Tilapia culture is usually characterized by excessive reproduction in the culture ponds. However, the deliberate spawning of large numbers of Tilapia is becoming important due to the advances made in hybridization, genetic selection, and to meet the growing demands of extension work. Cage and pen culture require large numbers of fry, and culture techniques using hormonal sex reversal require controlled spawning.
‘Mass production’ is an imprecise term, and this discussion confines itself to those techniques proven to produce at least 1 million fry per year. Discussion is further restricted to techniques applicable to Africa. It is obvious that the facilities described can be done on a smaller scale, should the need for fry be less.
An understanding of the natural spawning habits of the fish is important. Most tilapia have certain breeding habits in common, particularly nest building and territoriality. Once the fish have spawned, virtually all tilapias exhibit some degree of parental care, but this differs substantially among species. Tilapia are then divided into three groups; Oreochromis (female mouth brooders), Sarotherodon (male or both parental mouth brooders) and Tilapia (substrate spawners, no bucal incubation). These differences are important in the design and management of mass fry production facilities.
Probably the most popular tilapia species for culture is Oreochromis niloticus. For an example of how mass rearing of tilapia fry has evolved and the designs and management techniques now in use, reproduction of this species will be discussed in detail.
O. niloticus males will build a saucer shaped nest in 40 cm of water. While it will spawn in deeper and more shallow waters, in a natural environment, the nests are almost always at this depth. The size of the depression varies considerably with the substrate and the size of the male; a 50 g male may build a nest of only 10 cm in diameter, a 2 kg male may build one of a meter in size. Furthermore, the male quards a territory which may be several meters around the nest, chasing away other fish. The nest is used only for spawning. The male may spawn with one or several females before abandoning the nest.
With this maternal mouth brooder, the female collects the eggs from the nest and leaves the territory of the male. In nature, she will move extensively in deep and shallow waters while incubating the eggs and larvae, finally releasing them in shallow water near the shore. In culture conditions where high densities prevail, the incubating female changes to a yellowish color with vertical dark bars and moves to the centre of the pond or tank where she will not be disturbed by males defending their territories or cannibalistic fry.
The above spawning characteristics are considered in the design of mass production units.
In terms of land surface area, initial cost, ease of operation, and the number of fry produced, one of the most efficient systems for mass production of O. niloticus was done in Cote d'Ivoire in association with a large cage farm (Campbell, 1985). The production unit was in three parts; spawning ponds, first stage fry ponds, and ‘fingerling cages’ placed in an adjacent lagoon.
The ponds were built on laterite soil, the entire site measuring 60 × 150 m, or 0.9 ha. Water surface was slightly less than 0.5 ha. Ponds were constructed on two levels to allow the transfer of fish from one to the other by gravity. The bottom of the higher ponds was 1.5m above the water surface of the lower tier, and the bottom of the lower ponds 1.5m above the surface of the lagoon.
On the upper level, there were 11 spawning ponds which measured 10 × 20 m. This size was chosen as being the most manageable. It allowed territoriality of the males and a large area for brooding females, yet was easy to control and the quantities of fish were not excessive. Water depth was 40 cm over most of each pond, increasing abruptly to 50 cm at the point of drainage. Care was taken to insure that the bottom was properly sloped to insure rapid and complete drainage. The bottom and lower portions of the dikes exposed to water were mechanically compacted and covered with a thin coating of mortar or a cement and laterite mix to prevent nest building by the fish. If the pond bottom were left in the earthen condition, it would soon becomes a series of craters from the nests, making complete draining and removal of fry difficult.
The ponds were emptied through a concrete monk to avoid water over flow, with a PVC pipe 100 mm diameter passing under the front dike and emptying into a concrete channel used both for drainage and water supply to the first stage fry ponds below.
The 9 first stage fry ponds on the lower tier measured 10 × 25 m with an average depth of 60 cm. The bottom was compacted but no cement was used, as there was no nest building by the fry, and the soil/water interaction was useful in increasing pond productivity. Drainage was similar to the spawning ponds, however, the PVC pipes were 150 mm and led directly to the lagoon. Larger diameter pipes were used due to the increased amount of water in the pond, and even larger pipes could have been used to advantage.
