Williams and Ralston; Distnbution and co occurrence of Sebastidae off California and Oregon 



845 



species and listed in Table 4 closely follow the ordination 

 plots. In particular, dimensions one and three in combination 

 showed clear segregation of the eight distinct clusters (lower 

 panel in Fig. 12). Clearly, the A- 1 group in Table 4 stands out 

 as a distinct assemblage in the MDS ordinations (Fig. 12). 

 Because dimensions one and three were highly correlated 

 with depth and latitude, respectively, we concluded that the 

 species groups are defined by depth and latitude. However, 

 the low amount of variance explained by the dimensions, the 

 small sample sizes for some species, and the relatively low 

 average silhouette measures in the partitioning analysis, 

 would indicate that some caution should be exercised when 

 using these results. In fact, there appeared to be some dis- 

 crepancies in depth-latitude distributions and partitioning 

 groups for the rockfish species (Table 4, Fig. 11). 



Group A-1 in Table 4 represents the deep-water slope 

 species of rockfish, and the species in groups B-2 and B-3 

 represent the nearshore species of rockfish. The separa- 

 tion of halfbanded rockfish from the nearshore group sug- 

 gested by the ^ = 8 clustering was likely the result of a 



more southerly distribution for this species than that for 

 the other members of the group (Fig. 7). The remaining C 

 and D groups in Table 4 represent shelf species, group C 

 being a southern shelf group and D representing a north- 

 ern shelf group. A likely misclassification in the clustering 

 results was the inclusion of greenspotted rockfish in the 

 northern shelf assemblage, which was not warranted by 

 the distribution of catches shown in Figure 9. Within the 

 C group, the separation of cowcod as an isolate in the k = 

 8 analysis was due to this species' relatively deep distribu- 

 tion (Fig. 8). The separation of canary and yellowtail rock- 

 fish as a distinct cluster was the result of their relatively 

 northern distribution and their inclusion in the southern 

 shelf species group by the partitioning analysis was prob- 

 ably the result of a boundary effect (Fig. 9). Lastly, the 

 division of the northern shelf group into D-7 and D-8 sub- 

 groups, suggested by the k = 8 partitioning, was likely the 

 result of differences in the survey's total catch of rockfish 

 (MDS dimension 2), which is not evident in distributional 

 patterns (Table 4). 



