DIFFERENCES IN SCALE CHARACTERISTICS 



BETWEEN SEXES 



To determine if the sexes could be combined 

 in later analyses, I compared the scale char- 

 acteristics of scales fronn salmon in the Sj age 

 group from the 1961 Alsek River sample (1956 

 brood year), which included more readable 

 scales than the other samples. Differences in 

 mean numbers of circuli and lengths of radii 

 for 108 males and 177 females were not signi- 

 ficant (t = 0.9225 for numbers of circuli and 

 0.4017 for lengths of radii). I therefore com- 

 bined males and females in the analyses. 

 Clutter and Whitesel (1956) also found no sig- 

 nificant difference between sexes of sockeye 

 salmon for these scale characteristics. 



and the shaded rectangle represents +1 times 

 the standard error of the mean. Where shaded 

 rectangles overlap, differences between 

 sample means are not likely to be significant. 

 In both the Alsek and Anchor River samples 

 the shaded rectangles representing age groups 

 of the same brood year overlap broadly. Dif- 

 ferences between sample means are not sig- 

 nificant at the 95-percent level for any of the 

 eight comparisons shown in figure 2. 1 therefore 

 conclude that the numbers of circuli and lengths 

 of radii through the first annulus are not 

 correlated with total age of the adult, and that 

 for each stream the different age groups of 

 a brood year can be combined to compare 

 fresh-water growth of the scales. 



DIFFERENCES EST SCALE CHARACTERISTICS 

 BETWEEN AGE GROUPS OF A BROOD YEAR 



To determine if fish of different ages could 

 be combined for later analyses, I compared 

 the scale characteristics of samples taken 

 from the Alsek and Anchor Rivers in 1960 

 and 1961. The sample represented two age 

 groups of the 1955and 1956 brood years (fig. 2). 

 Comparisons could not be made with data 

 from the Copper and Taku Rivers because 

 I had samples from these streanns for only 1 

 year. A modification of the graphic method for 

 connparing samples described by Dice and 

 Leraas (1936) is used in figure 2 (and later 

 figures). The horizontal line defines the range 

 of sample observations; the vertical line inter- 

 secting the range denotes the sample mean; 



DIFFERENCES IN SCALE CHARACTERISTICS 

 BETWEEN BROOD YEARS 



The next step was to determine if scale 

 characters of two or more brood years could 

 be connbined for comparisons among popula- 

 tions from different rivers. Differences in 

 numbers of circuli and lengths of radii between 

 years for each system were usually not sig- 

 nificant (the 95-percent confidence limits 

 usually overlapped- -table 1), but the general 

 similarity among the four rivers in the year- 

 to-year changes (fig. 3) indicates that these 

 changes are real and not due to sampling 

 error. Therefore, since the observed range 

 of average numbers of circuli and lengths 

 of radii for the four rivers overlap, I have 

 restricted comparisons between rivers to 

 common brood years. 



H' ^^ 



Sj, ^ 



=2' *— 



"2' ^^ 



(955 



N = I6 



1956 



N:2e5 



CIRCULI (NUMBER) 



16 18 20 22 24 26 2B 30 32 



LENGTHS OF RADII (MM.) 



Figure 2. — Numbers of circuli and lengths of radii from the innermost 

 circulus to the end of the first fresh-water annulus for scales of chinook 

 salmon of 1955 and 1956 brood years, by age class, from the Alsek and 

 Anchor Rivers. The horizontal line defines the range of sample observa- 

 tions; the vertical line intersecting the range denotes the sample mean; 

 and the shaded rectangle represents ± 1 times the standard error of the 

 mean. 



