Fig. 3. Changes in cryophilic movement are attributed to changes in cryophilic bias. (A) Scatter plots show the correlation between run orientation and run duration of the detailed crawling trajectories exhibited by individual well-fed wild-type (WT) worms navigating a linear spatial thermal gradient at T>T_S (upper plot) and at T<T_S (lower plot). The starting point of all runs is set to the origin. Each black dot denotes the relative end-point of each run; duration is indicated by distance from the origin (see scale bar) and run orientation is indicated by the angle with respect to the thermal gradient (defined to be 0° for worms crawling up the spatial gradient, shown by the arrow). For wild-type worms tested at T>T_S (cultivated at 15°C and allowed to navigate on a gradient between 18–22°C), runs oriented down the gradient are extended, and runs oriented up the gradient are shortened. By contrast, for wild-type worms tested at T<T_S (raised at 25°C and allowed to navigate on a gradient between 18–22°C), there is no significant correlation between run orientation and duration. Each scatter plot represents run statistics collected from 1000 runs exhibited by 100 worms. (B) Plots of mean run duration as a function of run orientation of wild-type animals, corresponding to the scatter plots in A. Error bars represent 1 s.e.m. Cryophilic bias at T>T_S is represented as prolonged runs pointed down the gradient (grey data points, fit to a constant with P<10^–5). The weak or undetectable thermotactic response at T<T_S is represented as invariance of run duration with run orientation (black data points, fit to a constant with P>0.1). (C) Two measures of cryophilic behavior – the cryophilic bias, calculated using Eqn 1, and the mean of cryophilic migration after 15 min – are plotted for wild-type worms that had been starved for different durations. Errors bars are 1 s.e.m.