28 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



amined for signs of ova and younfr. The females 

 were also examined to see if ova or younfj were 

 being carried in the mouth. It was noted at tliis 

 time that all the brood stock appeared in excellent 

 condition but there were no signs of spawning. 



On December 5, 1956, one-half cubic yard of 

 white beach sand was placed in two of the brood 

 tanks (tanks 1 and 3), covering the bottom to a 

 depth of about 3 inches. It was our original in- 

 tention to determine if the tilapia would spawn on 

 the bare floor of the tanks, which seemed probable 

 in view of Chen's (1953: p. 7) observation that 

 tilapia were seen spawning successfully in a gar- 

 den pond with a concrete bottom. 



Good evidence of excavating and nest-building 

 was noted the next morning following placing of 

 tlie sand, and the nests continued to increase in 

 number during the next few days until they occu- 

 pied at least two-thirds of the sandy bottom. Two 

 weeks after the sand was placed in the tanks, the 

 first young were noticed on the surface of tank 1. 

 The young were removed and subsequently placed 

 in tank 2, which was drained and converted into 

 a fry tank. Six days later, the fish in tank 3 

 started to produce young. 



During the draining of tank 2 in preparation 

 for conversion into a fry tank, one large male was 

 observed carrying six yolk-sac fry in its moutli 

 cavity, although in this species the female is sup- 

 posed to brood the young. This behavior was 

 also observed by Vaas and Hofstede (1952), who 

 reported that the male incubated the ova in ex- 

 ceptional cases. Further evidence of spawning 

 was found in tank 2, where many ova and yolk- 

 sac fry were seen widely scattered on the tank 

 floor. It was our sujiposition that the adults had 

 become excited as the water level dropped and 

 ejected the ova and fry. 



At the Kewalo plant, where all the brood tanks 

 were supplied with sand, fry production started 

 11 days after the tanks were stocked. Further 

 data on fry production at the Kewalo jjlant will 

 be presented latei- in this report, together with the 

 results of the various experiments. 



FRY COLLECTION 



Newly hatched fry of T. monmrnVtca measure 5 

 millimeters in length, 5.8 nun. on the second day, 

 and S.O nun. at the end of the Kfth day. On about 

 the fifth day, they begin to spend less time in the 



mouth of the female or leave it altogether and 

 swim about in a tight school near the surface of 

 the water, feeding on tiny food particles (Panik- 

 kar and Tampi, 1954). This behavior of the fry 

 made their capture by dipnetting a relatively 

 simple process. Another behavior pattern that 

 became evident to us through daily observation 

 was the tendency for the fry to congregate along 

 the walls of the tank, especially in the corners and, 

 at times, directly under the inflowing water. 



At both plants, the young emerging each day 

 were captured with a fine-meshed dipnet and 

 counted as they were released into the fry tank. 

 Although our daily netting effort varied some- 

 what in efficiency, as indicated by the different- 

 sized young netted, the method proved to be quite 

 effective as evidenced by the few juveniles that 

 were removed from the tanks when they were 

 drained. 



Early in the operation of the pilot plant, the 

 displacement and weight methods for assessing the 

 daily production of young were considered, but 

 both had major drawbacks for enumerating the 

 very small tilapia fry. We concluded that only 

 \iy an actual count could we oljfain the accuracy 

 desired. 



In August 1957, a shelf -collect ion method of 

 capturing young was investigated at the pilot 

 plant. This method was originally the idea of 

 biologists of the Hawaii Division of Fish and 

 Game who observed that tilapia fry tended to con- 

 gregate in shallow water near the edge of the 

 ponds. Consequently, it was hypothesized that if 

 the fry had shallow water available to them along 

 the walls of the brood tank, perhaps their capture 

 would be simplified. 



A redwood shelf, running the length of the in- 

 side wall just below the waterline, was installed 

 in tank 1. The water level in the tank was main- 

 tained so that the outer edge or lip of the shelf 

 was ordinarily about one-half inch below the sur- 

 face. The water over the shelf could be drained 

 through a hole in the wall of the tank and the 

 young fish caught in a net. 



Comparison of the number of fry collected from 

 the shelf with that collected elsewhere in the tank 

 l)y dipnetting indicated that tiie shelf was effec- 

 tive in the removal of only about 25 percent of the 

 fry that emerged daily. Further, many of the fry 

 congregated under the shelf, rather than over it. 



