Even though selective breeding programmes can be designed to improve all sorts of production phenotypes, the single most important phenotype is growth rate. Improving growth rate will decrease the time it takes to grow a fish to market, which means a farmer can produce more crops in a given time period, which means he will make more money. It also means yields will increase. This will increase production efficiency, increase food production, and also increase a farmer's income. As an added bonus, increasing growth rate can improve other production phenotypes via indirect selection. Some studies have shown that faster-growing fish also have more efficient feed conversions and seem to be more disease resistant.
Selection can be used to improve other quantitative phenotypes, if improving them will improve production efficiencies or profits. In most cases, as important as these phenotypes are, they are not as important as growth rate. Occasionally, a breeding programme other than selection is needed to solve the most important goal. For example: in tilapia farming, controlling reproduction in grow-out ponds is the most important goal, and this is accomplished by interspecific hybridization and/or sex reversal; in grass carp farming in the U.S., producing sterile fish is the most important goal, and this is achieved by chromosomal manipulation.
The object of this chapter is to outline simple, relatively inexpensive selective breeding programmes that can be conducted by farmers on farms with about 2 ha of ponds and to provide examples of the types of data that must be collected, as well as data tables that can be used to record these data. This chapter will not discuss selective breeding programmes that use tandem selection; a selection index; or those that combine selection with crossbreeding, inbreeding, or some facet of biotechnology such as chromosomal manipulation.
When describing these simple selective breeding programmes, information will be provided on the number of ponds that will be needed in order to grow the fish that will be evaluated. This number will not include the number of ponds that will be needed to hold or to spawn brood stock. Additionally, this chapter will not discuss ancillary facilities that are needed, such as holding tanks, hatchery buildings, counting tables, etc.
The main goal for the selective breeding programmes that are outlined in this chapter will be to improve growth rate by selecting for length. As was described in Chapter 4, growth can be improved by selecting for either length or weight. Because it is far easier to accurately measure hundreds of fish than it is to accurately weigh them, and also because most farmers do not have access to accurate balances, most farmers should select for length.
Examples of selective breeding programmes that are designed to improve two phenotypes will also be outlined. In these examples, selection to improve growth rate is still the primary goal, but a second phenotype is added.
The selective breeding programmes outlined in this chapter are designed to be conducted by a farmer who is creating genetically improved stock for his farm. If a farmer wants to embark on a breeding programme so that he can sell genetically improved fingerlings to support a local or regional industry, the size and costs of the projects that are outlined must be increased: more ponds must be built; more cohorts must be created; and more select brood fish must be produced. In fact, if this is a farmer's goal, it is likely that he will convert his entire farm to this enterprise, and he will no longer produce food fish.
One of the things a farmer must do if he is to run a successful selective breeding programme is to determine how many select brood fish must be saved. If too few are saved, a farmer will be unable to produce enough select fish for future grow-out. If this happens, he will have to spawn unselected fish, which will negate much of his efforts.
An additional problem is that if a farmer saves too few fish, inbreeding can build to levels which will result in inbreeding depression. If this occurs, much of the improvements created by selective breeding would be used simply to counteract inbreeding depression.
A farmer should spawn at least 25 males and 25 females every generation to minimize inbreeding depression. This number is not carved in stone, but it can be used as a general rule. This means that a farmer should save a minimum of 100–200 brood fish. The reason why a farmer needs to save this many select brood fish is some will die before they can be spawned, and spawning success is seldom 100%. These guidelines also apply for a control population.
A second factor that determines how many brood fish a farmer needs is the size of his operation; that is, how many fingerlings are needed for the grow-out ponds. If a farmer saves too few brood fish, he will not be able to stock his grow-out ponds. Saving enough select brood fish can be a problem on large fish farms, but it is seldom a problem on medium-sized (2 ha) fish farms, especially if a farmer saves at least 100–200 select brood fish.
The selective breeding programmes described in this chapter are bare-boned skeletal outlines that are presented to demonstrate that relatively simple and relatively inexpensive programmes can be conducted by farmers and that, if the programmes are conducted properly, they can be integrated into everyday farming practices. The programmes outlined in this chapter are not absolute. They can be and should be modified in order to customize the programme for a particular species, for a farmer's income, and for his farm.
The breeding programmes that are outlined in this chapter are for species that do not exhibit sexual dimorphism. This means a single cut-off value can be used during selection.
If the species that a farmer grows exhibits sexual dimorphism a slight modification of the programmes that are outlined will be necessary. Sexual dimorphism can increase the effort needed to conduct selection, because once it occurs, the fish must be sexed and selection must occur independently in each sex. The age at which sexual dimorphism occurs will determine if a single cut-off value can be used or if separate cut-off values will be needed for males and females. In some species, sexual dimorphism occurs when the fish are small and separate cut-off values are needed at both the fingerling and food fish phases of selection. If sexual dimorphism does not occur until after the fingerling stage, a single cut-off value can be used at the fingerling stage, while separate ones will be needed when food fish are harvested. If sexual dimorphism occurs after the age at which food fish are harvested, a single cut-off value can be used during both phases of selection.
The selective breeding programmes outlined in this chapter do not describe the creation and use of a control population to assess the results of selection. This was described in Chapter 4.
If possible, individual selection should be used to improve growth rate. Individual selection is easier and less expensive than family selection, because it can be done in only one or two ponds and fewer fish need to measured.
