634 



Fishery Bulletin 89(4), 1991 



collections with a 38L/min capacity diaphram pump 

 and filtering approximately 0.3-0.5 m 3 through nested 

 63 and 25^ mesh nets on deck. Zooplankton samples 

 were preserved in 5% formalin. The average time be- 

 tween collection of fish larvae and zooplankton was 

 2.5 hours, with values ranging between 21 minutes and 

 4 hours. 



Statistical analysis of diel vertical patterns in larval 

 red drum abundance was conducted using nonstandard, 

 chi-squared, goodness-of-fit procedures (McCleave et 

 al. 1987). First, the standardized residual of larval 

 density at each depth was calculated, 



(Nj - EQ 



SR: = 



where N ; = observed catch at depth i, and E ; = 

 expected catch at depth i. E; was calculated by 

 multiplying the total number of red drum larvae caught 

 during a time period (all depths combined) by the pro- 

 portion of total fishing effort at that depth, i.e., the 

 volume filtered at that depth divided by the total 

 volume filtered during the time period. The sum of 

 squared SRj's yields the chi-squared (x 2 ) statistic for 

 testing the null hypothesis that within a single time 

 period the density of red drum larvae is uniform with 

 depth. A second null hypothesis— that the vertical 

 distribution of larvae remains unchanged over the diel 

 cycle, i.e., among the three sampling periods— was 

 investigated using the chi-squared test for hetero- 

 geneity (a nonstandard, goodness-of-fit test; Sokal and 

 Rohlf 1981). This was accomplished by subtracting the 

 X 2 value for the combined data set (three time periods) 

 from the sum of the individual x 2 's for each time 

 period. 



Results 



Depth-stratified ichthyoplankton collections during 

 afternoon, night, and morning, over a 24-hour period 

 during five cruises, were examined for patterns in ver- 

 tical distribution of red drum larvae. Larvae were ver- 

 tically stratified in both offshore and nearshore waters 

 (Fig. 2). Larvae were usually more abundant at 1 and/or 

 5 m than at 11, 12, or 16 m at the offshore location, and 

 than at 7 or 9m at the nearshore location. In two of 

 the three cruises in which both offshore and nearshore 

 sites were sampled, the depth of greatest abundance 

 (at a comparable time period) was the same at both 

 nearshore and offshore sites (Fig. 2A, D). 



Vertical stratification at the offshore location ap- 

 peared to have a diel component in four cruises (Fig. 

 2A,B,D,E). The most consistent diel pattern was a 



decrease in abundance, relative to afternoon values, at 

 1 m and a relative increase at 5 or ll-12m during night- 

 time hours. The following morning, abundance at 1 

 and/or 5 m was higher than the nighttime values at 

 those depths. During cruise 84-9-2 the diel shift in 

 maximum abundance occurred between 5 and 12 m, 

 with abundance at 1 m remaining relatively constant 

 throughout the 24-hour period (Fig. 2B). Mean density 

 of larvae during cruise 84-10-1 was highest at 12m 

 during both afternoon and night, but in the morning 

 maximum density values were observed at both 1 and 

 16m (Fig. 2C). 



The most distinct pattern of vertical stratification 

 was observed in cruise 84-9-1 where the center of abun- 

 dance of red drum larvae shifted from 1 to 5 m and back 

 to 1 m over the three time periods, with no overlap in 

 mean density or range in densities among adjacent 

 depths within a sampling period (Fig. 2A). Mean den- 

 sity at 1 m in the afternoon was remarkably similar to 

 the mean density at 5m at night, 214.5 vs. 234.5, but 

 by morning of the next day maximum abundance had 

 declined to 141.0. Although vertical stratification was 

 evident in the other three offshore data sets, the diel 

 pattern was less distinct. 



Nonstandard, chi-squared goodness-of-fit analyses 

 were used to test two null hypotheses (McCleave et al. 

 1987): that (1) red drum larvae were homogeneously 

 distributed over the three sampling depths, and (2) the 

 vertical distribution of larvae remained unchanged over 

 three sampling times spanning a diel cycle. In 17 of 18 

 cases, including both offshore and nearshore sites, null 

 hypothesis 1 was rejected at the 0.01 significance level 

 (Tables 1 and 2). In all five data sets from the offshore 

 site, null hypothesis 2 was also rejected at the 0.01 

 significance level, i.e., larval depth distribution did not 

 remain the same among the sampling periods (Table 1). 



The sign and magnitude of standardized residual 

 deviations, SRj values, were examined to determine 

 the diel pattern in vertical distribution of red drum 

 larvae (Table 1). In four of five cruises, the depth of 

 greatest larval abundance was higher in the water col- 

 umn during afternoon hours than at night (Fig. 2). In 

 two cruises (84-9-1 and 84-9-2), the depth of greatest 

 larval abundance on the following morning was the 

 same as on the preceding afternoon and, therefore, 

 higher in the water column than on the preceding night. 

 In one cruise (85-9-1), although the depth of greatest 

 abundance (5 m) in the morning was the same as at 

 night, abundance at 1 m (depth of greatest abundance 

 during the preceding afternoon) had increased relative 

 to the nighttime value. The morning sampling period 

 of cruise 85-10 was the only case where larvae were 

 homogeneously distributed throughout the water col- 

 umn. During the only cruise (84-10-1) in which depth 

 of greatest larval abundance was the same in afternoon 



