Specimens of age group were obtained 

 above the dam, while I, and II were found below. 

 An analysis of variance, using the 1953, 1954 

 and 1955 year classes and the three age groups, 

 was run to determine if these specimens could 

 be considered to be from the same population. 

 The resulting F - value based on the total num - 

 ber of gill rakers was highly significant and is 

 probably due to the variation among year classes 

 and age groups (table 16). 



The best way to compare fish having dif- 

 ferent standard lengths is by use of the analysis 

 of covariance (Marr 1955). In order to use this 

 test the regression slopes cf the samples must 

 belong to the same population . The F - value, 

 for the "homogeneity of within sample regres- 

 sion" based on the total number of gill rakers 

 was just significant at the 5 percent level (table 

 17). A test of adjusted sample means made us- 

 ing the total number of gill rakers (table 18) 

 gave an F value which was just significant at 

 the 5 percent level. When dealing with wild 

 populations, it is difficult to know just where to 

 draw the line for significance due to factors 

 that may cause the data to be non -random . 



Similar covariance tests were also run, 

 using the number of gill rakers on the upper arm 

 of the first branchial arch. The test determin- 

 ing the homogeneity of the slopes proved to be 

 barely significant (table 19), while the test of 

 adjusted sample means was not significant 

 (table 20) . As the above values were not highly 

 significant, indications from the gill raker data 

 are that the Santee-Cooper River System speci- 

 mens belong to one population. 



The Hudson River population. - -The most 

 numerous Hudson River collections available 

 are from the 1954 year class and the age group. 

 Specimens are available from Palisades State 

 Park to Coxsackie, New York. When an analysis 

 of variance using the total number of gill rakers 

 was run on these samples, the resulting F 

 value proved to be highly significant (table 21). 

 A similar test, based on the number of gill 

 rakers on the upper arm, was also highly sig- 

 nificant but with a lower magnitude (table 22). 



Based on a purely statistical outlook, the 

 results of these tests indicate that some hetero- 

 geneity of samples exists in the Hudson River. 



However, looking at it from a biological point of 

 view, the magnitude of these differences doesn't 

 seem to indicate more than one population. A 

 chi- square test was performed on the same data 

 from the total number of gill rakers. The total 

 chi-square value was 42.75, which was signifi- 

 cant at the 5 percent level. Bartlett's test for 

 homogeneity of variances was also worked out 

 for the above samples . The resulting value of 

 chi-square (1.01 with 5 degrees of freedom) was 

 not significant. By deleting the Coxsackie sample 

 from the hudson samples and running a chi-square 

 test (X^ = 28.08, with 20 degrees of freedom), 

 the results indicated that these (lower river) 

 samples were homogeneous 



A test was run to compare the samples 

 from Haverstraw to Palisades State Park with 

 Coxsackie, New York; t proved to be just above 

 the 5 percent level of significance (table 23) . 

 This difference was not enough to consider the 

 divergence between upstream and downstream tx) 

 be biologically significant. 



A test using the total number of gill rakers 

 was run for samples of the 1953 year class from 

 Coxsackie and Haverstraw and t was not sig- 

 nificant (table 24). An analysis of variance was 

 worked out for the 1949 year class with samples 

 from Port Ewen, Stoney Point and Haverstraw; 

 the F -value was not significant. The same was 

 true when a sample from Port Ewen was com- 

 pared with those from Stoney Point to Haverstraw 

 (table 25) . 



An analysis of variance was conducted 

 among six year classes (1936, 1949, 1950, 1952, 

 1953, 1954) from the Hudson River and was not 

 significant (table 26) . 



A correlation coefficient was determined 

 between the upper arm and the total number of 

 gill rakers on the first arch for the samples from 

 1954 year class from Hudson River. The result- 

 ing value (r = .885) indicates a good correlation. 



In summary, even though some significant 

 differences were found between samples within 

 the Hudson River, biologically it should be con- 

 sidered as one population. There is an overall 

 homogeneity among the several year classes 

 studied from the Hudson River. The samples 

 from the Hudson River had the hi^est gill raker 



