70 



Fishery Bulletin 90(1). 1992 



morphological and environmental variables. Assessing 

 difference matrices using these techniques represents 

 an alternate method with which to evaluate covaria- 

 tion of geographic patterns. For Postorbital W. (Fig. 

 8A), there are nine significant associations using the 



Mantel test (Table 8), with Sea 

 Surface Temp. (Jan.) being the 

 highest. The seven environmen- 

 tal variables displaying correla- 

 tions in Table 7 with Postorbital 

 W. also are judged concordant 

 using the Mantel test. In addi- 

 tion, based on interblock differ- 

 ence values, there are statistical- 

 ly significant associations with 

 the two measures of solar insola- 

 tion (variables 4 and 5; see Table 

 8). The concordance with these 

 two environmental variables is 

 primarily on the strength of pat- 

 tern similarities in the eastern 

 portion of the range. Postorbital 

 W. also shows significant associa- 

 tions with the first two environ- 

 mental principal components. 



Based on correlation tests for 

 block means (Table 7), L. Ros- 

 trum (frm. Pterygoid) exhibited a 

 geographic distribution of mean 

 block values that was statistical- 

 ly associated with those for six 

 environmental measures. Four 

 significant associations were 

 identified using the Mantel test 

 (Table 8), only two of which were 

 found by both tests (Sea Surface 

 Temp. [Jan.] and Oxygen Mini- 

 mum Layer [Depth]). It has a 

 significant association with en- 

 vironmental component II. 



In Table 8, a total of seven 

 significant associations of differ- 

 ence values are recorded for W. 

 Internal Nares with environmen- 

 tal variables, including the three 

 listed in Table 7 as having sta- 

 tistically significant associations 

 based on means. Difference 

 values for environmental com- 

 ponents I and III significantly 

 covary with those of W. Internal 

 Nares. 



Using difference values, W. 

 Premax. (at 1/2 L.) has only a 

 weak correlation with a single 

 environmental variable (Table 8); only a single signifi- 

 cant association was found using correlations of mean 

 values, and this was with another environmental 

 variable (see Table 7). W. Premax. (at 1/2 L.) is weak- 

 ly associated with environmental component I. 