There was no gravity feed water available, which would have made for a nearly perfect situation. Water was supplied from the lagoon by 2 large diesel irrigation pumps, each with a capacity of 300m3/hr and distributed to each pond from concrete supply canals with a maximum flow of 80 m3/hr per pond. Water was pumped once a week to compensate for seepage and evaporation. During fish transfers, which involved pond drainage and filling, pumping was done as needed.
The breeding ponds were stocked with females averaging 120 g and males 150 g. These sizes were the optimum chosen as larger fish were easily injured during handling, and smaller fish did not produce the required number of offspring. 7 females per male were stocked, but mortality usually resulted in an effective ratio of 5:1. The stocking density in the breeding ponds varied from 2 to 3 fish per m3 of surface area.
Fish were replaced at 3 month intervals, as fry production began to decrease at that point.
Brood fish were left in the ponds for 25 to 35 days with a water level of 40 cm. The fish were fed once daily to satiation with 28 % protein pellets. A bloom soon occurred in the pond, and breeding fish were not visible. After about 25 days, the pond was sampled by randomly dipping a 2 m length of mosquito net along the edge. When the fry were present in large numbers in most of the samples yet the size differentiation was not yet large enough for cannibalism to occur, the brood stock were collected with a 12 m long 25 mm nylon seine net heavily weighted along the bottom.
Brood fish were moved to another breeding pond. The PVC drainage pipe was then extended by joining the necessary number of 6 m lengths of the same diameter pipe to direct the water and fry from the spawning pond to a first stage fry pond in the lower tier. A plastic 22 mm mesh trap at the outlet caught any remaining brood stock.
In most instances two breeding ponds were used to stock one first stage fry pond. Stocking densities in the first stage fry ponds were not controlled to avoid mortality.
Fry were held in the first stage fry ponds for 3 to 7 weeks until they approached 1 g mean weight. Feeding was done 4 to 5 times daily with the same 28 % protein feed, although in the ponds it was given in powdered form. Fish were fed until they stopped actively feeding on the surface. The ponds were then drained into a cage measuring 3 × 3 m equipped with a 4 mm mesh nylon cage. The cage was placed in the lagoon beneath the pond. The drain pipe spilled into the middle of the cage and into a sorting pen measuring 1 m3 made of plastic 8 mm mesh pen. This retained fish of 1 g and larger. Fish retained in the pen were transferred to the fingerling cages. Those less than 1 g were returned to another pond containing fish of similar size. Any fish small enough to pass through the 4 mm mesh net were abandoned in the lagoon.
On the shore based facility, yearly production of fish larger than 1 g was 2.424 million weighing 4.083 tons. Although the growth rate of the fry was much less than possible, the numbers produced in such a small area were substantial. Mortality was very high (at least 50 %) in the first-stage fry ponds, caused by cannibalism and predation by frogs. This had no economic effect as the fish were so small and the numbers so great. The monthly production of fry from the entire unit was surprisingly steady at about 200,000 per month.
In practice, the system was easy to operate and control. About two and a half days were needed to drain and restock all the breeding ponds, and three days for the first stage fry ponds. The only mechanical devices were the pumps, and they used only 200 L of diesel fuel a month. Production could have been increased by better controlling predation by frogs, by sorting fish in the first stage fry ponds more frequently to lower the total mass, or simply by constructing more ponds.
The only drawback to the system was that hormonal sex reversal of the fry under the stagnant water conditions of the first stage fry ponds would not have been possible as the fry would feed more on the plankton.
This could have been changed by harvesting the breeding ponds at shorter intervals to insure that the fry were of the uniform proper size, and by building the first stage fry ponds more in the form of raceways and using a flow through system. Pumping, maintenance, and personal costs would have greatly increased. If gravity flow water in sufficient quantities was available, this would have been possible. As the fish were used in cages where over reproduction does not occur, sex reversal was needed.
If one considers only the land surface area, the number produced was 50,000 /are/yr, or 5 million/ha/yr. However, the size of the harvested fish was small, only 1.7 g, and ongrowing was necessary in small mesh cages.
