Steves et al : Settlement and nursery habitats for demersal fishes 



171 



and winter (December-February). Each environmen- 

 tal variable was standardized to have a mean of zero 

 and a standard unit of variance so that all variables 

 had equal weight despite differences in the scales of 

 their usual units. 



Although seasonal patterns are likely to be impor- 

 tant in juvenile fishes, these patterns are likely to be 

 influenced by the spawning patterns of adult fishes 

 rather than by habitat selection of settling fishes. These 

 seasonal effects may dilute the variation explained as 

 due to other factors. Some environmental factors such 

 as temperature, salinity, and the presence of ephemeral 

 biota that vary with season are possibly of more inter- 

 est than simply accounting for settlement timing. To 

 this end, a partial CCA was conducted using season as 

 a covariable. Partial CCA allows environmental vari- 

 ables to be treated as covariables, thus removing their 

 influence from the rest of the data set (ter Braak and 

 Verdonschot, 1995). 



A forward selection of environmental variables 

 was used to select a minimum set of environmental 

 variables that best explained the distribution and 

 abundance of age-0 fishes (ter Braak, 1992). For- 

 ward selection begins by the selection of the variable 



that explains the most variation in the data set. This 

 variable is then treated as a covariable, the data set 

 is then reevaluated, and the next most important 

 variable is selected. This procedure continues until 

 the desired subset of variables has been selected. 

 To test the advisability of adding variables, Monte 

 Carlo permutation simulations are conducted. Vari- 

 ables are added as long as their addition continues 

 to contribute significantly to the explained variance 

 (P<0.05). Forward selection was performed on the 

 data set, with season treated as a covariable. The 

 combination of partial CCA and forward selection is 

 not considered problematic with regard to the assump- 

 tions of the analysis (ter Braak, 1992). 



Results 



Faunal composition and distribution 



A total of 21,309 fish representing 47 species and 

 32 families were collected over the course of the ten 

 cruises and the 659 tows. Of these 47 species, 33 

 were collected as age-0 juveniles, including 25 spe- 



