268 



Fishery Bulletin 102(2) 





"T .^ 



Nonvegetated bottom (SNB) 



Marsh edge (ME) 



Submerged aquatic vegetation (SAV) 



Figure 4 



Spatial distribution of Galveston Bay bottom types used in the multivariate 

 regression model. 



Krumgalz et al., 1992; Garrison, 1999) and we tested 

 for possible interactions between the variables. Only the 

 interaction between bottom type and salinity zone yielded 



statistically significant results. ANOVA results for the 

 model including the bottom-type and salinity-zone interac- 

 tion term (Table 1) and variable coefficients (Table 2 (fitted 