Growth rate is the most important production phenotype. The following programmes outline how individual selection can be used to improve only this trait. When conducting individual selection, if a farmer can synchronize spawning, selection will be relatively simple. If the species spawns asynchronously, care must be taken to ensure that age-related size differences do not confound and obscure genetically-produced size differences. If the selective breeding programme is not modified to accommodate asynchronous spawning behaviour it will be impossible to identify fish which are genetically superior from those which are environmentally superior (older).
Synchronous spawning: If a farmer can synchronize spawning and can produce at least 25 families on a single day (or at most, over a 48-hour period), he can conduct a simple, inexpensive selective breeding programme in only one or two ponds. In addition, an extra pond will be needed for the select brood fish, once they have been saved.
To initiate the selective breeding programme, the fish should be spawned using normal management techniques. If possible, egg masses should be collected and incubated. If the eggs are usually incubated by the females in the ponds, they should be closely monitored, and fry should be collected as soon as they hatch or begin to swim away from each female.
If possible, family size should be equalized before fish are stocked. This will prevent one family from skewing the results of selection; additionally, it will help minimize inbreeding. Because of this, families should be isolated until they are equalized. In addition, if families are isolated until they are stocked, the complete mortality of one or more families will be noticeable and can be recorded.
If the fish are traditionally raised in a two-phase process (phase one is stocking fry and raising them to fingerlings; phase two is stocking fingerlings and raising them to food fish), selection can be done when fingerlings are harvested and when fish are harvested for market; if this is the case, selection will require two ponds. If production is only a single-phase process (no fingerling phase; fry are stocked and raised to food fish), selection will occur when fish are harvested for market, and selection will require only one pond.
If fish are raised in a two-phase process, fry should be stocked in a single pond and fingerlings should be produced using normal production techniques. Just prior to harvest, a random sample of 100–200 fingerlings should be measured to the nearest millimeter in order to determine the phenotypic value that corresponds to the desired cut-off percentage, as was outlined in Figure 20. At fingerling harvest, the top 35–50% should be saved and stocked in the food fish production pond. The culled fingerlings can be grown for food or sold. If they are grown for food, no fish from this population should be saved and spawned.
The select fingerlings should be raised using normal production management. Prior to harvest, a random sample of 100–200 fish should be measured to the nearest millimeter to determine the phenotypic value that corresponds to the desired cut-off percentage. At harvest, the top 10–20% of the fish should be saved, and these fish will become the select brood fish. Culled fish can be eaten or sold.
If fish are raised in a single-phase process, a random sample of 100–200 fish should be obtained to create the F1 control brood fish when fish are harvested. The population of F1control brood fish should be created prior to selection. If fish are raised in a two-phase process and if selection will occur at both the fingerling and food fish harvests, the F1control brood fish need to be saved at the end of the fingerling phase. If the F1 control brood fish are saved just prior to the second act of selection during a two-phase selection process, the control will not be a true control population, since some selection will have occurred prior to its creation.
If fish are raised in a single-phase process, fry should be stocked in a single pond, and fish fish should be produced using normal production techniques. Selection at harvest is done as described above.
These simple selective breeding programmes are outlined in Figures 29 and 30. Figure 29 outlines the procedures that will be used when selection will occur twice-at fingerling harvest and at food fish harvest. Figure 30 outlines the procedures that will be used when selection occurs only once at food fish harvest.
The cut-off percentages were mentioned as ranges, not as specific values (35–50% for fingerlings and 10–20% for food fish). This is because the exact cut-off values are not that critical. The intensities of selection that are used are individual decisions. A farmer can increase the rate of gain by using higher cut-off percentages (saving a smaller percentage). But rate of gain must be balanced against inbreeding-related problems and with the ability to produce enough fish in the next generation. Most selective breeding programmes conducted on medium-sized farms will encounter few problems and will also be able to achieve desired results if cut-off values in these ranges are used.
The pond(s) used in this project should be 0.04–0.4 ha. Pond size is determined by the stocking rate that will be used to grow the fish and by the intensity of selection. Once a farmer decides what his cut-off percentages will be and how many select brood fish will be saved, he can determine how big the ponds must be. This will tell him the percent of the farm that will be devoted to the breeding programme. The procedure used to determine pond size is outlined in Table 14. The values derived in Table 14 are valid only for the assumptions that were given in Table 14.
Figure 29. Schematic diagram of a simple and inexpensive selective breeding programme to improve growth rate by selection at two ages-at fingerling harvest and at food fish harvest. This breeding programme can be conducted in two ponds. Fish that are culled at the fingerling stage can be sold or grown for food. Fish that are culled at harvest can be eaten; sold as food; or some can be retained and used as brood fish to produce the production fish (fish raised and sold as food), if the select fish cannot produce enough offspring for both the selective breeding programme and for the production ponds.
Figure 30. Schematic diagram of the least expensive selective breeding programme that can be designed to improve growth rate. In this breeding programme, fry are stocked in a pond, and the fish are not harvested until they are ready for market. Selection occurs when the pond is drained and the fish are harvested. This breeding programme can be done in one pond. The fate of the culled fish was described in Figure 29.
If ponds of approximately the correct size already exist, a farmer does not have to destroy them and custom-build new ponds in order to conduct a selective breeding programme. A farmer who uses existing ponds should adjust stocking rates and/or cut-off values to achieve the desired results. The formulae presented in Table 14 can be used to generate this information.
