Previous Page Table of Contents Next Page


8.1 Feeding Rate
8.2 Feeding Frequency and Other Factors
8.3 Biomass Assessment

8.1 Feeding Rate

This topic is vitally important for efficient aquaculture. Underfeeding can result in loss of production. Overfeeding will cause a wastage of expensive feed and is additionally a potential cause of water pollution, a condition resulting in loss of animals or requiring expensive corrective measures. Thus, both overfeeding and underfeeding have serious economic consequences which affect the viability of the farm.

Sometimes you may read a vague instruction, like 'feed 5% of biomass per day' for a dry feed. This might be applied throughout the growing cycle. This would almost certainly result in near starvation in the early stages and gross overfeeding and water quality problems later. Feeding rates should not stay steady throughout the whole of the growing cycle to market size. They must be modified according to the size and age of the fish or shrimp, and to the water conditions.

The quantity of feed to be given to a pond or cage each day should normally be based on a percentage of the biomass present (total weight of animals). Thus, if a pond contains 10 000 fish weighing 10 g on average and the recommended feeding rate (see later) is stated to be 7% per day, the amount of feed to be given daily is:

The percentage of biomass to be fed is not a fixed amount. It should decrease as the animals grow, to reflect their decreasing metabolic rate. Thus the ratio of weight of feed per day to animal weight (biomass) is high at the start of the growing period and lower towards the time when the animals reach marketable size.

Applying a feeding rate accurately depends on an accurate estimate of average animal weight and of the numbers of animals in the production unit (pond, cage, etc.) (survival). Average weight can be obtained directly from weighing samples or by measuring the length of the animals, where an accurate length/weight relationship has been established. Comments on these topics are contained in section 8.3. Accurate record keeping is essential not only to aid efficient feeding now but also to enable you to examine the effect of past actions and to help you predict the result of planned actions during the next growing cycle.

Feeding tables have been constructed for various aquatic species. Many are to be found in the references given in 'further reading' at the end of this section; manufacturers of compound feeds always give a feeding guide for their products. Examples of feeding tables are provided in Appendix XIII. The tables for those species, such as trout, which are reared under highly intensive conditions tend to be more elaborate and reliable because they are based on many decades of accurate observation and measurement. They usually specify feed type and size as well as the daily feeding rate. Similarly elaborate tables will become available for other species as more becomes known about the most efficient ways to present their feed. Meanwhile simple feeding guides for some species are available (Appendix XIII).

It is emphasized that the feeding rates given in tables must not be applied without reference to other factors. Feed should be reduced in quantity or omitted during times of low temperature and, based on operational experience in a specific location and environment, increased when growth rates are predicted to be highest. Daily feeding rates must also be based on your observation of the animals during feeding. At this time feeding activity, water quality (colour), presence of old feed, etc., must be assessed. All feeding tables are merely a guide which, if applied with careful judgement, will markedly improve economic viability. However, if applied rigourously without complementary assessment of conditions, they can result in disaster.

As stated before, feeding rates should be decreased as the animals grow. Feeding a steady percentage of biomass throughout the growing cycle usually results in underfeeding when the animals are small, depressing growth and survival rates, and overfeeding when the animals are larger. The total effect is depressed growth rate, poor survival, and poor AFCR. The more often feeding rates are adjusted, the more efficiently will feed be utilized. Ideally they should be adjusted daily but this would require too much paper work and would confuse those who do the feeding. Also, it is only possible to adjust the feeding rate accurately when the biomass can be estimated by measurement or when it can be accurately predicted from past experience. This is where accurate, long-term records are so important.

In practice, feeding rates are adjusted weekly or twice-monthly for salmonids and catfish, based on estimates of biomass, together with a knowledge of environmental conditions (mainly temperature). Feeding tables for other species are at present less refined and feeding rates are adjusted less frequently. For example, using a feeding table given in Appendix XIII for carp, the amount of feed to be given daily (at 20-23° C) begins at 9% for animals of less than 5 g in size. It changes to 7% for animals of 5-20 g and to 6% for those of 20-50 g average weight. Further reductions, to 5%, 4%, and 3%, respectively for animals weighing 50-100 g, 100-300 g, and 300-1 000 g are recommended in that table. Obviously, changes in feeding rate will be less frequent when this type of table is used, than is possible with the more detailed tables, such as are given in Appendix XIII for salmonids. In practice, more frequent adjustments can be made to feeding rates even when simple feeding tables are used. In the example just mentioned, feeding rates at 20-23° C could be adjusted more frequently in the following way:

