Clark et al.: A habitat-use model for juvenile Farfantepenaeus aztecus in Galveston Bay 



267 





Low salinity (April-June) 



Increasing salinity (July) 



&m 





High salinity (August-October) 



Decreasing salinity (November-March) 



Salinity Zone (ppt) 



□ 0-0.5 CZZI 51-5 CD 51 " 15 CZ] 15 1-25 H > 25 



Figure 3 



Galveston Bay seasonal salinity distribution maps. 



to spatially display model predictions in each 10 x 10 m 

 cell. The resulting values for each cell (predicted mean log 

 density) were converted to numbers/m 2 and reclassified 

 into 5 percentiles based on their resultant distribution: 

 0-20%, 21-40%, 41-60%, 61-80%, and 81-100%. All 

 statistical analyses were conducted with JMP statistical 

 software (SAS Institute, Cary, NO. 



Due to difficulties in creating continuous salinity and 

 temperature contour maps in GIS, these variables were 

 classified as follows: salinity was classified by one of the 

 five isohaline zones described previously and analyzed as 

 such to determine its influence on brown shrimp distri- 

 bution; and water temperature was classified by season 

 determined by cluster analysis and analyzed to examine 

 possible temporal effects of brown shrimp distribution. 



Spatial patterns were evaluated by comparing the pre- 

 dicted mean log density values with the observed mean 

 log density values from Galveston Bay drop samples. Addi- 

 tionally, the model's predictive performance was assessed 

 by comparing the predicted mean log density values with 

 observed mean log density values from samples collected 

 in Matagorda, Aransas, and San Antonio bays using the 



same collection method. With this approach, the assump- 

 tion was made that brown shrimp modeled in Galveston 

 Bay respond similarly to the range of biotic and abiotic 

 factors in the other bay systems. 



Drop sample data collected during July-September 1984 

 (/i=128), and April- June 1985 (re=144) from West Bay (ME, 

 SNB) and Christmas Bay (ME, SAV, and SNB) were used 

 to examine bottom-type preference or selectivity. Tukey- 

 Kramer multiple comparisons test was used to compare 

 log density patterns in areas where ME and SAV occurred 

 together and in areas where SAV was not present. 



Results 



Brown shrimp model 



ANOVA and Tukey-Kramer pair-wise comparisons 

 showed significant differences in brown shrimp log density 

 between the three bottom types, five salinity zones, and 

 four seasons (Fig. 5). Multiple regression models were run 

 with these discreet variables (Mahon and Smith, 1989; 



