Rabe and Brown: Behavior, growth, and survival of Glyptocepholus cynoglossus lan/ae in relation to prey availably 
467 
Table I 
Definition of the modal action patterns (MAPs) observed in developing witch flounder larvae, after Barlow ( 1968). 
MAP 
Definition 
Locomotory MAPs 
Swim: 
Turn: 
Nondirected MAPs 
Pause: 
Sink: 
Shake: 
Foraging MAPs 
Orient: 
Fixate: 
Lunge: 
Forward movement of the larva through the water column resulting from undulations of the caudal region. 
A rapid lateral bend initiated by the head results in rotating the body approximately 180°. 
Larva is motionless (similar to “non-swimming” of Munk, 1995). 
Larva is motionless and descends through the water column, often head first. 
Rapid lateral undulations of the entire larval body. 
The head movement towards a prey item (similar to “orientation” of Brown and Colgan, 1985). 
The larva is stationary and bends its caudal region into an “S” shape position, typically follows orient MAP 
(Braum, 1978). 
The larva moves towards prey from the fixate position in an attempt to capture prey (Braum, 1978). 
first daily feeding, larvae were arbitrarily selected from 
the rearing tank and placed in 2-L glass bowls containing 
the appropriate density of prey. For each treatment, a 
total of ten larvae were serially observed for two minutes 
each by using the focal animal technique (Altman, 1974). 
After all larvae had been observed, they were returned 
to the general rearing tank. The daily order of observations 
on prey density treatments was varied systematically over 
the study period. The light intensity was 200 lux during 
observations. On observation days, standard length was 
measured on 12 live larvae from the rearing tank. 
Larval behaviors were categorized into modal action 
patterns (MAPs, Barlow, 1968). During observations, the 
frequency and duration of the following MAPs were re- 
corded with an event recorder: swim, turn, pause, shake, 
sink, orientation, fixation, and lunge. Individual MAPs are 
grouped into three classes: locomotory, nondirected, and 
foraging (Table 1). 
Data analysis The effects of prey density and larval size 
(mm standard length) on MAP frequency and duration 
were analyzed with a model III ANCOVA (Zar, 1999), with 
size as the covariate (a=0.05; the fixed factor was prey 
density and size was random). Both size and age will influ- 
ence the larval response to prey density. We chose to focus 
on the effect of size (as a surrogate for structural char- 
acteristics), rather than age (or experience) because this 
factor likely has a greater effect on the foraging response 
to prey density during this early life history stage. 
The response variable represents the mean value for the 
10 individual larvae per prey density for each observation 
day. Means for each treatment were weighted by the inverse 
of the standard deviation (SD) around that mean in the AN- 
COVA. In cases where the SD for a treatment was zero, the 
mean SD for that MAP (for all prey density-size combina- 
tions) was used to weight the mean for that treatment. 
A linear model was used to describe the MAPs. For the 
turn, pause, sink and shake duration analyses, only data 
for the size range prior to the near decrease or disappear- 
ance of that MAP were used, in order to satisfy the as- 
sumptions of equal variance of the ANCOVA. For the swim 
duration analysis, only data for the size range prior to the 
larvae spending most (90%) of their time swimming were 
used. Most behavioral response variables were logarith- 
mically transformed to satisfy model assumptions. Plots 
of residuals and predicted values were examined for het- 
eroscedasticity and normality for each test, and ANCOVA 
assumptions were satisfied. 
The orientation MAP frequency data could not be easily 
fitted to a linear or polynomial equation and were analyzed 
differently. An ANOVA was used to determine the effects of 
prey density on orientation frequency within the size range 
where orientation frequency was variable between treat- 
ments ( 10.5-20.8 mm). A Tukey test was then used to de- 
termine which treatment means differed (a=0.05). 
We compared the lunge frequency of early (<2 weeks) 
and late (>6 weeks) stage witch flounder larvae to that of 
yellowtail flounder (Pleuronectes ferrugineus) and Atlantic 
cod ( Gadus morhua ) to examine differences in prey con- 
sumption rate between species. Yellowtail flounder were 
observed at 8000 prey/L because this prey density pro- 
motes good growth and survival for this species (Rabe and 
Brown, 2000). Similarly, the data used for Atlantic cod 
were taken from the behavioral observations conducted at 
4000 prey/L presented in Puvanendran and Brown ( 1999). 
The 4000 prey/L level was the prey density that optimized 
growth and survival in that experiment. 
Results 
Experiment 1 : Growth and survival 
At hatching, the mean standard length of larvae was 5.62 
mm (±0.12 mm SE). The standard length, dry weight, and 
body height of larvae did not differ between prey density 