Animal Size (g)


% of Biomass to be Fed per Day

Recommended in Table

Adjusted Actual Rate




























Increase in growth can be measured by length or by weight. With some species, notably salmonids, so much is known about their characteristics that simple methods of predicting growth rate under each environmental condition have been derived which enable more complex feeding schedules to be planned one or two months ahead. The ability to do this, particularly for a large farm with many different tanks and animal batches of differing sizes, is essential for forward planning of feed purchases, storage requirements, cash flow, etc.

The following example, adapted from Piper et al., (1982) is given of the way in which this type of feed table can be used:

Salmonids, when reared at constant temperature, increase their length at a constant rate for the first 1 ½ years of life. The weight of fish at each given length is also known. For a certain temperature the amount of daily feed needed can be calculated from the formula:


FCR = the amount of feed necessary to produce a unit of animal weight increase (e.g., 1.2 kg feed to produce an increase of weight of 1 kg is equivalent to an FCR of 1.2).

A = the daily increase in length in centimetres

B = the length of the fish in centimetres at the present time

To calculate 'A' in the above formula, an average monthly growth in centimetres is established from the records of previous years on the farm at the same temperature. This is divided by the number of days in the month to obtain a value for 'A'. The value of FCR is also extracted from previous experience, recorded in the farm data, of animals of the same size category reared with the same type of feed at the same temperature.

This equation enables feeding rates to be predicted in advance and adjusted frequently even when animals are not sampled and measured so frequently. It is emphasized that such a sophisticated system is only possible where there have been many years of experience in rearing the same species and where long-term record keeping has been faithfully attended to. An example of the calculation involved is given below:

Suppose, on 13 April, there are 200 000 fish present in the pond. Their feeding rate was last determined on 1 April, when the fish were 9.07/kg per 1 000 fish (or 1.45 cm in length) 1/. We wish to adjust the feeding rate again now, knowing from past records that at this operating temperature, the average length increase per day (A) will be 0.0075 cm during April and the expected FCR is 1.2.

The new feeding rate is then calculated:

Length, 1 April:

1.45 cm (9.07 kg/1 000) 1/

Plus growth, 13 days x 0.0075 cm (A):

0.0975 cm

= Calculated length today (B), 13 April:

1.5475 (11.03 kg/1000) 1/

1/ From tables of length/weight relationship

Applying the formula, the amount of feed to be given, as a percentage of biomass, is:

Thus the weight of feed to apply is:

As the daily increase in length in constant, and the length/weight relationship is known, this formula can be used to calculate new daily feeding rates frequently even though actual measurements of body length are infrequent.

The above example assumes 100% survival. In practice this is extremely rare so the biomass figure must take into account estimates of animal population as well as size. It is foolish to feed for 200 000 fish if there are, for example, only 160 000 left. (see section 8.3).

Very small fish and shrimp fry are generally fed to excess. Fry should be fed to satiation several times per day. Feeding rates for fry can be as high as 50% of biomass/day for catfish and 100-200% of biomass/day for very young shrimp or prawn postlarvae. Feeding at such high rates, especially if stocking rates are also high, places a strain on water quality which must be relieved by greater water exchange, the removal of waste feed and detritus, etc. The extra trouble and expense involved in such treatments is more than recompensed by the production of healthy, fast-growing fry with good survival rates.

Feeding rates for channel catfish in the U.S.A. average 3-5% of body weight per day throughout the season (see tables in Appendix XIII) because of seasonally changing water temperatures, except when the animals are close to market size. Feeding rate does not, in this case, start high and decrease gradually as it would do if the whole of the growing cycle were conducted at near constant temperature level (as it would for fish grown in tropical zones). An example of feeding rates for channel catfish kept at constant temperatures is also given in Appendix XIII for comparison.

Feeding rate tables for trout, salmon, channel catfish, common carp, tilapia and marine shrimp are provided in Appendix XIII.