The information generated in Table 14 can also be used to determine how many fish are needed from each family. For example, if a farmer needs to stock 8,892 fry and if he has produced 25 families, he needs 355.7 fry/family (8,892/25). Because you cannot have 0.7 fry, he needs 356 fry/family. Because the number is rounded up, the farmer will stock 8,900 fry. If he maintains the stocking rate at 200,000 fry/ha, the size of his fingerling pond would change only by 0.001 ha to accommodate the 8 additional fry, so a larger pond would not be needed (pond size had already been rounded to the nearest thousandth). The rounding up of family size also means that the initial culling will produce an additional 2 select fingerlings. This will have little effect on stocking rate in the food fish pond, nor will it require a larger food fish pond. If the farmer does not want to alter stocking rate one iota from his planned rate, he can randomly cull 2 select fingerlings once that population has been created.
Brood fish that are saved should be stocked in a select brood stock pond. No other fish should be stocked with the select brood fish. When mature, these fish will be spawned to produce the F1 select generation. No selection for secondary sexual traits and no selection for other criteria should be done with the brood fish prior to the mating season. The only goal of this selective breeding programme was to improve growth rate by selecting for length, and the select brood fish were selected for that phenotype and should be selected only for that phenotype; they should not be selected for any other reason. The select brood fish should be allowed to mate among themselves in a random manner or, if they will be paired in pens or manually stripped, they should be randomly paired.
If a farmer has enough select brood fish, he can use these fish to produce both the F1select generation and the production fish (fish that the farmer will stock and grow for market in his production ponds). If the select brood fish cannot produce a sufficient number of offspring for both purposes, their fry should first be used to produce the F1select generation; any surplus fish can be grown for food. If the select brood fish cannot produce enough fingerlings for both purposes, the farmer should use other brood fish (culls from the harvest selection) to produce fry for the production ponds.
Table 14. Procedure that can be used to determine the size of the ponds that are needed in a selective breeding programme.
|Goal:||To have 200 select brood fish|
|Given:||Cut-off at food fish harvest: select top 10%|
Cut-off at fingerling harvest: select top 50%
Fry-fingerling mortality: 50%
Fingerling-food fish mortality: 10%
Stocking density in fingerling pond: 200,000/ha
Stocking density in food fish pond: 7,000/ha
|Step 1.||How many food fish must be harvested to produce 200 select fingerlings, if you save the top 10%?|
|Number of food fish harvested = number saved/percent saved = 200/0.1 = 2,000.|
|Step 2.||How many select fingerlings should be stocked in the food fish pond, if mortality is 10%?If mortality is 10%, survival is 90%:|
|Number of fingerlings stocked = number harvested/survival rate = 2,000/0.9 = 2,223.|
|Step 3.||How many fingerlings must be harvested from the fingerling pond to produce 2,223 select fingerlings, if you save the top 50%?|
|Number of fingerlings harvested = number saved/percent saved = 2,223/0.5 = 4,446.|
|Step 4.||How many fry must be stocked in the fingerling pond to produce 4,446 fingerlings, if mortality is 50%? If mortality is 50%, survival is 50%:|
|Number of fry stocked = number harvested/percent survival = 4,446/0.5 = 8,892.|
|Step 5.||How big should the fingerling pond be, if it is stocked at 200,000/ha and if you will stock8,892 fry?|
|Size of fingerling pond = number of fry that will be stocked/stocking rate =8,892/200,000 = 0.0446 which is rounded to 0.045 ha.|
|Step 6.||How big should the food fish pond be, if it is stocked at 7,000/ha and if you will stock2,223 fingerlings?|
|Size of food fish pond = number of fingerlings that will be stocked/stocking rate =2,223/7,000 = 0.3175 ha which is rounded to 0.318 ha.|
The second, third, etc. generation of selection can proceed as described above. If the select brood fish cannot produce enough fish for both the selective breeding programme and for the production ponds, genetic improvements that are being achieved in the select population can be transferred to the production fish beginning with the second generation of selection, if the farmer uses the culls from the breeding programme as brood fish to produce fry for his production ponds. Even though these brood fish were culled, they came from the selective breeding programme, and their parents were select brood fish. This will enable a farmer to transfer the genetic gain to the farmed fish, but it will be with a one-generation delay. Additionally, when a farmer replaces one generation's select brood fish with their successors, he can transfer the previous generation's select brood fish to the production brood fish population. This will also enable the farmer to transfer genetic gain to the production population. Some of the ways genetic gain can be transferred to the production ponds are illustrated in Figure 31.
If selection will occur only when food fish are harvested, a farmer will not be able to transfer the genetic improvement to the production population as quickly, unless the select fish can produce enough offspring for both the breeding programme and for the production ponds. If the select brood fish can produce only enough offspring for the breeding programme, a farmer must use unselected brood fish to produce fry for the production ponds after the first generation of selection. After the second generation of selection, he can either use F1 select brood fish or fish that were culled when he created F2 generation brood fish. This means a farmer will realize no genetic improvement in his production fish for one generation, but thereafter, he will be able to transfer it with a one-generation delay.
