T.^LE 2.— \'ariance ratios obtained from analysis of \ariance tests between (i) tJiree different statistical transformations of 

 pheasant abundance as expressed in pheasants obser\ed by rural mail carriers jjer 100 miles of driving (dependent variables) 

 and (ii) 10 different land-use statistics (inde{)endent variables for the 102 coimties of Illinois. 



• Significant at the 0.01 level of probabilitv. 



f Logio of 1 plus the mean number of pheasants observed by rural mail carriers per 100 miles of driving. 



The three dependent variables were (i) the mean num- 

 ber of pheasants per 100 miles, (ii) the square root of 

 the mean number of pheasants per 100 miles, and (iii) 

 the login of 1 plus the mean number of pheasants per 

 100 miles of driving. The log transformation of the 

 mean number of pheasants (iii"! proved to be better than 

 the two other dependent variables because with it more 

 of the variations of data could be explained by the 10 

 independent variables (Table 2) : it was used as the de- 

 pendent variable in all subsequent statistical analyses. 



The total correlation (positive or negative 1 of in- 

 dividual independent variables with the log transforma- 

 tion of pheasant abundance was found to be statistically 

 significant at the 0.01 level (with 101 degrees of freedom) 

 for 7 of the 10 independent variables (Table 3). This 

 finding suggested that variations in 7 of the 10 land-use 

 factors tested statistically were associated with variations 

 in the abundance of pheasants in Illinois. 



The significant correlation between the abundance of 

 pheasants and the percentage of cropland ( r = 0.629, 

 Table 3) demonstrated that pheasants were most abun- 

 dant in counties having a high proportion of cultivated 

 land (Table 1). In 9 of the 10 coimties in which pheas- 

 ants were most abundant, at least 75 jier cent of the 

 land was in crops, w hereas in 20 of 28 soudiem coimties 

 from which pheasants were absent less than 60 per cent 

 of the land was in crops. 



The amount of woodland in Illinois counties varied 

 inversely with the relative amount of cropland and the 

 abundance of pheasants (Table 1 and Fig. 3i. A highly 

 significant negative correlation ( r = — 0.6 1 2 ) was ob- 

 tained between pheasant abundance and the relative 

 sunount of woodland in the counties (Table 3) . The best 

 pheasant counties of Illinois had little or no woodland 

 (Fig. 2 and 3). 



The abundance of pheasants was significantly cor- 

 related (r = 0.4761 with the relative number of cash- 

 grain farms among all farms in the counties (Table 3). 



Cash-grain farms comprised 74 and 71 {>er cent of all 

 farms in Livingston and Ford counties, the counties in 

 which pheasants were most abundant, and more than 

 50 f)er cent of all farms in many of the other counties 

 in which pheasants were numerous f Table 1 ' . The 

 abundance of pheasants was less associated with the rela- 

 tive number of dairy and livestock farms in the countie 

 than with the relative number of cash-grain farms 

 (Tables 1 and 3 ■ . 



In most of the counties within the range occupied 

 by pheasants in Illinois, a hi?h projxji^on of cropland 

 ( referred to as cropland and pastureland in Table 1 i w: 

 devoted to row crojjs, mainly com iZea may5\ and sov- 

 beans (Glycine max). The proportion of cropland 



Table 3.— Test of significance by analv-sis of total correlation 

 for each of 10 independent variables v\ith the log transforma- 

 tion of pheasant abundance* in the 102 counties of Illinois. 



* LA7gi« of 1 plus the mean nunit>er of pheasants otkserved 

 by rural mail carriers per 10<i miles of drivine. 

 t Not significant at 0.03 level of probability. 



' 



