Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 



481 



Rock sole 



□ Unstratified 



 Poststratified(H) 



 Poststratitied(D) 



1991 1992 1993 1994 1995 1996 



Yellowfin sole 



D Unstratified 



 Poststratilled(H) 



 Poststratified(D) 



1991 1992 1993 1994 1995 1996 



Pacific halibut 



D Unstratified 



 Poststratified(H) 



 Poststratified(D) 



1991 1992 1993 1994 1995 1996 



Flathead sole 



1991 1992 1993 1994 1995 1996 

 Year 



Figure 5 



Three standard error estimates of annual total abundance. Standard 

 error estimates are for the unstratified, poststratified by habitat 

 (poststratified [H]), and poststratified by habitat and fish density 

 (poststratified [D]) estimates. 



unstratified estimates but were not consistently more 

 precise than the estimates poststratified by habitat 

 alone. The six cases in which estimates poststratified 

 by habitat and fish density were the most precise show 

 that some species have strong density gradients within 

 habitat areas and that the incorporation of fish density 

 information from neighboring years can be beneficial for 

 increasing precision. Being able to predict the distribu- 

 tion of fish density in one year from that of neighboring 

 years indicates annual consistency in species distribu- 

 tion in relation to habitat characteristics. 



The present study indicates that when estimating 

 abundance from haphazardly sampled data, the estima- 

 tor poststratified by habitat is superior to the unstrati- 

 fied estimator regardless of sample size. The estimate 

 poststratified by habitat was more precise than the un- 

 stratified estimate in 18 of the total 24 species-year 

 combinations. These 18 species-year combinations oc- 

 curred across nearly the full range of habitat stratum 

 sample sizes, from 12 to 45. The six cases in which the 

 estimate poststratified by habitat was less precise than 

 the unstratified estimate were affected by the propor- 



