336 



Fishery Bulletin 92(2). 1994 



12, thermocline depth (summer), has 

 strong positive correlations with a 

 large number of variables, particu- 

 larly those reflecting measurements 

 in the anterior portion of the skull. 

 Given its covariation with water 

 depth, it is not surprising that ther- 

 mocline depth (winter) has signifi- 

 cant correlations with principal com- 

 ponent I and canonical variable 1. 

 Surface dissolved oxygen (variable 

 13) has only a few weak statistical 

 associations with morphological char- 

 acters. 



Environmental principal compo- 

 nent I (Fig. 9A) has a pattern simi- 

 lar to those for sea surface tempera- 

 tures in January and July (variables 

 6 and 7). The highest correlation 

 (0.733) of this component is with 

 width of temporal fossa (Fig. 10A). 

 The second environmental compo- 

 nent (Fig. 9B) is strongly associated 

 with numerous characters (Table 7), 

 reflecting the general trends from 

 the northeast to the west, southwest, 

 and south. The third component, 

 which is negatively associated with 

 canonical variable 2 (Table 7), has 

 only one strong association with a 

 morphological variable, that being 

 with tooth width (r=-0.680; Table 7). 

 The third environmental component 

 exhibits decreasing values as one 

 moves away from the equator. Tooth 

 width shows an opposite pattern, 

 which is particularly emphasized 

 with the relatively thick teeth in 

 specimens from the Hawaiian Is- 

 lands (block 0802). 



In Table 8, Mantel ^-values and 

 matrix correlations are provided for 

 associations of environmental vari- 

 ables (including environmental prin- 

 cipal components) with the five 

 morphologic characters selected for 

 inclusion during the canonical-vari- 

 ates analysis. With this approach, 

 covariation patterns are assessed on the basis of 

 difference values between all block pairs. Preorbital 

 width shows a strong association with water depth, 

 the two measures of solar insolation, the sea surface 

 temperatures in January and July, oxygen minimum 

 layer (depth), and surface salinity (Table 8). It also 

 exhibits a pattern that is statistically associated 



with all three environmental components. The two 

 measures of the temporal fossa show concordance 

 with patterns for solar insolation (January, as well 

 as annual), all sea surface temperature measures, 

 oxygen minimum layer (depth), surface salinity, and 

 the first environmental principal component. The 

 length of temporal fossa also has a weak statistical 



