MUSICK: COMPARISON OF THE HAKES 



\ 



"^r.X 



/ 



72° 



3r 



□ U. ttnu.»,N-S, 

 A U i»nui5. P.E.I. 

 # U Chun 



A 



66" 





O 



'O 





^tt-._,^ '" - ^( Uy NOV 



:^ 



n 



Figure 3. — Localities where Uropliycis were collected for a study of meristics and mor- 

 phometries. U. chuss was compared with samples of U. tenuis from New England (N.E.), 

 Nova Scotia (N.S.), and the Gulf of St. Lawrence (P.E.I.). 



smaller than about 600 cm were made with 

 fine-pointed dividers. Larger specimens were 

 measured with calipers consisting of a meter 

 stick and sliding brass "jaws." When possible, 

 measurements were made on the left side of 

 the specimens. 



The number of fin rays in the second dorsal 

 fin and in the anal fin, the number of abdominal 

 vertebrae, and the total number of vertebrae 

 were each subjected to an analysis of variance 

 (Snedecor, 1956:246). Abdominal vertebrae are 

 those anterior to the caudal vertebrae. An 

 analysis of variance of the number of lateral 

 line scales was performed only among the three 

 samples of U. te)iHis because the ranges of 

 variation of U. chuss and U. te)niis do not 

 overlap. The gill rakers were counted on the 

 epibranchial of the first gill arch. The raker 

 found at the joint between the epibranchial 

 and ceratobranchial was not included in the 

 count. 



The meristic data have been summarized in 

 Figures 4 to 8, and the method of graphing 

 follows that of Hubbs and Hubbs (1953). For 

 each sample the range of variation is shown by 

 a heavy horizontal line, the mean (x) by a 

 small narrow triangle. The blackened part of 

 each bar comprises two standard errors of the 



mean {2s^) on either side of x. One-half of each 

 black bar, plus the white bar at either end, 

 outline one standard deviation (.s) on either 

 side of x; s indicates dispersion; 2s-, reliability 

 of .7. 



Tests of significance were performed at the 

 1% level, and if no difference was found, at 

 the 5% level. Statistical significance is desig- 

 nated as follows: NS = not significant; 

 * = significant at the 5% level; ** = signifi- 

 cant at the 1% level. If significance was found, 

 Duncan's Multiple Range Test was conducted 

 at the 1% and 5% probability levels (Duncan, 

 1955, 1957). The results of Duncan's test in- 

 clude within the same parentheses those samples 

 which are not significantly different. 



The effect of allometric variation on char- 

 acter validity in Urophycis has not been con- 

 sidered in the past. In the present study, an 

 attempt was made to measure specimens of 

 all sizes. All values were transformed to loga- 

 rithms to reduce the correlation between the 

 variance and the mean (Mottley, 1941), and 

 regressions were computed for each character 

 for each of the four samples. Every sample 

 was tested against each of the others by an 

 analysis of covariance (Steel and Torrie, 1960). 

 Variances were tested for homogeneity. Hetero- 



481 