Asynchronous spawning: If a farmer cannot spawn his fish synchronously, he should divide the population into age cohorts and select for growth rate independently within each cohort. To initiate this breeding programme, a farmer should spawn the fish using normal management techniques. As before, it is better if egg masses can be collected and incubated. If the eggs are usually incubated by the mother in the ponds, females should be closely monitored, and fry should be collected from each female as soon as they hatch or begin to swim away.
Egg masses or newly hatched fry should be grouped into daily (24-hour) age cohorts. If it is not possible to collect enough families within a 24-hour period, the time interval for each cohort can be stretched to 48 hours. Each cohort should be composed of at least 5 families, and there should be at least five cohorts. These numbers are not carved in stone. A farmer does not have to discard a cohort simply because it is made up of only four families. There are two basic premises behind this work plan: the first is that a cohort should be composed of several families; the second is that at least 50 parents (25 males and 25 females) should produce offspring for the breeding programme.
As before, if this species is usually produced by a two-phase production process, selection for increased length will occur when fingerlings are harvested and when food fish are harvested. If production is usually a single phase with no fingerling phase, selection will occur only when fish are harvested for food. Individual selection will be used to select the best fish from each cohort.
Figure 31. Schematic diagram illustrating the ways genetic gain can be transferred from the selective breeding programme to the production fish that a farmer grows for market. If the select brood fish can produce enough for both populations, the transfer will be immediate, and the means of the two populations will be the same (path 1). If selection is a two-phase process, the culls from the second phase can be used as brood fish to produce the production fish (path 2). If this approach is used, some of the genetic gain will be transferred immediately, but the production mean will always lag behind the select mean. If selection is done only at harvest, unselected fish will have to be used to produce the first generation of production fish, and either the culls from the breeding programme or the previous generation's select brood fish can be used thereafter (path 3). This approach will transfer the gains with a one-generation delay if the previous generation's select brood fish are used; if culls from the breeding programme are used to produce the production fish, the production mean will be slightly better than the mean of the previous generation of select fish. All assumptions about mean values were generated using the premise that there is no environmental influence on the phenotype and the trait has a large heritability.
This breeding programme will require five to ten 0.04-ha ponds. The number of ponds depends on the size of the project (how many cohorts will be produced) and whether production is a two-phase or a one-phase process. The exact size of the ponds can be determined as was described previously and as was outlined in Table 14. The exact size is not that important, but all ponds should be similarly sized. The reason you want small ponds is because a large number can be squeezed into a small space. Additionally, it is cheaper to build a 0.04-ha pond than a 0.1-ha pond.
If selection is a two-phase process, select fingerlings could be restocked in the ponds where they were produced, if the ponds can be drained and refilled in a single day. This would reduce the number of ponds needed for the programme by one-half. The only prerequisites are: all fish must be harvested; the farmer must have holding facilities where he can safely hold the select fingerlings until they are restocked.
Farmers who cannot afford to build the ponds could conduct this breeding programme in large hapas (20–40 m2) that are placed in one or more ponds. Even though hapas are less expensive than ponds, it is better to use ponds, because if the fish are grown in ponds they should be selected on the basis of growth in ponds, not on the basis of growth in hapas.
Each cohort should be stocked in a single pond; stocking rates for each cohort should be identical, but it is not critical, since selection will occur independently in each cohort. Even though it is not necessary to have identical stocking rates in all ponds, it is advisable to have similar stocking rates, or selection could select for slightly different genes in the different cohorts.
The best way to create the population of fish that will be stocked in each pond is to choose an equal number of fish from each family within a cohort. The fish chosen from each family must be randomly selected; they cannot be chosen because they are the largest, etc. Therefore, families must be isolated until family size is equalized. If the families within a cohort are mixed before the proper number is chosen to create the desired stocking rate, the largest family will be over-represented and the smallest will be under-represented.
When the fingerlings are harvested, the top 35–50% from each cohort (pond) should be stocked in a single grow-out pond. To determine the fingerling cut-off value for each cohort, a sample of 100–200 fingerlings from each cohort should be measured to the nearest millimeter, and the phenotypic value that corresponds to the desired cut-off percentage should be determined, as was outlined in Figure 20. The cut-off value must be determined independently for each cohort, because they are managed as temporary, separate sub-populations during the selection process, and selection will occur independently in each cohort.
As before, the intensity of selection is not that critical, but it must be the same for each cohort. Fish from different fingerling ponds (cohorts) should not be mixed when they are harvested and measured. After selection, the select fingerlings from each cohort should be stocked in a separate food fish ponds. Fish from different cohorts should not be mixed.
As was the case when stocking the fingerling ponds, the stocking rates for the food fish ponds should be the same or at least similar. The select fingerlings in each cohort should be grown using normal production techniques. Just prior to harvest, a random sample of 100–200 fish from each cohort should be measured to determine the cut-off value for each cohort. The top 10–20% from each pond (cohort) should be saved to form the population of select brood fish. Again, the intensity of selection is not that critical, but it must be the same for all cohorts.
Once the select brood fish have been chosen from each cohort, they can be mixed and stocked into one or two select brood stock ponds. These ponds should contain no other brood fish. As before, a major goal of the selection process is to have at least 100–200 select brood fish. The two-phase selective breeding programme is outlined in Figure 32.
The brood fish should be managed and used to produce fry for the breeding programme and for the production ponds, as was described in the previous sub-section.