The cages were made from 8 mm plastic mesh, the same size of the sorting pen used in transfer from the first-stage fry ponds. Nylon net could have been used, but plastic mesh was preferable due to the inherent semi-rigidity. The body of the cage was sewn with nylon monofilament fishing line of 1.2 mm diameter, the dimensions were 1.5 × 3 × 1.5 m depth for a total volume of 6.75 m3.
The cage frames were built in a series. There were two continuous parallel lines made of 6 m galvanized steel scaffolding poles (4.45 cm dia) and standard right angle attachments. Bamboo could have been used, but these materials were available, and the scale of the operation was such that the most durable materials were used to lower future maintenance costs. The lines were set 3,5 m apart in the lagoon. Each cage was then suspended from two bamboo poles 4 to 5 m long, which rested at 90 degree angles on the steel lines. Three of the top edges of the cage were attached with nylon fishing twine, the two 3 m sides to the bamboo and one 1.5 m end to one line of the steel frame. The effective water volume in the cage was at least 3 m3, and there were a total of 110 fingerling cages.
Stocking density varied from 1.3 – 3000 fish/m3, or 4–9,000 per cage Because of the large number of cages, and a constantly shifting stock due to almost daily sorting, stocking, and growth, the fish were stocked so that 1.5 kg food/day/cage resulted in a food ration of 4 to 8 % of the biomass. The pelleted feed was distributed 3 times daily at 8, 12, and 16:00 h using automatic feeders. The feed was placed canister made from scraps of 8 mm plastic mesh, with a lining made from pieces of the 4 mm nylon net. The volume was roughly 2 1, and the 4 mm mesh nylon net allowed for slow disintegration of the pellets. The plastic mesh stopped the fish from chewing to pieces the nylon net.
A floating platform measuring 6 × 9 m was made from steel scaffolding poles and empty oil drums. The raft was equipped with a trellis made of the same scaffolding poles, forming a triangle some 5 m above the water level at one end of the raft. This was used to harvest and sort the fish from the fingerling cages. A pulley was attached at the peak of the triangle, giving a leverage point for lifting the cage.
The raft was brough to the side of the cage where the plastic mesh was attached to the steel frame. A man in a small boat passing on the other side of the frame crossed the 2 bamboo poles to form an ‘X’ and passed the cord leading from the pulley around them. Other people on the raft then pulled on the end of the cord, and the cage swung up and out of the water, pivoting on the first steel frame.
The fish were then easily scooped out by dip net and placed into automatic sorters made from various sizes of mesh. A cage containing 8–9,000 fish could be completely sorted and the small fish returned in 10 minutes.
In one year, 2.007 million fish of about 14 g mean weight removed from the fingerling cages, with a production weight of 30.76 tons. When the fish that were left in the cages at the end of 1 year are considered, mortality was 12.93 %. This was due to equipment failure when cages were not properly maintained, losses in handling, and an influx of fish eating birds. However, cannibalism was suspected as the major factor as dead fish were rarely found.
The total amount of feed given in fingerling production process, including spawning ponds, the first stage fry ponds, and the the fingerling cages, was 54 tons in one year, giving a rough food conversion of 1.75.
If the mean weight of all the fish in the fingerling cage ranges from 8 to 12 g, the sorted fish have a consistent mean weight of 14 g (min 12, max 18). As the mean weight in the fingerling cages increases beyond 12 g, the mean weight of the sorted fish increases. Since the minimum 12g weight retained by sorting remains the same, the result is a population of more diverse size.
The total weight of large fish sorted increases in direct proportion to higher stocking rates, and the largest total weight of fry sorted was attained at a stocking rate approaching 35 kg/m3. Growth was not affected to this point.
From an economic and management stand point, the total weight of the fish is the most important factor throughout the fingerling production and sorting process. At each sorting, a relatively small number, about 46%, of the fish in the rearing cage is ready for transfer, but the proportion of the weight of these larger fish is much higher, usually 60 to 70% of the cage. The mean increase in total weight was 80.9% per month.
