I08 ECOLOGY AND LIFE HISTORY OF THE COMMON FROG 



Since the relationship with wind direction was obviously curvilinear, 

 a further attempt was made on the problem by a multiple curvilinear 

 correlation method, using temperature, rainfall and wind direction as 

 the independent variables. The result was unsatisfactory, for early in 

 the analysis it became obvious that these variables were only minor 

 factors. The value of including the results here can be questioned, and 

 is discussed in Chapter lo. Appendix i^, but after considering their 

 suppression, I decided that, poorly based as they are, they may be the 

 best so far available. Fig. 27 shows the curves obtained by the analysis. 

 The rainfall curve suggests that once there is a certain amount of rain, 

 more has no effect on migration, but it may be that the effect shown 

 really measures the probabihty that it is raining at the time of migration. 

 Light rain may have fallen in the morning only, but heavy rain very 

 likely goes on all day and into the evening. The temperature curve 

 shows no sign of a threshold value. It is common to read in accounts 

 of amphibian behaviour some statement to the effect that "Activity 

 has never been seen below a temperature of :\:°." It is doubtful if such 

 statements have any real value. If a curve such as that in Fig. 27 

 expresses the results, then activity will get more and more rare the 

 lower the temperature, but there may be no real end-point at all. The 

 supposed limit will then merely depend on the quantity of experience 

 brought into the discussion. For example, in this figure no migration 

 is recorded below i°C, but I have seen frogs migrating in a snowstorm, 

 and there is a record of them migrating over the snow in the Pyrenees 

 (Wolterstorff, 1904). There is presumably some point at which 

 movement stops, but it is probably not much above freezing point, 

 and the attempt to fmd the lowest point of a curve nearly asymptotic 

 to the axis is not likely to be rewarding. We may be looking at the 

 tail of a distribution curve, in which the probability of fmding a 

 migrating frog at decreasing temperatures gets lower and lower. 



It is possible that the reason for the difficulties in the analysis was 

 that the main cause was not in the data under examination, but it 

 could be merely because only a small part of the migration — that over 

 the road — was being considered. Migration goes on for a long time, 

 often for weeks, and if movement from the winter quarters goes on 

 when wind, temperature, and rainfall are suitable, it does not at all 

 follow that frogs from a long way away will reach the road on that 

 night. On the other hand, if a large number do get near the road, but 

 are halted by daylight before they cross the road, they may complete 



