However, in the space of two weeks, apart from on the large central lawn, only a few people with

dogs and small children were observed to go off the footpaths – these people represent a tiny percentage of the total number of humans walking around the complex GSK-3 inhibition (∼25 000 residents plus others who can walk through the complex). For every focal individual approached (animals that were not running: i.e. foraging slowly or motionless), we recorded whether the animal fled or not (‘yes’ or ‘no’). Animals that did not flee were recorded as having an FID of 0 m (the minimum distance from the observer passing by 2 m away). The proportion of animals that fled was compared by Pearson’s χ2 analysis, with expected values calculated assuming an equal proportion of animals either fled or did not flee across all the treatments. Separate analyses were also carried out to test for the effects of approach trajectory and direction of attention. The observer also measured: (1)  Start distance – the distance from observer to squirrel when the observer began moving towards it. Distances

were measured by dropping coloured pen lids at each point and recording the distances afterwards (with a practiced, measured pace; ±1 m). Although previous studies have indicated that start distance may influence FID, we had no significant difference in start distance (averaged 13 ± 4 m; range 7–27 m) for our four treatment categories (median test, χ23 = 1.86, P = 0.603) Autophagy activator and start distance was not significantly correlated (Spearman rank–order correlations) with alert distance, FID or distance fled for any of the four treatments. We therefore MCE公司 have not further considered the effect of start distance on FID (which allowed us to use one-way non-parametric analyses, which were required on account of the nature of the escape behaviour data). Alert distance and FID were not normally distributed because

of the high proportion of animals (57%) that did not flee the approaching observer. These data were therefore analysed by non-parametric median test for the four approach treatments for each measure (alert distance, FID and distance fled) independently followed by post hoc multiple comparisons (using Kruskal–Wallis H-tests). If focal animals are used to people, as in the present case, and a tangential approach has been used, then animals can be passed-by without them moving at all (e.g. Bateman & Fleming, 2011). We have recorded these animals as having an FID and distance fled of 0 m (although the FID values could also be scaled up, making the minimum value 2 m). Individuals that do not move are central to the data on the repertoire of responses seen in habituated animals, and therefore should not be omitted or ignored. Using non-parametric analyses allows the analysis of these data, where attributing a minimum value to these animals will not affect their ranking in the calculation of statistically significant differences.