Production could have been improved by better controlling the food ration of each cage. Previous work (Campbell, 1978) has shown a good potential for sorting largely male populations from fingerling cages on the basis of size, although the exact holding time in the cages and the necessary mesh sizes were not determined in this particular situation. There is also a possibility of overall selection for the fastest growing individuals in the population for further rearing, discarding the slow growers. At such an early stage, the economic loss should not be so important.
When comparing annual production figures, cages can not easily be compared to stagnant ponds. Compared to other cage operations, the production obtained of slightly over 100 kg/m3/yr shows a good utilization of cage capacity.
In the Philippines, a 10 ha farm is used for the production of O. niloticus with organic manures (Broussard et al. 1983). 6 ponds measuring 0.45 ha are considered spawning ponds. These were stocked with broodfish at rates between 150 – 356 kg/ha with a sex ratio of 1 male to 3 females. All ponds received a base application of 2 t/ha air dried chicken manure and 100 kg/ha inorganic fertilizer (N:P:K 16:20:0). Dried chicken manure was applied weekly at a rate of 3 T/ha/month. Inorganic fertilizer was applied weekly at a rate of 100 kg/ha/month. No feed was provided.
About 60 days after stocking, ponds were seined twice with a 6 mm bag seine. Broodfish and fingerlings were graded in suspension nets of varying mesh size, sampled, and weighed. Broodfish were returned to the pond and fingerlings were harvested for dispersal in other ongrowing ponds for advanced fingerling production or sold.
Harvesting was repeated about every 30 days. All ponds were drained completely after about 250 days.
Mean harvest during the 250 day period was about 650,000/ha, with 2.8 T/ha of fry weighting 4.3g mean weight.
The pond serves both as a spawning and rearing pond. Fish harvested from this system are larger than in other fry production systems, which helps in their survival rate upon harvest. The number of fingerlings harvested declined throughout the harvest period as the mean weight of the harvested fish increased. After 150 days, the number harvested declined significantly (100,000/ha) and ponds should be drained and reconditioned then. With a 5 month production cycle, production should exceed 1,2 million /ha.
If larger, advanced fry are needed, fingerlings should be transferred to other ponds and stocked at about 25–30 fish/m2. Production of about 56 kg/ha/day were obtained by using very similar fertilization rates.
In a fertilization only system, the number of brooders is not as important as the total weight of the broodfish. Fry production was good from both big and small brooders, although the size ranges were not given. The decline in production after 150 days is probably due to many factors, including cannibalism, over crowding, competition from the fry that were missed when seining, etc. Better production could be obtained by seining more frequently, for example once every 2 weeks instead of once every month.
In Israel, hormone sex reversed all male populations are produced using large breeding ponds and circular running water tanks (Rothbard et. al,, 1983). Two earthen ponds of about 1 ha each, were used for group spawnings. The bottom of each pond sloped slightly in the direction of the monk, which was placed at the edge of a harvesting sump or catchment area. The bottom of this catchment area was about 500 m2. In harvest, it was covered with a 25 mm mesh net. The net could be lifted to enable separation of the spawners and the small fry. The ponds were refilled with good quality well water after each spawning cycle. Brood fish were fed daily with a 19 % protein diet. 3000 to 5000 brooders were placed per pond, at a sex ratio of about 3 males to 4 females.
Every 17 to 19 days, the ponds were drained and the broodfish and fry removed separately. After lifting the net, females holding eggs in the mouth were placed in a container equipped with a net framed 10 cm above the bottom. The eggs released spontaneously sank through the net which protected them from mechanical damage.
All females were checked individually and additional eggs were washed out with a light stream of water. All spawners were transferred to a temporary holding facility. The fry of 9 to 11 mm were seined from the pond using mosquito netting. Eggs and larvae collected from the females during the harvest were transferred into the hatchery where they were incubated in funnels supplied by running water. These fry were treated as a separate batch.
Immediately after harvest, the pond was thoroughly disinfected by spraying rotenone. One day later the pond was refilled with fresh well water, and after another day restocked with the spawners for a new breeding cycle.