If this selective breeding programme were going to be conducted by a scientist at a research station, a major aspect of his experimental plan would be to make all aspects of management identical for all cohorts. But a farmer does not have to worry about this. It would be nice if all cohorts were managed identically during each phase of selection, but because each cohort is stocked in a single pond and because selection will occur independently within each cohort, minor management differences among the cohorts will not affect selection. For example, if one food fish pond is stocked with 5,000 fingerlings/ha and the others are stocked with 4,000 fingerlings/ha, the different stocking rates could affect the average growth rate of the cohorts, but since selection will occur independently in each pond (cohort), it probably will not affect the selection process.
However, large differences in management among the ponds could affect the outcome. For example, if one food fish pond is stocked with 15,000 fingerlings/ha and if cattle manure is used as the sole source of nutrients while the others are stocked with 4,000 fingerlings/ha and if fish are fed rice bran, selection may select for different genes in the differently managed cohorts.
The breeding programmes that have been outlined can be expanded to include another phenotype. If a farmer wants to improve phenotypes other than growth rate, he can easily add traits such as body conformation and/or harvestability. Farmers should use independent culling or modified independent culling to simultaneously improve two phenotypes. Aquaculturists at fingerling production centers should also use independent culling; however, they could use a selection index if they are technically sophisticated and have the expertise and labour needed to conduct this programme.
When independent culling is used, a farmer needs to determine the overall intensity of selection in order to calculate the cut-off percentage for each trait. This process was described in Chapter 4. A farmer should try to save 10–20% of the population at the food fish phase of selection. The phenotypic values that correspond to the desired cut-off percentages should be determined as was described previously and as was outlined in Figure 20.
Body conformation is an important phenotype, and improving this trait can increase yields. A deeper-bodied fish or a thicker-bodied fish will carry more muscle on its frame, which means each fish weighs more per centimeter body length than a normal, streamlined fish.
Figure 32. Schematic diagram of a selective breeding programme that uses individual selection to improve growth rate when a farmer cannot synchronize spawning. This selective breeding programme divides the population into age cohorts, and selection occurs independently within each cohort. In the programme outlined in this figure, there are five cohorts (A-E), and there are five families within each cohort. If there is no fingerling phase, step one can be eliminated, and the selective breeding programme can be conducted in five ponds. If the number of families can be increased in each cohort, fewer cohorts will be needed, which will decrease the number of ponds that are required. The fate of the culled fish was described in Figure 29.
Selecting for weight might improve body conformation, but this type of selection does not guarantee that body conformation will be improved. Selecting for weight simply improves average weight, and heavier fish might be fish that are longer, that have larger heads, etc. Improving body conformation has been a continuing goal of many cattle, swine, sheep, and poultry breeding programmes, because animals with better body conformations produce more meat.
One way to improve body conformation is to select for both length and body depth at the anterior margin of the dorsal fin (at the first dorsal fin spine). Since the primary goal of the breeding programmes that are outlined in this chapter is to select for length as a way of increasing growth rate, all a farmer has to do is add selection for body depth, and he will add body conformation as the second phenotype.
Alternatively, body conformation can be improved by selecting for a body length:body depth ratio. This has been done with common carp and it succeeded. However, even though body conformation was improved, average weight did not increase. This occurred because fish were selected only for the body length:body depth ratio. Consequently, small, but deep bodied fish could have become select brood fish. To improve both growth rate and body conformation, both traits should undergo selection.
If selection will occur at the end of both the fingerling and food fish phases of production, the initial act of selection (fingerling phase) can be for length only, as was described earlier. The rate of progress will be slower for body conformation (body depth) than for growth rate (length) if selection for body depth is conducted only on food fish, but if independent culling is conducted twice, a farmer may end up with only a few select brood fish.
When fish are raised in ponds and harvested by seining, harvestability is a phenotype that a farmer might wish to improve. Anyone who has seined a pond knows all too well that fish are experts at escaping a seine. Most farmers never consider the costs of harvesting when they determine annual production budgets, but it can be significant in terms of labour and equipment. Hard-to-capture fish can become stressed and killed as a result of repeated seinings. Finally, fish that are not captured cannot be sold or eaten.
If a farmer wants to produce fish that are easier to capture, he can add this phenotype to his selective breeding programme and select for both growth rate (length) and harvestability by using independent culling. If a farmer decides to select for harvestability, he needs to define it as “fish that are captured during the first seine haul.” If this is done, independent culling will be a two-step process: The initial step will be to save only those fish that are harvested during the first seining; all other fish will be culled. The second step will be to select for length in the fish that were saved (seined).
It may be that so few fish are harvested during the first seining that a farmer cannot select for length efficiently or meaningfully. If this is the case, the goal should be modified so that fish that are captured during the first two seinings are saved.
The ability to escape a seine is not only a detrimental phenotype in that it increases production costs, it is detrimental because it can produce slower-growing fish, if a farmer conducts selection for growth rate improperly. If a farmer cultures fish using multiple batch production, selection for growth rate can be done only during the first harvest after the pond is filled.
Thereafter, size and age are confounded, especially if fingerlings are stocked to replace fish that are harvested or if fish can reproduce in the pond.
If a farmer chooses to select for growth rate and for a second phenotype, growth rate will be improved more slowly than it would be if it were the only phenotype under selection. A farmer could select for growth rate, harvestability, and body depth (or any other phenotype he wants to improve) by using independent culling, but the rates of improvement for all three traits would be small.
