416 



Fishery Bulletin 97(3), 1999 



space" term ( DSP), as the difference between the first 

 dorsal insertion (DID less the second dorsal origin 

 (D20). When the analysis was run with this new 

 variable (DSP), it was selected as an important term 

 in the functions (Tables 4 and 9). 



The centroids for yearling fish do not fall into two 

 classes, which would have suggested either a geo- 

 graphic north-south or temporal spring-summer 

 spawned classification (Fig. 1); rather they fall into 

 three clusters of cohorts by year class, regardless of 

 geography of collection site (Table 10). For example, 

 collections during 1987 at Hatteras, North Carolina, 

 and Chesapeake Bay, Virginia (the 1986 year class), 

 are separate and distinct from both the 1988 and 

 1989 collections (1987 and 1988 year classes) and 

 also from the Hatteras and Chesapeake Bay collections. 

 Further, the 1989 collections ( 1988 year class ) were clas- 

 sified into two groups, one of which was the same as 

 the 1988 collection. The 1990 col lection ( 1989 year class ) 

 fell in between 1988 and 1989, overlapping both. 



The centroids of the large fish (>400 mm), a mix- 

 ture of up to 10 year classes in a sample, showed 

 little geographic, temporal, or year-class classifica- 

 tion (Fig. 2). 



