466 
Fishery Bulletin 99(3) 
results with those for other species. One can then use this 
comparative evidence to make speculations concerning ob- 
servations from the field. We used cultured zooplankton 
as live prey for the larvae. Although this is not the natu- 
ral prey for larval witch flounder, it offers desirable char- 
acteristics necessary for experimentation, such as uniform 
shape, size, and swimming speed and although prey densi- 
ties used in our experiments were typically higher than 
averages reported from the field (Myers et al., 1994), these 
levels were necessary in the laboratory to promote growth 
and survival (see Houde, 1978; Puvanendran and Brown, 
1999). 
Experiment 1 : Growth and survival 
Six 33-L rectangular black tanks were used for the rearing 
experiment. All tanks were kept in a water bath to mini- 
mize temperature fluctuations and were supplied with fil- 
tered (25 pm) seawater. Each tank was fitted with two air 
stones that provided light aeration to promote a homoge- 
neous distribution of prey. The light level at the water sur- 
face was 200 lux and continuous lighting (24 h) was used. 
The temperature ranged from 10° to 15°C (mean 12.6°C). 
On day zero, newly hatched larvae were transferred to 
each experimental tank to achieve a stocking density of 
six larvae per liter. Three replicated feeding treatments 
were chosen: 2000, 4000, and 8000 prey per liter. Previous 
results in our laboratory for other northwest Atlantic spe- 
cies suggested that this range of prey densities was suffi- 
cient to observe the effects of prey density on larval growth 
and survival (Puvanendran and Brown, 1999; Laurel et 
al., in press). 
Rotifers (Brachionus plicatilis) enriched with culture 
selco (INVE, Belgium) or Artemia franciscana nauplii en- 
riched with DHA selco (INVE, Belgium) or Algamac (Bio- 
Marine, Hawthorne, CA), were used as prey for the larvae. 
Prey densities were determined by sampling 5-mL ali- 
quots from different depths within the tanks (below sur- 
face, mid-depth, and above bottom) and were adjusted as 
needed to maintain nominal densities three times a day 
(around 10 AM, 4 PM, and 10 PM). Microalgae ( Isochrysis 
and Nannochloropsis) were added to the experimental 
tanks prior to each feeding. 
At week 5, the larvae were transferred to larger 65-L 
circular tanks because they had grown too large for the 
smaller tanks; all other protocols remained unchanged. 
The rearing experiment was stopped at week 12. At this 
point most larvae had begun eye migration but were still 
pelagic. 
Data collection Larvae were sampled weekly for standard 
length (SL, measured from tip of snout to posterior end of 
notochord). For weeks 0-3, larvae were measured to the 
nearest 0.1 mm with a dissecting microscope. Standard 
length was measured to the nearest 1 mm after week 3. 
On weeks 1, 5, 8, and 12, five larvae were lethally sampled 
(killed by an overdose of MS-222) from each tank for deter- 
mination of SL, body height (BH, myotome height posterior 
to anus), and dry weight (DW ). These sampled larvae were 
kept in beakers on ice and measured immediately after 
death to prevent shrinkage from dehydration. Larvae were 
rinsed in 3% ammonium formate, placed on preweighed 
aluminum foils (weighed to nearest 0.001 mg), dried at 
55°C for at least 48 hours, and reweighed. For all other 
weeks, the SL of ten live larvae per tank was measured. 
The absolute growth rate was calculated according to 
the equation 
G = (L t — Lq)H, 
and the length-specific growth rate (SGR) was calculated 
according to the equation 
SGR = (ln(L,) - ln(L 0 )/t) x 100, 
where L t = the mean final length (mm); 
L 0 = the mean initial length; and 
t = the period of growth (days; Busacker et al., 
1990). 
All tanks were examined for mortalities twice daily from 
day 14. Dead larvae decomposed too quickly to be observed 
prior to this time. At the end of the experiment, the num- 
ber of surviving larvae in each tank was recorded. 
Data analysis For each growth measurement (SL, BH, 
DW), a mean value was calculated for each replicate and 
this value was used in the analysis. Growth measure- 
ments were analyzed by treatment using a model I anal- 
ysis of covariance (ANCOVA, Zar, 1999) with age of the 
larvae (weeks after hatching) as the covariate (a=0.05). 
Dry weight data was logarithmically transformed to sat- 
isfy the assumptions of the ANCOVA. A one-way analysis 
of variance (ANOVA) was used to test for differences in 
survival at the end of the experiment. 
Experiment 2: Behavior 
Larvae were stocked into a 250-L cylindroconical upwell- 
ing tank on day zero. This tank served as a general rearing 
tank for larvae used for behavioral observations. The light 
was continuous (24 h) at an intensity of 200 lux at the 
surface. Preliminary results with witch flounder larvae 
showed that this light intensity and light regime resulted 
in good growth and survival (Rabe, 1999). The tempera- 
ture ranged from 4-14°C (mean 12.4°C) during the 8-week 
experiment. 
Feeding began on day 1 after hatching. Rotifers or Ar- 
temia nauplii, or both, were used as prey for the larvae 
and were enriched with commercial products as described 
previously. Larvae were fed three times daily at 4000 prey 
per liter (prey/L). The prey density in the rearing tank 
ranged from 0-4,000 prey/L throughout the day. Micro- 
algae ( Isochrysis and Nannochloropsis) were also added 
twice daily. 
Data collection Behavioral observations were conducted 
every 3-4 days from week 2 to 8 beginning on day 8. The 
prey densities used in the feeding trials were 250, 500, 
1000, 2000, 4000, 8000, and 16,000 prey/L. Prior to the 
