480 



Fishery Bulletin 103(3) 



area. As a result, the estimate poststratified by habitat 

 and fish density was greater than the estimate post- 

 stratified by habitat, but lower than the unstratified 

 estimate. According to our results, it is unlikely that 

 the estimates poststratified by habitat and fish density 

 were the most representative estimates of abundance 

 because poststratification adjusted for the disproportion- 

 ate distribution of samples between areas. 



Another reason to poststratify the data is to increase 

 the precision of abundance estimates. Poststratified 



estimates in our study were generally more precise 

 than unstratified estimates, given sufficient sample 

 sizes (Table 5). Poststratification by habitat character- 

 istics increased the precision of abundance estimates 

 in three-quarters of all species-year combinations. This 

 finding indicates a close link between habitat type and 

 fish abundance and agrees with poststratification re- 

 sults in other studies (Pollock et al., 1994; Reilly and 

 Fiedler, 1994). Estimates poststratified by both habitat 

 and fish density were also generally more precise than 



