over the 24-h period siicges ti-d ;i tidal In f 1 nonce . 'I'lic tidal period 

 observed during both studios was 12 h, A harmonic regression as 

 described by Lorda (1983) using terms of sin(hours) and cos(hours) over 

 a 12-h period with slack ebb occurring at hours and 12 and slack high 

 at hour 6 was fit to log-transformed (n/500 m^ + 1) data (Fig. 14). The 

 harmonic regression accounted for about 45% of the total corrected sums 

 of squares (TCSS) with the two sampling dates combined. Based on this 

 model, collection densities increased on a flood tide with a peak prior 

 to slack high and then declined during ebb tide. Analyses of 

 covariance, with tidal effect as described by the sine-cosine function 

 as the covariate, was used to examine sampling data and day-night 

 effects (Table 15). An interaction term for sampling date by day-nipht 

 effect accounted for less than 1% of the TCSS and was pooled with the 

 error. The two sampling dates were significantly different and 

 accounted for an additional 19% of the TCSS. In agreement with the 24-h 

 plot (Fig. 13), the day-night effect was not significant. Weinstein et 

 al. (1980) reported that three post-larval fish taxa (spot, Atlantic 

 croaker, and Paralichtys spp. flounders) used vertical migration in 

 response to tides as a retention mechanism in the Cape Fear River 

 estuary. The day and night differences in frequency of Stages 3 and 4 

 larvae at stations B, NB, and EN (Fig, 10) showed that winter flounder 

 larvae of these developmental stages were capable of vertical movements. 

 At station C the lack of diel differences for Stages 3 and 4 larvae 

 suggested a modification of behavior in response to tidal currents. The 

 vertical movement from the bottom during during a flood tide would act 

 as a retention mechanism in the Niantic River. 



Table 15. Summary of analysis of covariance for 24-h dlel study with harmonic 

 components of the tidal effect used as covariates. 



Sum of squares Z of total 



4A.6 *'' 

 19.4 * 

 3.5 ns 



— Includes both sine and cosine components 



* - significant at p £0.05 

 ns - not significant 



40 