In general, family selection is used when heritability is small and/or when there are uncontrollable sources of environmental variance which obscure genetic differences and which make individual selection ineffective.
Within-family selection is usually used when there is a large environmental source of variance that has a major influence on phenotypic variance at the family level. Chief among these factors are spawning date and age and size of the mother.
Between-family selection is usually used when most of the phenotypic variance is due to environmental sources of variance, and they are felt at the individual rather than at the family level. When this occurs, an individual's phenotypic value does not accurately reflect its breeding value; consequently, individual selection will be ineffective and between-family selection must be used. By comparing family means, a farmer can neutralize much of the environmental component of phenotypic variance, and the family means can be used to assess the average breeding value of every fish in each family.
Between-family selection is also used when animals must be killed before their phenotype can be measured. For example, between-family selection is usually used to improve carcass traits, because animals have to be slaughtered in order to be measured.
The simplest type of family selection that can be used is within-family selection. This is because each family can be considered a separate sub-population, and selection will occur independently within each family, in a manner similar to that described for individual selection when the population was divided into age cohorts. In this case, each family can be considered a cohort. Unless families can be given permanent family marks, families will have to be cultured in individual ponds or hapas.
The size of the ponds that are needed in this type of breeding programme will depend on the fecundity of the species. With some species, pond size could be as small as 100 m2. For example, despite their reputation for being incredibly prolific, tilapia produce relatively small families. In general, family size ranges from 50–1,500, depending on the female's size. If the desired stocking rate is 5,000/ha, a family of 50 would have to be stocked in a 100-m2 pond to achieve that stocking rate. If family size is 50–100 fish, it might be more efficient to raise the families in 10-to 20-m2 hapas that are suspended in a 0.1- to 0.2-ha pond.
Twenty-five to 50 families should be produced and entered into the selective breeding programme. This means that 25–50 ponds must be constructed for the project.
If the ponds can be drained and filled in a day and if selection is a two-phase process, the number of ponds that are needed can be reduced by one-half if the farmer has holding facilities. When the ponds are refilled, the fingerlings from each family can be restocked into the pond where they were produced.
Twenty-five families is the minimum number that should be entered into this selective breeding programme, because you want 25 males and 25 females to produce offspring for reasons that were described earlier. If a farmer wants to use the minimum number of families, he should produce 27–35 families, because mortalities in some families may reduce family size below that needed to stock a pond with the desired number of fry.
Fish should be spawned and families should be produced as was described for individual selection. Families must be isolated, because selection will be at the family level.
Because each family will be treated as a temporary sub-population and because selection for length will occur independently within each family, minor differences in management among the ponds (families) will not affect the selective breeding programme. This means that a family does not have to be discarded because a farmer cannot stock it at the desired rate. As was the case for individual selection when the population was divided into age cohorts, minor differences in management among the ponds are of little consequence, but a farmer should try to manage all ponds similarly.
At harvest, a sample of 30–100 fish from each family (or every fish if family size is small) should be measured to the nearest millimeter to determine where the cut-off should be placed. In this case, the cut-off value is expressed as the largest 5 or 10 or 20, etc. fish. A farmer should save the top 5–10 males and the top 5–10 females from each family. If no sexual dimorphism exists at harvest, a farmer can simply save the top 10–20 fish from each family. This selective breeding programme is outlined in Figure 33.
This type of selective breeding programme obviously requires more effort than individual selection. If 25 families are raised and if a farmer measures 30–100 fish from each family, he will measure 750–2,500 fish to determine cut-off values, as opposed to the 100–1,000 fish that were needed in the programmes that were outlined for individual selection.
Furthermore, this type of selective breeding programme could stress the fish. A farmer has to measure every fish in each family twice-once to establish the cut-off value and once to determine which fish should be saved.
Once the select brood fish from each family are saved, they can be mated using either of two techniques: The first and easiest is to stock the fish in a single pond and spawn them in a random manner. The second is to use rotational mating, which was described in Chapter 4. If rotational mating is used, either each family must be given a unique mark and stocked communally until the next breeding season when they will be isolated once again for spawning, or each family must be stocked in a separate pond. Rotational mating is very expensive in terms of facilities and labour, and it greatly increases the costs of the breeding programme. For these reasons, the first approach is recommended for most farmers.
The genetic gain should be transferred to the production fish as was described earlier. If possible the select brood fish should be spawned to produce both the F1 select generation and fry for the production ponds.
If a farmer wants to improve two phenotypes by within-family selection, he can add a second phenotype such as body depth or catchability as was described earlier for individual selection. A farmer must be judicious in his use of within-family selection to improve two or three phenotypes. If family size is small, only one or two fish from each family may be able to meet or exceed all cut-off values. If this occurs, a farmer must either greatly relax his cut-off values or evaluate 100–200 families, which might not be possible on a 2-ha farm.
Figure 33. Schematic diagram of within-family selection to improve growth rate. A minimum of 25 families should be evaluated, and each family can be stocked in a single pond. The fate of the culled fish was described in Figure 29.
Between-family selection is more expensive than individual selection and within-family selection, because it requires more ponds. Because whole families will be saved or culled, a farmer should enter 25–50 families. It is better if a farmer can evaluate 50 families, but the costs will probably be prohibitive for most farmers.
