648 



Fishery Bulletin 97(3), 1999 



500 meters 



Pacific 

 Ocean 



Figure 1 



Location of sampling sites (Bay Entrance, Belmont Shore, and Marine Stadium) in Alamitos 

 Bay, California, U=unvegetated habitats, E=eelgrass habitats. 



except during September 1994 through December 

 1994, when dissolved oxygen values were not taken 

 owing to equipment failure. Bottom water tempera- 

 ture and dissolved oxygen were recorded with a YSI 

 Model 5 IB oxygen meter. Surface salinity was mea- 

 sured with a temperature-compensated refractometer. 

 Eelgrass bed length, bed width, shoot density, and 

 blade lengths were measured by two divers using 

 SCUBA. A 300-m transect tape was laid out in the 

 center of the bed along its longest axis. Width was 

 measured perpendicular to the tape at three equi- 

 distant points. Divers sampled 1/16 m^ quadrats in 

 the bed at predetermined random points perpendicu- 

 lar to the tape. Within each quadrat, all shoots were 

 counted and the lengths of two randomly selected 

 blades were measured. A total of 10-20 quadrats were 

 sampled at each site during August 1992, March 

 1993, December 1993, and December 1995. 



Data analysis 



Nonparametric statistics were used because data and 

 their transformations were heteroscedastic and not 

 normally distributed. A nonparametric two-factor 

 analysis of variance (ANOVA) on ranked data (Zar, 

 1984) was performed separately on the number of 

 California halibut, barred sand bass, and 13 other 



common fishes captured per tow with habitat and 

 site as factors. If significant main effects with no sig- 

 nificant interactions were found, Tukey-Kramer 

 multiple comparisons were performed to determine 

 which pairs of means were significantly different. The 

 results were considered significant if P was <0.05. If 

 significant interactions between habitat and site 

 were found, analyses of main effects were considered 

 dubious and subsets of data were formed for each 

 level of one factor within each level of the other fac- 

 tor and vice-versa (Underwood, 1981). Kruskal-Wallis 

 and Dunn multiple comparisons (Hollander and 

 Wolfe, 1973) were performed on these subsets. For 

 example, a Kruskal-Wallis test was used to deter- 

 mine if there was an overall difference in abundance 

 among sites for unvegetated habitats, and then for 

 eelgrass habitats. If a significant difference was 

 found, Dunn multiple comparisons were made to de- 

 termine which pairs of sites were significantly dif- 

 ferent. For fishes other than California halibut and 

 barred sand bass, comparisons among sites were 

 made only for the habitat where they were most abun- 

 dant overall. Abundance differences between habitats 

 were determined by a Wilcoxon two-sample test on 

 unvegetated and eelgrass habitat data within each site. 

 Two-sample Kolmogorov-Smirnov tests were used 

 to compare length-frequency distributions between 



