FISHERY BULLETIN: VOL. 



Table 8. — Y intercepts in millimeters, slopes, coefficients of determination ir^). 

 and N for regression lines calculated on Enchelyopus cimbnus from six geographi- 

 cal areas. Independent variable is standard length. 



submitted to a Newman-Keuls multiple range test 

 in order to determine which population sample or 

 groups of population samples were different from 

 others. This procedure failed to detect differences 

 between any slopes, a not uncommon occurrence 

 due to the fact that the analysis of covariance is a 

 more powerful test than is the multiple range test. 

 The sample from Iceland had the lowest slope at 

 0.10, the Northern Atlantic, Gulf of Mexico, and 

 Intermediate samples each had a slope of 0.12, the 

 Southern Atlantic sample had a slope of 0.15, and 

 the sample from Europe had a slope of 0.16. 



The hypotheses concerning the equality of Y 

 intercepts was rejected at the 0.0.5 level of sig- 

 nificance for all four characters tested. These re- 

 gression data also were submitted to a Newman- 

 Keuls multiple range test in order to determine 

 which population sample or groups of population 

 samples were different from others. Again, this 

 procedure failed to detect significant differences 

 between any Y intercepts. 



If a more stringent 0.001 level of significance is 

 used, only orbit length tests as being significantly 

 different with respect to overall coincidence. Data 

 for this regression from each of the six samples 

 were submitted to a continuation of analysis of 

 covariance to determine whether differences in 



the regression lines were attributable to the slopes 

 and/or the Y intercepts. We accept equality of the 

 slopes with a probability of 0.85. However we re- 

 ject the equality of the Y intercepts after calculat- 

 ing a probability of equality of 2.06 x 10"''. Re- 

 gression data were submitted to a Newman-Keuls 

 test, which failed to detect differences between 

 any pairs of intercepts. Inspection of our data 

 shows that rocklings from Iceland appear to have a 

 proportionally larger eye than do other rocklings; 

 however, our sample is small and may be biased by 

 larger fishes; hence we do not draw inferences 

 from this apparent difference. 



Although differences between samples appar- 

 ently exist, we do not interpret them as represent- 

 ing the kind of discontinuity that indicates dis- 

 tinct species. Their significance is beyond the 

 scope of this paper. 



Discussion 



We believe that the fourbeard cockling is best 

 considered as a single species throughout its en- 

 tire range. Differences in pigment pattern, meris- 

 tics, and the relative size of several body parts do 

 exist; however, there are neither trenchant dis- 

 continuities in variation nor is there any overall 



96 



