ponded to the changes at other places through- 

 out the area, since most of the winter storms 

 are generalized. However, local precipitation 

 conditions and such factors as blowing and 

 drifting caused some variation in snow depth 

 from place to place. 



Assuming that the figures as graphed in 

 figure 3 are representative of the entire area, 

 it may be seen from the configuration of the two 

 curves of this figure, representing the amount 

 of snow and the fishing success respectively, 

 that a broad relationship did exist whereby the 

 fishing did tend to become poorer as the snow 

 cover increased. The correlation coefficient 

 was -0.37 (with P somewhat less than 0.01). 



When the same set of figures for amount 

 of snow on the ground was used in computing 

 the correlation with the fishing success at some 

 of the individual localities, the following results 

 were obtained Fountain City Bay, r= -0.64, 

 P= 0.01; Lake City, r= -0 35, P = more than 

 0.1; Slough "22", r = zero; Gremore Lake, 

 r = very small . The differences between in- 

 dividual localities may have been due in part to 

 different effects upon different fish species. 



With the same type of statistical treat- 

 ment, some of the figures for the winter 1945- 

 46 were: total (all pools), r = very small; Lake 

 City, r = zero; Gremore Lake, r = -0.22, 

 P = 0-2. Thus the correlation which apparently 

 held true in the first winter cannot be demon- 

 strated statistically for the second winter. 



Relationships and correlations 



Two successive years. The all-over average 

 fishing success for the two winters 1944-45 and 

 1945-46 was fairly close to identical, being .54 

 and 0.46 fish per hour, respectively. As 

 pointed out above, it is difficult to show that the 

 difference was significant. For the 11 localities 

 of table 16 (not including Winter Creek) for 

 which there were data for the two winters, a 

 mathematical correlation does not exist (r = 

 virtually zero) . The localities, therefore, 

 which furnished good (or poor) fishing in the one 

 winter were not necessarily the ones that fur- 

 nished good (or poor) fishing in the next winter. 



The configuration of the curves repre- 

 senting the week-to-week variation in fishing 

 success for the two successive winteis(figure 4) 

 makes it appear that the trend of changes 

 throughout the winter was similar for the two 

 years. However, a calculation of the correla- 

 tion coefficient (table 22) shows only a weakly 

 significant correlation (r = 0.57; and P = 0.1). 

 Furthermore, for each of two individual local- 

 ities. Lake City and Gremore Lake, the week- 

 to-week correlation coefficient is so small as to 

 have no statistical significance even though there 

 is a suggestion of similarity in the graphs of 

 Gremore Lake for the two winters (figure 5). 



The tendency of the fishing to fall off 

 following the first few days or weeks of the win- 

 ter season is perhaps significant and apparently 

 is the common rule. It may be related to the 

 fact that the snow cover is usually Ug^t at the 

 beginning of the ice fishing period and tends to 

 increase with successive snowfalls. Another 

 factor which may make a contribution is that the 

 fish at the beginning of the winter have not yet 

 acquired the semi -lethargic state which they ap- 

 parently possess later in the winter and therefore 

 they are still doing some feeding. It must be 

 remembered that the mathematical comparisons 

 just referred to are made on the basis of corres- 

 ponding calendar periods and not necessarily on 

 a phenological basis . The onset of winter con- 

 ditions and the course of events within a body of 

 water throughout the winter may vary by a con- 

 siderable period of calendar time. 



Figure 6 presents a comparison of the 

 week-to-week fishing success for the two con- 

 secutive summers. There is no statistical 

 correlation and in deed the trends of the two 

 curves have little or no visual similarity. To an 

 even greater extent than in the winter, the sum- 

 mer conditions are apt not to follow the same 

 calendar sequence year by year; particularly 

 such things as flood periods may occupy dates 

 which are considerably different one year from 

 the next. This in turn has an influence upon 

 other factors, such as water turbidity and water 

 temperature. Therefore, it cannot be reason- 

 ably expected that the best (or poorest) fishing 

 will fall within the same week in successive 

 summers. Essentially the same statement has 

 been made by Frey and Vike (1941). 



42 



