524 ENRLICH, McGOWEN, AND MUSZYNSKI 



sponses. The gradients allowed experimentation with groups and 

 permitted these pelagic fish to swim freely. This apparatus combined 

 some of the best attributes of spatial (e.g., Norris, 1963) and 

 temporgJ (e.g., Beitinger, 1976) gradients. Larvae up to 18 days 

 post-hatching were tested in a l.S-m-long round-bottom gradient 

 5 cm in diameter and filled to a depth of 1.5 cm. This apparatus 

 contains 11 evenly spaced thermistor probes and operates on a 

 countercurrent principle, with hot and cold water entering opposite 

 ends of the outer temperature-controlling trough and the inner 

 experimental one. 



Potentially harmful changes in water quality were avoided by 

 using only six larvae at once and having aerated seawater flowing 

 slowly through the gradient. Seawater velocities of 0.1 to 0.9 mm/sec 

 were selected on the basis of known larval swimming speeds (Blaxter, 

 1969; Rosenthal and Hempel, 1970; Hunter, 1972). Velocities in this 

 range did not present the larvae with any significant difficulty in 

 maintaining a chosen position. Wilson (1974), studying behavioral 

 responses to pollutants, successfully used velocities of up to 

 10.8 mm/sec with pelagic marine fish larvae. Our observations in the 

 present study showed that the flow rates were low enough to avoid 

 any potential rheotactic interference. 



The larger gradient (which is 300 cm long by 15 cm wide by 

 10 cm deep) is static and is controlled by heating and cooling the 

 opposite ends. Surface and bottom baffles, in conjunction with 

 gentle aeration between them, disrupt convection currents. This 

 aeration also maintains saturation levels of oxygen and helps to 

 remove any supersaturated gases, which Gift (1977) reported as 

 potential problems in temperature gradients. Eleven thermistor 

 probes are evenly spaced along the test chamber. 



Both experimental chambers are enclosed in lightproof cabinets 

 with viewing ports. Observations were made from a darkened room. 

 A temperature differential of at least 10° C was established before 

 data were collected for determining thermal preference. 



Low levels of lighting, based on minimum intensities for larval 

 feeding (10 to 15 ft-c) and schooling (60 to 70 ft-c) (Blaxter, 1970), 

 were used during the experiments on larvae and juveniles, respec- 

 tively. These levels of illumination did not appear to disturb the fish 

 as brighter light sometimes did. Additionally, Sullivan and Fisher 

 (1954) reported that temperature selection was more precise at low 

 light intensities. Natural day lengths were used during preexperi- 

 mental holding, as well as during testing. Hasler (1956) reported that 

 fishes could align themselves with small deformities in a tank's 



