596 



Fishery Bulletin 104(4) 



model were sampling date (day of year), sampling time 

 of day (sine-transformed), station distance from coast- 

 line (km), surface temperature (°C), surface salinity 

 (%o), Secchi depth (m; a coefficient of light extinction 

 or water clarity), water depth (m), and cross- and long- 

 shelf wind speed (positive east and north, respectively). 

 The significance level for entry into the regression mod- 

 el was set at P<0.05. Environmental variables that sig- 

 nificantly affected the spatial and temporal distribution 

 of bluefish density and mean size (i.e., those parameters 

 incorporated into regression models) were also analyzed 

 with an analysis of covariance (ANCOVA) model, with 

 sampling date as the covariate and station transects as 

 the discrete explanatory variable. 



Synoptic analysis of habitat use 

 by summer-spawned bluefish 



Several other sampling programs initiated in 1998 moni- 

 tored the distribution and size composition of YOY blue- 



fish across multiple habitats in southern and central New 

 Jersey (Table 1, Fig. 1) and the data from these programs 

 were then compared to our data from surface waters on 

 the inner continental shelf. A complete description of the 

 sampling protocols and schedules for monitoring YOY 

 bluefish in these habitats is provided elsewhere (Able et 

 al., 2003; Rowe et al.'). Habitat use by different onto- 

 genetic stages was ascertained from the size and stage 

 (larval, transitional, juvenile) composition of bluefish 

 observed in estuaries, inlets, ocean beaches, and the 

 inner continental shelf. Accordingly, length-frequency 

 histograms of bluefish were created for each habitat over 

 four time periods (August, early to mid-September, mid- 

 to-late September, and from October to November) by 

 using class intervals of 5.0 mm SL for Methot trawl and 

 beach seine samples, and 10.0 mm SL intervals for otter 

 trawl samples. Moreover, a length-based key derived 

 by McBride and Conover (1991) and adopted by others 

 (Juanes and Conover, 1995; Wilber et al., 2003) was 

 used to delineate between spring- and summer-spawned 