Because family means will be compared, a farmer must raise each family in at least three ponds, and the assignment of the families to the ponds must be random; consequently, a farmer will need 75–150 ponds for this programme. When fish are measured to determine which families will be saved and which will be culled, the means from the three ponds are averaged, and it is the overall mean from the three ponds which is used as each family's mean. Families must be raised in replicated ponds because this is the only way to isolate pond-related size differences from gene-related size differences. If each family were stocked in only one pond, the largest family might be the largest simply because the pond had the best algal bloom.
Because this type of selection requires the use of so many replicated ponds, it might be cost-prohibitive for farmers who only want to produce genetically improved fish for their own use. For example, if a farmer evaluates 50 families, he will need 75 ponds. If each pond is 0.01 ha, he will use 37.5% of his 2-ha farm for the breeding project.
This unfortunate side-effect of family selection can be circumvented only if a farmer is able to give each family a unique mark. If he can, the fish can be stocked communally in one or two 0.1- to 0.25- ha ponds. Even if this is possible, each family must be raised in an individual tank or pond until it can be marked. At harvest, the fish must be separated into families once again, so the farmer must have adequate holding facilities.
Because whole families will be saved or culled, even if production is a two-phase process, selection needs to be conducted only when food fish are harvested. If the farmer wants to select the families twice, he can cull the worst 5–10 families during the fingerling phase of selection. This will make the grow-out phase less expensive.
If there is a large genetic correlation between fingerling size and food fish size, selection can occur at the fingerling phase instead of the food fish phase. This will reduce the costs of the breeding programme because it will reduce the number of ponds that are needed for the food fish phase, since only the select families will be grown to market size. This approach has been used with rainbow trout, and it improved harvest size by indirect selection.
At the end of the food fish phase of production, the top 5–10 families should be saved; these are the select brood fish. A farmer can save either the entire family or a random and equal number from each select family. As before, a farmer should save at least 100–200 select brood fish. This selective breeding programme is outlined in Figure 34.
The one aspect of between-family selection for increased growth rate that can be inexpensive is obtaining data to determine which families should be saved. Because selection is based on family means, the fish from each pond can be batch weighed. If the number of fish that are weighed is known, the mean weight can be easily determined. This will enable the farmer to improve growth rate by selecting for weight rather than for length.
Figure 34. Schematic diagram of between-family selection to improve growth rate. A minimum of 25 families should be evaluated. Because family means are compared, each family must be grown in at least three ponds, and the overall mean of the three ponds is the value that determines whether a family is saved or culled. If the fish can be given family marks, they can be grown in a single pond, but the families must be segregated when they are measured and selected. The fate of the culled fish was described in Figure 29.
In general, between-family selection precludes selection for two or more phenotypes, because only those families whose means meet or exceeded two cut-off values would be saved, and it is unlikely that the top five families for one phenotype would also be the top five for a second. A farmer could improve two phenotypes by using between-family selection to improve one phenotype and another type of selection to improve a second phenotype.
The select brood fish should be managed and spawned as was described for the within-family selective breeding programme. Again, because it is less expensive, most farmers should mix the select families and randomly spawn the select brood fish. However, if this is done, farmers should be aware of the fact that inbreeding will build to levels that will cause problems within a few generations. This occurs because between-family selection makes the breeding size of the population far smaller than the actual number of brood fish. This type of breeding programme is designed to save fish from only 5–10 families each generation, which means the breeding size of the P1 generation is retroactively lowered to 10–20, and it is smaller thereafter.
A combinational approach that is often used is to combine between-family and within-family selection. If this combination is used to improve two phenotypes, the first step is to improve growth rate via between-family selection. The second step is to use within-family selection to improve the second phenotype. Of course, between-family and within-family selection can be combined in a two-step selection programme to improve only growth rate. A logical approach, and one that has been successfully used with rainbow trout and coho salmon, is to use between-family selection at the fingerling phase of selection and to use within-family selection at harvest.
If the between-family selection and within-family selection combination is used, it will be expensive. It will combine all of the costs of between-family selection with some of the costs of within-family selection.
Selective breeding programmes work only when farmers keep records. It is probably the least appreciated but most necessary aspect of any breeding programme. Without records, there is no way a cut-off value can be determined, so it is impossible to create a population of select brood fish. Without records, there is no way to determine if a selective breeding programme is succeeding. Without records, a farmer will not know which ponds contain select fish, which ponds contain select brood fish, which ponds contain control fish, and which ponds contain fish that are being grown for market.
Farming is one occupation where record keeping should be considered an integral part of everyday life. Farmers should gather information about spawning success, the stocking rates used, mean size at stocking, the amount of fertilizer used, the amount of feed used, percent survival, mean weight at harvest, yield, etc. If a farmer has this information, he knows what is going on and does not rely on guesses. He knows if yields decreased because of bad weather, and he knows how much they decreased. He knows if yields increased because he used a better quality fertilizer, and he knows how much they increased. If there are no records, he can only guess.
Farmers should keep records of these parameters for every pond. Ponds have individual personalities. Records will indicate which ponds are bad and which are good. This information can also be used to customize management for each pond.
Many farmers do not want to or are unable to keep accurate records. These farmers probably should not be encouraged to conduct a selective breeding programme. If a farmer will not gather the data that are needed to evaluate a programme, he probably will not conduct the programme properly. This could be detrimental on a regional basis because the farmer might tell his neighbors that selective breeding is a waste of time-he will not tell them that it did not work because he did not conduct it properly.
