APPENDIX 1 



FISHERY BULLETIN: VOL. 82, NO. 1 



On 24 July, 30 August, and 22 September 1978, 

 vertically stratified samples were taken at one station 

 along the 8 m isobath and at another along the 13 m 

 isobath. A sample set, or profile, consisting of five 

 strata was sampled at each station: Neustonic, three 

 midwater strata, and the epibenthic layer (the mid- 

 water strata were chosen with regard to the depths of 

 power plant cooling structures). At the 8 m station, 

 the midwater strata were 1) the lower 3 m of the water 

 column, 2) 3 m above the bottom, and 3) the water 

 column above stratum 2. At the 13 m station, the 

 lower midwater stratum was the lower 2 m of the 

 water column, while the upper two depended on the 

 vertical thermal structure. When a thermocline was 

 present, as during the September cruise and inter- 

 mittently during the August cruise, the middle 

 stratum extended from 2 m above the bottom to the 

 base of the thermocline, and the upper stratum from 

 the top of the thermocline to just below the surface. 

 In the absence of a well-defined thermocline, the 

 water column above 2 m from the bottom was divided 

 into two equal parts. Sample sets were replicated 

 four to six times in the day and again at night, result- 

 ing in 325 samples in the vertical migration study. 



Data from the two stations were analyzed 

 separately, since all sampling depths (except the 

 neustonic layer) differed between stations. No 

 analysis was done of the effects of the thermocline, 

 since its extreme movements with respect to the ver- 

 tical scale of interest would require a more intensive 

 sampling program. In this analysis nominal sampling 

 depths were treated as constants. 



Because of patchy distributions of ichthyoplankton 

 and movements of the thermocline (August and Sep- 

 tember), inherent variability was expected among 

 the sets of profiles taken on a given date. In order to 

 separate this variability from variability due to sam- 



pling date (cruise), time of day, and "error", we 

 analyzed the data in a repeated-measures type 

 analysis of variance design (App. Table 1). In this 

 design, the depth effect was contained within the 

 fixed-effect time of day and the random-effect cruise. 

 The questions addressed were 1) whether there was a 

 depth effect, i.e., significant differences among 

 strata, within cruise X time-of-day blocks, and 2) if a 

 depth effect did exist, whether there was a significant 

 depth X time-of-day interaction. This interaction, 

 interpreted (when significant) as daily vertical migra- 

 tion, was evaluated as the F-ratio of the depth X time 

 of day to the depth X time of day X cruise mean 

 square errors. When the three-way term was 

 insignificant (in this case, P > 0.75), the error sums 

 of squares and the three-way sum of squares were 

 pooled, and this pooled term was used as the 

 denominator in the F-ratio (Sokal and Rohlf 1969: 

 266). 



The 10 most frequently occurring taxa were 

 analyzed (App. Table 2). (A high frequency of 

 occurrence was important to keep cell variances 

 relatively homogeneous.) To reduce the effect of day- 

 night differences in apparent abundance (most likely 

 from visual net avoidance), we reduced each profile 

 to a set of differences, or A's between adjacent strata, 

 e.g. 



A, = (abundance at depth 1) — (abundance at 



depth 2). 



Abundance was expressed as log,,, (A' +1), where X 

 = larvae/1 00m'. Any daily change in the relative 

 abundance in two strata would thus be manifest in a 

 change in sign and/or magnitude of the correspond- 

 ing A. 



Appendix Table 1 .— ANO VA model applied in the analysis of daily vertical 

 migration. The last two terms can form the error estimate (e) in Appendix 

 Table 2. 



i/ km ' 

 where V, 



+ CT M + CD 



I'* 



+ TD m + DP [mlllJk] + CTD (l/k) + e, lkm 



i /km 

 M 

 c , 



oT 



OP trnt,,lk) 



crb m 



i/km 



Density 



Mean effect 



Sampling date (cruise) effect 



Time-of-day effect (day-night) 



Depth profile within cruise and time-of-day 



Depth effect 



Interaction, cruise X day-night period 



Interaction, cruise X depth 



Interaction, day-night period X depth 



Depth k for profile m within cruise and time-of-day 



Interaction, cruise X day-night period X depth 



Residual error 



110 



