Research Training Scholarships
Kenneth Rutherford
“Fractal analysis of behaviour: a non-invasive method for assessing animal stress and welfare”
In 1999 Kenneth Rutherford, with support from UFAW, began his PhD research investigating whether fractal analysis could be used in the assessment of animal welfare. A fractal is made by repeating a simple process again and again, and the result can be infinitely complex. There are many fractal patterns in nature, such as branching fractals in lungs, or spiral fractals in ferns. Fractal analysis is a mathematical technique originating in geometry but it may also be applied to time series data (time series data is a set of observations, results, or other data obtained through repeated measures over time, usually at regular intervals). It involves using fractal descriptions (whether they are spatial or temporal) to identify changes or differences between groups.
Measures of on-going behaviour used in welfare assessment are commonly limited to durations and frequencies; however, these provide only a partial description of behaviour. Extracting additional information from behavioural sequences could benefit animal welfare assessment, for example, fractal analysis may be able to extract an indicator of reduced behavioural complexity and thereby reduced welfare.
Kenneth applied the fractal analysis methodology of Detrended Fluctuation Analysis (DFA) to continuous focal observations of hens and instantaneous scan-sample observations of growing pigs. Kenneth found that the complexity of vigilance behaviour increased in young hens; both during an open field exposure test or following a five-minute period of restraint. The total duration of vigilance was increased in the open fields but not following restraint. DFA measures revealed an alteration in behavioural organisation under stress not identified during standard analysis. DRA also identified behavioural differences between treatment and control groups when a series of mild stressors were applied to groups of growing pigs (although it was unclear if these were directly related to the stressor treatment).
The data sets generated in Kenneth’s experiments were then used to further investigate the DRA method. Analyses showed that alterations in the duration of observation and the frequency of behavioural sampling could affect the end result. Although the analysis had some limitations it did allow novel dimensions of behavioural organisation – not identified during standard analysis – to be measured. These dimensions were independent of total durations of behaviour and they were sensitive to stressful stimuli in some circumstances. Overall, Kenneth concluded that fractal analysis of behaviour showed promise as a tool for measuring stress but further validation was required.
Published papers arising from Kenneth’s project supported by UFAW:
Rutherford, K.M.D., Haskell, M.J., Glasbey, C., & Lawrence, A.B. 2006. The responses of growing pigs to a chronic-intermittent stress treatment. Physiology & Behavior 89, 670-680. DOI: http://dx.doi.org/10.1016/j.physbeh.2006.08.006
Rutherford, K.M.D., Haskell, M.J., Glasbey, C., Jones, R.B., and Lawrence A.B. 2004. Fractal analysis of animal behaviour as an indicator of animal welfare. Animal Welfare. V13 suppl 1: 99-103(5).
Rutherford, K.M.D., Haskell, M.J., Glasbey, C., Jones, R.B., and Lawrence, A.B. 2003. Detrended fluctuation analysis of behavioural responses to mild acute stressors in domestic hens. Applied Animal Behaviour Science. 83(2): 125-139. DOI: http://dx.doi.org/10.1016/S0168-1591(03)00115-1.