Many prawn farmers, for example in Thai freshwater prawn culture, feed 'to demand' rather than according to a feeding rate table. Daily feeding rate is adjusted according to how much of the previous day's feed is consumed. This is reasonably effective where the farmer is experienced particularly where prawns, rather than fish, are being reared. Fish tend to eat more than is really necessary, resulting in poor FCR, if too much feed is presented. Phytoplankton density, measured with a Secchi disk or by hand, which obscures at 25 cm or less, is also used to detect overfeeding and as a means of controlling feeding rate. For a freshwater prawn batch-culture pond, stocked at 5 animals per square metre, the feeding rate starts at about 6.25 kg/ha/day, rising to about 37.5 kg/ha/day at harvest time (New and Singholka, 1982). The final feeding rate is equivalent to about 3% of biomass per day.

8.2 Feeding Frequency and Other Factors

8.2.1 Salmon and Trout
8.2.2 Catfish
8.2.3 Tilapia
8.2.4 Carp
8.2.5 Other Fish Species
8.2.6 Shrimp and Prawns

The most effective method of feeding with respect to location, time of day and frequency varies from species to species. Its cost effectiveness depends also on other factors such as the availability of feeding labour or automatic feeders, size of pond or tank, cost of labour, and the personal preference of the farm manager, based on observations and results. In this section of the manual a brief review of information about each of the major groups of cultured aquaculture species dealt with here is presented.

A number of basic rules, suggested by Piper et al., (1982) are summarized first:

a) for optimum growth and feed conversion, each feed should ideally be only 1% of the body weight. Thus, if 5% of biomass per day is being fed, there should be five feeds per day

b) survival rate is not significantly affected by feeding frequency once the initial feeding stage of the animals has been passed

c) frequent feeding reduces starvation and stunting of small fish; thus the group has better uniformity

d) infrequent feeding results in feed wastage, poor FCR, water quality problems, and the leaching out of water soluble nutrients

e) dry feeds should be distributed more frequently than moist feeds

f) 90% of the feed (for fish) should be consumed within 15 minutes or less of the feeding time.

In narrow or small ponds for fish, feed should be spread evenly around the perimeter. For larger ponds, other methods have to be used to give adequate distribution, especially for species which are territorial in nature. These methods include feeding from a boat, feed blowers towed by a tractor, and (in the U.S.A.) distribution by aeroplane for very large ponds. Boats are, of course, essential for the feeding of moored cages which are not connected to the shore by a walk-way. Information on feeding devices is given in Appendix XIV.

Returning to the subject of feeding frequency, the following sub-sections summarize information on a species by species basis.

8.2.1 Salmon and Trout

In common with other very young fry of fish and shrimp, very frequent feeding is most effective for young salmonids. For swim-up fry of salmonids the daily feed ration is split up into very small quantities fed as often as 20-24 times per day, either manually or automatically. Sometimes a 24-hour lighting regime is used for the first few days to encourage the fry to take dry feed. Feeding frequency is gradually reduced to 1-3 times per day as the fish grow. Rainbow trout start to take food about twenty-one days after hatching when reared at 10° C. Most hatcheries feed at ½ to 1 hourly intervals during an 8-hour day, reducing this to three times per day. After the fish are about 5 inches long (23 g) feeding frequency is reduced to twice per day. Brood fish are fed only once per day. Feeding frequencies quoted by Piper et al., (1982) for coho salmon, autumn chinook salmon and rainbow trout are given in Table 22.

Salmonids are frequently reared in tanks or cages so feed distribution is not a problem. Many different types of automatic feeders are utilized, some of which are mentioned in Appendix XIV.

Table 22 Suggested Feeding Frequencies for Salmonids



Fish Size (g)










No. of Feeding Times Per Day

Coho salmon









Chinook salmon








Rainbow trout










Source: Piper et al., 1982

Salmonids tend to feed to satiation and then do not eat again until most of the meal has left the stomach. Once past the fry stage therefore, a feeding frequency of 1 or 2 times per day is sufficient.

8.2.2 Catfish

The information presented here refers to channel catfish; it is reasonable to assume that it is applicable to other species of catfish.

Channel catfish fry begin to feed 5-10 days after hatching, when the yolk sack reserves are used up. As with salmonids, swim-up fry are best fed many times per day. One commercial feed manufacturer, whose feeding table is reproduced in Appendix XIII, recommends 8-10 feeds per day, reducing quickly to 6 per day by the time the fish are about 2.5 cm long. Feeding frequency is further reduced to 3 times per day when the fish reach 7.6 cm in length. Juvenile catfish grow best with two feeds per day, one at mid-morning and one late in the afternoon, seven days per week.