The fry were counted and stocked in outdoor concrete circular tanks 6 m in diameter with a capacity of 28 m3. Densities were 8000– 17000 fry/m3. Every tank sloped slightly to the central outlet which was protected by a filter net. The tanks were equipped with an central telescopic pipe to control the water level. A permanent flow of high quality well water created a circular current driving waste and feces close to the outlet. All solid waste could be easily drained by briefly lifting the center filter. The bottom and wells of the tank were brushed once a week and the whole volume of water was changed. The flow rate of the inlet water was constantly kept at 20–30 l/min in order to exchange the water volume at least once a day. Daily treatment of the water with 25 mg/l of an algaecide controlled and prevented development of algae, which could otherwise have been used by the fish as food instead of the androgen containing feed.
Ethynyltestosterone was dissolved in 95% ethanol and mixed with 1 kg high protein feed (trout starter). The fry were fed a daily portion of 12% of the biomass daily by using 3 clockwork feeders on each tank. Ration was adjusted weekly. All tanks were shaded to avoid direct sunlight. Treatment was 28–29 days.
In one 3 months period, from 2, 1 ha ponds, about 4.6 million fry were harvested. Only about half survived the hormone reversal treatment (2.3 million) due to a variety of reasons including over crowding in the tanks and ectoparasites.
One major draw back was the slow growth rate of the fry in the tanks, where at the end of the treatment they weighed only 0.25 g, which is much less than possible. This is largely due to high crowding.
A system for mass production of tilapia (O. niloticus, S. melanotheron, T. guineensis) fry was developed in Cote d'Ivoire using large concrete tanks or “raceways” and a water flow through system (Campbell, unpubl.). The fish were used on a large cage farm. The basic fry production unit consisted of 20 concrete tanks which measured 3 × 18 m with a water depth of 30 to 40 cm. The entire land based facilities were much larger, using other circular tanks, raceways, and reservoirs. The farm was supplied by 4 electric pumps with a capacity of 150 m3/hr each delivering water from a nearby brackish water (3 to 8 ppt) lagoon, and a bore hole delivering 100 m3/hr of fresh water The usual flow to the 20 tanks used for fry production was 150 m3/hr, 24 hours a day, although the full 700 m3/hr capacity of the farm was available should the need arise.
In general, the raceways were divided into breeding and nursery tanks. The nursery tanks were operated similarly for all fish. Fry of the various species were captured within a few days after the yolk sac was absorbed and stocked at a density of about 30 – 50,000 fry per tank depending on the availability of fry. The species were generally kept separately. Water flow in each tank began at about 2 to 4 m3/hr, and as the fry grew and biomass increased, the flow rate became as much as 20 m3/hr/tank.
Fry were fed using clockwork automatic feeders, usually 2 feeders per tank. The diet was 28% and delivered in a powdered form. If so desired, methyltestosterone was added to the diet to change fish to an all male population. The automatic feeders broke down repeatedly, so feeding was further supplemented by hand feeding the tanks 3 or 4 times a day. Food ratio was generally 6 to 15% of the biomass in the tank, but was not closely followed as the total quantities used (10 – 15 kg/day) where so small compared to the overall food consumption on the farm (several tons a day) that it wasn't worth the effort. Water quality did not usually deteriorate due to adequate flow rates.
After about 3 weeks in the fry tanks, the fish were drained and sorted using a rigid mesh which retained fish large enough for transfer to cages, or those above 1 g. The smaller fish were returned to a tank. The next week, virtually all were large enough for transfer.
In the breeding tanks, management differed due the reproductive habits in the various species.
O. niloticus brooders of 200 to 400 g mean weight were stocked at a ratio of 5 females to 1 male, with between 300 and 400 per tank. The fish were then left undisturbed, and breeding males would create a territory around some imperfection or mark along the concrete walls. Females, either brooding or not, and inactive males would tend to stay in the middle of the tank. After about 15 days, schools of fry were seen, particularly around the screened outlet. These fish were collected using two people and a length of mosquito net, sweeping the entire breadth of the tank at the outlet end. The process was repeated once every 15 minutes until no more fry were captured. About once every 10 days, the water level in the tank was lowered, the sides and bottoms scrubbed, and the brooders left in very shallow water at the outlet end. Fry were then attracted to the inflowing current, and most of the ones missed previously were then captured.
