PERILLO, M. A.; MARIN, R.H.

Angra dos Reis, Río de Janeiro, Brasil

Workshop; Workshop in Mathematical Methods and Modeling of Biophysical Phenomena; 2006

Fractal analysis provides a novel tool to analyze temporal patterns of behavior offering more information than the standard measurements of duration and frequency of a specific behaviour. Long-range correlations in behavioural time series have been well documented. For a reliable detection of long-range correlations, it is essential to distinguish trends of different orders or other periodic oscillatory behavior from non-periodic intrinsic fluctuations in raw data. However this is not taken into account in most of the biological studies that evaluate long range correlations in time series. In the this study we evaluated whether long range correlations were present in the walking pattern of female Japanese quails, and if they were altered on exposure to a mild acute stressor. The walking behavior of birds was evaluated in their home pen without disturbance (control) or submitted to a feeder withdrawal for three hours prior testing (stressed).  First, the size distribution of the length of time intervals spent not walking was analyzed. A power law distribution was observed in all cases and the scaling factors were determined. A negative correlation between the time spent walking and the scaling factor was found.  Second, the whole walking pattern was evaluated with a detrended fluctuation analysis (DFA). Trends in data were systematically investigated and eliminated from the data set by using a detrending order of two. Also, a power spectrum analysis was performed on all data sets to ensure there was no periodic oscillatory behavior affecting the time series. Long-range autocorrelations were observed when time series were submitted to a DFA. Birds that were exposed to the acute stressor had a significantly more negative scaling factor. In addition, they showed a significantly lower auto-similarity parameter (?) when compared to controls. These results highlight the non-gaussian distribution of the waking pattern and suggest that stress decreases the autocorrelation time and increases the complexity of the walking behavior.