Feeding frequency in ponds depends on water temperature. At 13-29 C, feeding 6-7 times per week is recommended. At times of particularly high or low temperatures, less frequent (4 or 5 times per week) feeding is suggested. Feeding 6, instead of 7 days per week is said to encourage the fish to consume any surplus feed in the pond and lessen the chance of over-feeding. Catfish in cages should be fed daily. There is some evidence that feeding catfish twice per day, especially under raceway conditions, results in a faster growth rate.

Automatic feeders and mechanical distribution of feed is common in channel catfish culture (see Appendix XIV).

8.2.3 Tilapia

In the wild, tilapia feed more or less continuously throughout the day. Manual feeding, several times per day is best for intensively grown tilapia, in cages or raceways for example.

Automatic feeding can be employed and the blower type of feeder is said to distribute feed more adequately for tilapia. Tilapia fry should be fed at least 4 times per day, preferably 8 times per day, in daylight hours. In an experiment with 9 mm total length fry of Oreochromis aureus, New et al., (1984) showed that survival improved with continuous (mechanical) feeding compared to either five or three manual feeds per day. Feeding rates can be less frequent, 4 or 5 times per day for fingerlings. Adult tilapia thrive best on 2-3 feeds per day.

8.2.4 Carp

Again, common carp (and probably Chinese and Indian carps) thrive best on frequent feeds. Jauncey (1982) reports that one researcher found that optimum feed utilization by common carp (at 40 g size) was achieved when the feed was split into nine equal feeds. The best feeding frequency can only be assessed on an individual farm basis, determined by the cost of repetitive feeding operations. From the biological and nutritional point of view it would appear best to feed as frequently as possible.

8.2.5 Other Fish Species

Specific feeding frequency recommendations for the other cultured species of fish covered in this manual are not available. It is therefore recommended that until more information on the optimum conditions becomes available those frequencies found best for catfish and tilapia should be applied. A golden rule would be, when in doubt, to feed as frequently as economics allow. There has been a report, however, that groupers gave best biomass increase and good FCR, fed 'trash' fish to satiation in cages, when fed every second day, as compared to other frequencies varying from once every 5 days to three times per day. However, this result was apparently caused by poorer survival at the higher feeding frequencies. Best growth rate was, again in this case, achieved by feeding two or three times per day.

8.2.6 Shrimp and Prawns

There is a good deal of controversy about the optimum time and feeding frequency for marine shrimp and freshwater prawns. Some species burrow during the day and feed most actively at night. Others feed in the shallower parts of the pond but avoid these areas in daylight when temperatures are highest. For these species it would seem best to feed in the late afternoon or early evening. Most farmers feed once, or at the most twice per day, usually first thing in the morning and last thing in the afternoon.

Shrimp do not consume all of the feed presented at once, unlike most fish. This fact has led to much discussion and research on methods of binding shrimp feeds to prevent wastage and the loss of water-soluble nutrients. Some commercial feeds are extremely water stable (>24 hours) but may not be so palatable as softer feeds. The apparent need to produce such well-bound diets has, in part, been caused by the reluctance to feed shrimp and prawn ponds more frequently than once a day, even though labour is often available and otherwise unoccupied. Feed presented in smaller quantities more frequently would not need to be so efficiently bound and should therefore be cheaper.

In Taiwan many intensive farmers (usually farms are run by a family enterprise, so there is always someone on site) feed tiger shrimp (Penaeus monodon) four to six times per day, the feeds being evenly spaced over the whole 24 hour period.

I suggest that feeding shrimp and prawns as frequently as possible, spreading the daily ration between those feeds, would be the most effective technique. It certainly pays off with young post-larvae, as it does with fish fry. Shrimp larvae do not thrive at all unless maintained in a conditions where food (live food or artificial feed with neutral buoyancy) is constantly available.

8.3 Biomass Assessment

Calculation of the amount of feed to present daily must be based on an assessment of the biomass (total weight of fish or shrimp) in each pond or enclosure.

Regular, accurate, data for average animal size must be obtained through weekly, bi-weekly or, at the worst, monthly sampling of the rearing unit. In tanks or cages it is easier to take representative samples but in ponds it is common to get a biased picture, particularly when a species with uneven growth rate is stocked, such as freshwater prawns. Care must be exercised to take samples in several parts of the pond, not only at feeding points where the larger or more active individuals may congregate. Samples may be taken by seine, cast net or lift net.