Production varied considerably from tank to tank, and in the same tank from month to month. There were a few exceptions. If the brooders were particularly well domesticated showing no fear of man and literally feeding out of ones hand, the fry production was much higher, usually exceeding 10,000 fry/day and per tank. Fry were collected daily, and it became necessary that two people schedule the entire Monday morning, following the 2 day weekend, to fish fry from a single tank.
With S. melanotheron, broodfish of 100 to 200 g were placed at a density of about 6 fish/m2 (300 to 400 fish/tank) at a sex ratio of 1:1 as the male broods the eggs. Spawning began soon after stocking. Little territorial activity was noticed, but one could easily identity brooding males by the extended lower jaw. This species will retain the fry in the mouth for periods of up to 21 days, so about 3 weeks from stocking, the water level in the tank was lowered to leave 2 to 4 cm in the lower end of the tank; the brood fish swimming on their sides. A small water flow was maintained. The induced stress caused the fish to release the fry, but rarely the eggs. The fry were attracted to the inflowing current, and were collected by using a length of mosquito netting. The brood fish were left in very shallow water until fry were no longer seen. The process was repeated on a weekly basis.
T. guineensis, a substrate spawner showing a highly territorial reproductive behaviour, was bred by placing fish of 75 to 150 g at a density of 4 to 6 fish/m2 with a sex ratio of 1:1. Along each side of the tank, concrete bricks were placed at 2 to 3 meter intervals to allow for territoriality.
Spawning began 7 to 10 days. A couple would create a territory around a brick and defend it. As soon as the yolk sac was absorbed, but the fry were still remaining in the defended territory, 2 people with a length of mosquito net would approach the area. The breeders would flee, and the fry were then easily collected.
Spawning continued sporadically until the tank was drained. This became necessary as the few fry that would be missed when collecting the broods would soon grow to a size where they became extremely cannibalistic and preyed on the newly hatched fry of other broods. Production varied considerably, from 20,000 to over 1000,000 fry/tank and per month. The variation could not be explained.
Fry production of fish weighing 1 g or more from the 20 tank unit varied due to the need s of the farm. When necessary, over 600,000 fry were produced in one month. If the need was less, fry collected from the breeding tanks were simply released into the drainage canals. Yearly production varied from 3 to 5 million fry over a 4 year period.
The described raceway system has several disadvantages. The investment necessary in concrete raceways, pumps, drainage canals, etc. was high. The energy cost of pumping 24 hours day was a lot, particularly on this farm where electric power was supplied solely by diesel generators. Qualified, well trained mechanics and electricians were needed for maintenance of pumps, generators, and so on. Should the water flow stop for any length of time over a few hours, there was a high risk of loosing the fry.
If water was supplied by gravity, the system would be much easier any economical to run, particularly in Africa where spare parts for mechanical devises such as pumps and generators are very expensive and hard to come by.
Broussard, M., R. Reyes, and F. Raguindin, 1983. Evaluation of hatchery management schemes for large scale production of Oreochromis niloticus in central Luzon, Philippines. In: proceedings, International Symposium on Tilapia in Aquaculture, Tel Aviv University, Tel Aviv. p 414–424.
Campbell, D. 1985. Large scale cage farming of Sarotherodon niloticus. Aquaculture, 48: 57 – 69.
Campbell, D. 1978. Formulation des aliments destinés à l'élevage de Tilapia nilotica (L.) dans le Lac de Kossou, Cote d'Ivoire. Rapp. Tech, 45, 31 p. Authorité Aménagement Vallés du Bandama, Centre Dével. Pêches Lac Kossou.
Rothbard, S., E. Solnik, S. Shabbath, R.Amado and I. Grabie, 1983. The technology of mass production of hormonally sex-inversed all-male Tilapias. In: proceedings, International Symposium on Tilapia in Aquaculture, Tel Aviv University, Tel Aviv. p 425–434.
