334 



Fishery Bulletin 92(2). 1994 



1 001 

 1 23. 



1 47i 

 1.701 

 1.931 

 2.171 

 2.401 



2 63i 



2 871 



3 101 

 3.331 

 3571 

 3801 

 4.031 

 4271 



>, *, *, *, «, ' 



'<? O v v v 





Blocks 



\\\\% 



%% 



\x\% , 



'S- ^ ** "■* "<!■ . 

 ? ? ? ? ? 

 u> V V V V 



Blocks 



Figure 6 



Modified skyline diagrams for 29 latitude-longitude blocks, indicating groups formed 

 using function-point clustering procedures and based on: (A) all 30 characters; (B) 

 five characters that, in combination, best discriminate among blocks (preorbital 

 width, width of temporal fossa, length of temporal fossa, length of rostrum [from 

 pterygoid], length of braincase). For given upvalue (i.e. row), blocks connected in 

 common line are in same cluster. 



Morphological-environmental covariation 



Douglas et al. (1992: fig. 9) included a dendrogram 

 summarizing absolute correlations among 13 envi- 

 ronmental variables (listed in Table 2) for blocks 

 having either S. attenuata or S. longirostris or both. 

 These variables were partitioned into five clusters. 

 Sea current (N., winter) is separated by itself, 

 whereas sea current (W., winter) and oxygen mini- 

 mum layer (depth) form a second cluster, which 

 groups with an assemblage of five variables involv- 

 ing surface measures of temperature, oxygen, and 

 salinity (variables 6, 7, 8, 10, and 13). The fourth 

 cluster has the two solar insolation variables (4 and 



5), while the fifth includes three measures indicat- 

 ing water and thermocline depths (variables 3, 11, 

 and 12). 



A principal-components analysis was conducted to 

 obtain variables that would summarize general en- 

 vironmental trends; three components were pre- 

 sented from Douglas et al. (1992: table 6). Highest 

 loadings for environmental variables on principal 

 component I included those for sea surface tempera- 

 tures (variables 6, 7, and 8), particularly July tem- 

 peratures. The correlation with sea surface tempera- 

 ture (annual variable) is negative. The second com- 

 ponent reflected thermocline depth (variable 11 and 

 12), as well as water depth and surface salinity. The 



