gradient, thereby influencing the ability of fish larvae to feed upon them. To deal with these
problems, a testing apparatus was designed that consisted of a clear acrylic "wheel" partitioned
into six chambers to hold seawater, sediment, fish and brine shrimp (Artemia franciscana)
nauplii. Each chamber had a mean volume of 2.733 liters and had two access ports with
expandable plugs for filling and draining. During a test, the wheel was slowly rotated on the
vertical axis on a motorized base to maintain the sediment in suspension and to continuously
alter the direction of light.
Fishes used in the experiments included spot (Leiostomus xanthurus), pinfish (Lagodon
rhomboides), Atlantic croaker (Micropogonias undulatus), Atlantic menhaden (Brevoortia
tyrannus), and flounder (Paralichthys spp.). Larval and young juvenile fishes were collected on
flooding tides using a 945-m mesh net with a 1- 2-m opening and live-box attached to its
terminal end. Fishes were held in flow-through tanks and were fed freshly hatched brine shrimp.
Four total suspended sediment (TSS) concentrations (20, 200, 2000, and 20,000 mg/L) were
created by blending the appropriate weight of pre-wetted kaolin clay with aerated and filtered
seawater to form a slurry before each trial. Freshly hatched brine shrimp nauplii were used as
prey in all of the experiments during the first phase of the study. Natural plankton assemblages
were used in subsequent experiments to see if fish reacted differently to prey they were more
likely to encounter in nature. Marine plankters were collected just prior to their introduction into
the test apparatus. Three levels of brine shrimp concentration were used (1.0/ml, 0.1/ml, and
0.01/ml). An additional prey concentration of 0.001/ml was used for the experimental trials that
used natural plankton.
Clay and prey concentrations were randomly assigned to the 12 chambers of the two wheels.
Each wheel chamber was partially filled with filtered seawater while it was in a horizontal
orientation. The assigned clay slurries were then poured into the chambers followed by the prey.
Each chamber was filled to within 1-2 cm of the top with filtered seawater. Dissolved oxygen,
temperature, and salinity were measured in one randomly selected chamber from each apparatus.
A fish that had been held without food for 24 hr was then introduced to each chamber. The
access ports were plugged and the two wheels were placed in a vertical orientation on top of a
motorized roller system powered by a variable-speed motor. The wheels were set in motion at
3.5 rpm for 1 minute to resuspend any settled silt-clay, and then slowed to 1.75 rpm for a 30-
minute feeding period. At the end of the feeding period, both wheels were removed from the
motorized roller and the contents of each chamber were strained through a dip net to recover the
fish, which were processed for gut content analyses. Fish prey consumption was recorded as 0, 1-
9, 10-99, or 100+ prey items consumed.
Logistic regression models were applied to the data for each species and prey type. These
analyses allow estimations of probabilities that a fish will feed under different levels of turbidity
and food concentration. The parameter estimates for the models also provide a convenient means
of making comparisons between species concerning the relative importance of turbidity and prey
concentration in determining feeding success. The logistic regression models were fit to the
logarithms of the sediment and prey concentrations.