526 



Fishery Bulletin 94(3), 1996 



nual and sex-stage effects on growth rates were 

 tested with an Analysis of Covariance (ANCOVA) 

 with the following model: 



Length = Year Sex Year*Sex Date Year*Date 

 Sex*Date Year*Sex*Date, 



with the SAS GLM procedure (Freund et al., 1986; 

 SAS Institute Inc., 1988), where year and sex are 

 class variables and date is a covariate. The test of 

 the main effects (year and sex) represents a test for 

 differences in y-intercepts, whereas the test of the 

 interaction with the covariate represents a test for 

 heterogeneity of slopes (Freund et al., 1986). Growth 

 in terms of body weight was similarly tested. 



Habitat use patterns 



A statistical comparison of smooth dogfish abun- 

 dances among marsh creeks was made by using an 

 ANOVA, based on night-time weir sampling in 

 Foxboro, Schooner, and New creeks conducted dur- 

 ing 1989. We previously reported a comparison be- 

 tween day and night abundances of smooth dogfish 

 in Foxboro and Schooner creek sampling during 1988 

 as part of a study of diel variation in marsh creek 

 faunal composition (Rountree and Able, 1993). De- 

 scriptive statistics, based on night-time gill-net sam- 

 pling in 1990, were used to compare catches between 

 marsh creek and adjacent bay shoal habitats as well 

 as between ebb- and flood-tide stages. Because time 

 of day, tide stage, and seasonal effects could not be 

 partitioned in the 1990 gill-net sampling, no attempts 

 were made to statistically test hypotheses of habitat 

 or tide stage differences in catches. For example, gill- 

 net sampling was biased between habitats because 

 nets stretched across creek mouths were presumably 

 more efficient than those placed in the open bay. Ad- 

 ditionally, because tide and diel cycles are not corre- 

 lated, sampling on flood and ebb tides were some- 

 times conducted weeks apart (i.e. it was impossible 

 to sample on both flood and ebb tides at the same 

 location and time). 



Foraging habits 



Smooth dogfish collected in 1988-90 during gill-net, 

 hook-and-line, and trawl sampling were measured 

 live, packed in ice, and transported to the laboratory 

 for freezing. Because of the large number of smooth 

 dogfish captured during 1990 gill-net sampling, a 

 subsample of at least three fish was retained from 

 each net check for food habit analysis. The remain- 

 ing fish were tagged and released. The released fish 

 were tagged to prevent bias in the catch estimated 



from recaptures. Fish were tagged just posterior to 

 the first dorsal fin with individually coded yellow T- 

 bar anchor tags (total tag length, 40 mm; Hallprint 

 Pty. Ltd., Holden Hill, South Australia). Because no 

 tagged individuals were recovered during our gill-net 

 sampling, CPUE adjustments were not necessary. 



In the laboratory, thawed smooth dogfish were 

 remeasured and weighed prior to stomach removal. 

 The total stomach contents were then weighed and 

 the prey items were identified to the nearest taxon, 

 enumerated, and weighed (wet WT). An index of gut 

 fullness (%full) that incorporates body weight rather 

 than gut capacity (Hyslop, 1980) was calculated as 



%full = [(prey WT)l(total body WT - total gut WT)] 

 x 100, 



where prey WT - the weight of a given prey species 

 (or sum of all prey species); 

 total body WT = the total weight of the smooth dog- 

 fish; and 

 total gut WT = the weight of the entire gut con- 

 tents (excluding the stomach it- 

 self). 



Tidal and temporal (hour of the night) effects on 

 stomach, large intestine, small intestine, and total 

 digestive tract fullness were examined from 1990 gill- 

 net samples. After having been thawed in the labo- 

 ratory, the entire digestive tract was excised and sec- 

 tioned into stomach, large intestine (spiral valve), 

 and small intestine (colon and rectum). The total 

 contents of each section, as well as the total for all 

 sections combined, were then weighed. After pooling 

 data from all collections for each tide stage, hourly 

 mean fullness values were calculated for each gut 

 section separately, as well as for the total. The effect 

 of tide stage (ebb vs. flood) on gut fullness indices 

 was tested separately for each section of the diges- 

 tive tract and for the total digestive tract, with analy- 

 sis of variance (ANOVA; Sokal and Rohlf, 1981). Use 

 of the fullness index, described above, allowed us to 

 reduce bias due to variation in dogfish size in the 

 ANOVA. The data were arcsine-square-root trans- 

 formed prior to analysis. 



Results 



Life history stages and seasonal abundance 

 patterns 



Most smooth dogfish collected in the study area were 

 YOY according to size at date of capture (Fig. 2) and 

 according to an assumed length of 80-90 cm at ma- 