If accurate length/weight relationships for the species have been pre-determined under the environmental conditions being used, length measurements are a more accurate means of monitoring growth rate. This is particularly true of crustacea which hold uneven quantities of water under their carapace. Measuring length can be a rapid process with a skilled operator and is less stressful to the animals than trying to determine weight by a standardized technique. Fish length can either be total or, to avoid inaccuracies due to damage to the tail fin, is more accurately measured to the anterior end of the caudal fork. Care must be taken that the method of measurement corresponds with that used when the length/weight ratio was determined. Similarly total shrimp length, because of frequent rostrum damage, is less reliable than measuring from the posterior of the eye orbit to the tip of the telson.

For a pond of 5 000 m2, stocked at 10/m2, at least five samples should be obtained at each sampling time and 50 animals from each sample measured. Average weight can either be calculated directly from the total weight of the sample obtained by weighing or by referring the average of the measured lengths to a length/weight ratio.

An assessment of survival is also necessary to calculate feeding rate effectively. This can be illustrated as follows. If a pond is stocked with 50 000 animals and, at a sampling date the average weight is 10 g and the feeding rate to be applied is 3% of the biomass/day, the amount of feed would be:

if it is assumed that all the animals originally stocked are still present. If however, there has been a 20% mortality up to the 10 g size, the correct amount of feed should be:

Besides saving 20% of daily feed costs in this case, an accurate assessment of survival, as well as growth rate, would prevent possible water quality deterioration caused by over-feeding.

Although a good estimate of survival aids an effective feeding programme it is often extremely difficult to achieve. In small cages and tanks it is often possible to make accurate visual observations or to count all the animals during their transfer to another tank or cage. This is normally impossible or impractical in pond culture, except when stock transfer from one pond to another takes place for other reasons. In this case the numbers present can be calculated by taking the total weight of stock and dividing by the average animal weight obtained from samples. Visual observation is also impossible in ponds. In my view, no-one has yet devised a satisfactory way of assessing the stock in aquaculture ponds, which is why this important subject is rarely mentioned.

In practice most farm managers apply an arbitrary survival factor based on the number of days since stocking. This factor is derived from knowledge of previous culture cycles on the same farm or elsewhere in similar circumstances, modified by observation of actual mortalities, knowledge of water quality or disease problems, etc. The factor derived depends on accurate measurement of the number of animals originally stocked and the numbers harvested in each cycle for its accuracy. Thus the importance of careful farm records becomes obvious.

If those records show, for example, that 50% of the animals stocked normally reach market size (barring individual accidents) on the specific farm, it would be reasonable to assume that future growing cycles would show the same survival rate. Similarly, if there is no transfer of animals within the growing period, the highest mortality rate probably occurs after initial stocking (due to the handling stress and the ease of damage and of predation on young animals). Thus, if a 50% mortality is known to occur normally between stocking and harvesting in a 16 week growing cycle, for example, it would be reasonable to assume that 20% of the losses occur within the first 4 week period with a further 10% loss occurring during every 4 week period after that. Thus assessments of biomass in this example at two weekly intervals of a pond stocked with 50 000 animals would be based on multiplying the average animal weight, obtained by measurement of samples by the following number of animals:

Weeks 1 and 2

Weeks 3 and 4

Weeks 5 and 6

Weeks 7 and 8

Weeks 9 and 10

Weeks 11 and 12

Weeks 13 and 14

Weeks 15 and 16

Thus, to continue this example, if the average weight of prawns at the beginning of week 13 (12 weeks after stocking) was 20 g, and the percentage of biomass to be applied was 5%, the daily amount of feed would be:

Clearly assessment of biomass, particularly in ponds, depends partly on accurate sampling and size measurement but also very much on the manager's judgement based on an accurate record of past experience modified by the history of the particular batch being cultured.

Further reading for section 8:

Piper et al., (1982); Halver (1972); Phillips (1970);Gaudet (1967); Lee (1981); Ralston Purina (1974); Marek (1975); Foltz (1982); NRC (1983); NRC (1981); Pullin and Lowe-McConnell (1982); Jauncey and Ross (1982); New (1986a); New and Singholka (1982); Winfree (1979); Jauncey (1982).

Previous Page Top of Page Next Page