24 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



siphons clogged or failed. Each compartment 

 ■was served by a fresh- water tap. 



In June 1957 we acquired a number of surplus 

 assault boats from tlie U.S. Army, four of which 

 were installed at the laboratoiy and used as brood 

 and fry-rearing tanks. These undecked plywood 

 boats measured approximately 4 by 12 feet on 

 the bottom and each had a capacity of about 840 

 gallons. Although of very light construction, 

 tliey were fairly watertight for more than a year. 

 "Water flowed in through a hose attached to the 

 stem: the drain consisted of a plastic siphon at- 

 tached to the bow. Several 1-inch holes were 

 drilled in one side of the square bow just above 

 the water line for controlled overflow if the siphon 

 failed. Figure 2 sliows an assault boat converted 

 into a fisli-liolding tank, while figure 3 illustrates 

 the plan of the pilot plant. 



A filter system, consisting of a sand filter box 

 and a pump, was installed experimentally on tank 

 2. It soon became apparent that the filter box 

 was not adequate and that beach sand was in- 

 appropriate as a filtering medium. Cleaning and 

 backflusliing of the filter was necessary two or 

 three times a week, requiring considerable time 

 and etfort. The sand was eventually replaced with 

 several layers of fine-meshed Monel screen, which 

 was not a very effective filter but did remove large 

 amounts of fecal matter and other detritus. The 

 turbulence and the splash on tlie surf ac« caused by 

 the water as it was returned to the tank under 

 pressure also increased the oxygen concentration. 



Figure 2. — An as.sault boat cnnverted into a flsh-liolding 

 tank, Kevvalo jilant. 



Figure 3. — Plan of the tilapia pilot plant. 



Otlier minor modifications in tlie pilot plant in- 

 cluded installation of an aeration system and flood- 

 liglits on tank 3. Tlie aeration system consisted 

 of an air compressor, a rubber air hose, and a pipe 

 (drilled witii lioles) that ran crosswise of the floor 

 in tlie center of tlie tank. Air was pumped 

 tlirough this pipe and slowly bubbled through the 

 water in the tank. 



Two 150- watt projector floodlights were in- 

 stalled on tank 3, one at each end, approximately 

 5 feet above tlie surface of the water. It was 

 hoped, by day-and-night illumination, to increase 

 the algal content of the tank, wliicli was very low, 

 and also possibly increase the production of 

 yoimg. No changes were detected, however, and 

 the floodliglits were removed after 3 months. 



Kewalo Plant 



In July 1958, the redwood tanks (three brood 

 tanks and one fry tank) at tlie pilot plant were 

 dismantled and reassembled at the Kewalo Basin 

 docksite. A filter system (fig. 4), consisting of a 

 sand filter box 24 inches wide by 26 inches long 

 and 18 inches deep and a pump, was attached to 

 each of the three, brood tanks. Tlie fry tank was 

 modified by increasing the width 10 inches and by 

 partitioning it with plywood separators (instead 

 of screens) into six compartments. Essentially, 

 each compartment was a separate tank with its 

 own drain and tap. A filter box, 30 inches on the 

 sides and 18 inclies deep, and a pump were in- 

 stalled to filter and to recirculate the water. 



The four assault-boat tanks at the pilot plant, 

 after being in use for about a year, were not worth 

 salvaging. Thirteen assault boats were removed 

 from storage and converted into 12 brood tanks 

 and 1 filter tank at the Kewalo plant. Drain 

 water from the tanks was carried down a flume 

 to the sand-filter tank and pumped back to each 

 tank. All filter boxes contained a bottom layer 

 of crushed rock and a top layer of coarse black 

 sand (volcanic cinders), which was found to be 

 much more effective than the fine beach sand used 

 initially. 