Extension agents could provide record keeping services for hard-working, reliable farmers who are unable to keep records. If selective breeding programmes are kept relatively simple, an extension agent could provide this service for a number of farmers. The only drawback to this approach is that these farmers will become totally dependent upon the extension agent and often will move from one phase of their programme to another only when the extension agent appears. If the extension agent is transferred to another region or retires, the selective breeding programmes could collapse if the new extension agent has other priorities.
What kind of data should a farmer take and what kinds of records should he keep if he is going to conduct a selective breeding programme to improve growth rate or other quantitative phenotypes? First and foremost, a farmer must be able to describe the phenotype he is going to improve via selection. This means he must be able to measure it accurately and quickly and without stressing the fish.
Farmers who take data routinely will be able to incorporate what is needed into everyday management. In fact, if a farmer already records the kind of information that was mentioned earlier in this section, he may be recording these data routinely, so there will be little extra work.
If a farmer is going to conduct a selective breeding programme to improve growth rate by individual selection, he must obtain and maintain data on: the number of fish that spawned, the number of families that were produced, and the dates the families were produced; the families that contributed offspring to the selective breeding programme; the number of fry from each family that were grown; when the fish were stocked; where the fish were stocked; when the fish were harvested; when the fish were measured; phenotypic values; the cut-off value; the spawning success of the select brood fish; the performance of the F1 select generation, as well as that of the control population. This process is then repeated for the second generation of selection, etc. In addition, a farmer should maintain data about normal, everyday management for every pond used in the selective breeding programme.
Examples of data tables that can be used to gather the information needed to conduct a selective breeding programme by individual selection are presented in Tables 15, 16, 17, 18,19, and 22 and Figure 20 (in Chapter 4); data tables that can be used to record normal, everyday management will not be presented. The data tables illustrated in this chapter are not immutable. They are simply presented as examples. Any permutation may be used, as long as it is well organized and as long as the data are easily accessible.
Tables 15 and 16 are examples of data tables that can be used to record spawning information for each generation of individual selection. Table 15 is a data table that can be used to record data for a simple selective breeding programme, where the species can be spawned synchronously and where the fish will be grown in a single pond. Table 16 is a data table that can be used to record data for a selective breeding programme where the species cannot be spawned synchronously, and where the fish will be raised and selected in cohorts. Both data tables provide data on the date each spawn was produced, the size of each family, the number of fish from each family that was used in the selective breeding programme, the pond where each family was stocked, and the date the pond was stocked.
Tables 17 and 18 are examples of data tables that can be used to record phenotypic values for individual selection. Table 17 is for a selective breeding programme that will select for only one phenotype, and Table 18 is for one that will incorporate two phenotypes. The tables include information about the group of fish: when the group of fish was produced, when it was stocked, and when the fish were measured (so the age can be determined); the pond in which the population was stocked; the number of families that contributed fish to the population. Both tables are designed for species that do not exhibit sexual dimorphism. If a species exhibits sexual dimorphism and if separate cut-off values will be used for each sex, the tables can be divided into male and female sections or separate tables can be kept for each sex.
Table 15. Example of part of a data table that can be used to record spawning data for a selective breeding programme that will improve growth rate by individual selection; the fish will be stocked in a single pond. The table also includes data on the number of fish stocked per family, the pond in which they were stocked, and the stocking date.
|Dates: May 1 – May 15, , 1995|
Pond Nos.: 1, 2, and 3
|Species: Any fish species|
Generation: P1 generation. Fish grown will be selected to become F1 select brood fish.
|Stocked on : April 30, 1995|
|Date||Spawn||Pond||Weight egg mass (g)||Number of eggs||Number hatched||Number of fry||Used in breeding programme?||Number stocked||Pond stocked||Dated stocked|
Table 16. Example of a data table that can be used to record spawning data for a selective breeding programme that will improve growth rate by individual selection; the population will be divided into age cohorts. The table also includes data on the number of families in the cohort, the number of fish stocked from each family, the pond in which the cohort was stocked, and the stocking date.
|Dates: May 5 – May 15, , 1995|
Pond Nos.: 3, 4, and 5
|Species: Any fish species|
Generation: P1 generation. Cohort A. Fish grown will be selected to become F1 select brood fish.
|Stocked on : May 1, 1995|
|Date||Spawn||Pond||Weight egg mass (g)||Number of eggs||Number hatched||Number of fry||Number stocked||Pond stocked||Dated stocked|
Table 17. Example of a data table that can be used to record length measurements at harvest. This data table is designed for a species that does not exhibit sexual dimorphism. In this example, only 30 lengths are recorded.
|Species: Any fish species|
Date: October 1, 1995
Stocking Date: March 1, 1995
Number stocked: 1,000 fingerlings
Number offamilies: 27
|Lengths in mm|
|mean: to be calculated|
Table 18. Example of a data table that can be used to record harvest data for two quantitative phenotypes. This data table is designed for a species that does not exhibit sexual dimorphism for body size. In this example, length and body depth are recorded for only 4 fish.
|Harvest Length and Body Depth|
Species: Any fish species
Date: October 3, 1995
Stocking date: March 4, 1995
No. stocked: 1,230 fingerlings
Spawned: April 2, 1994
Number of families: 29
|Fish No.||Length (mm)||Body depth (mm)|
|Means||= to be calculated||= to be calculated|
The data tables illustrated in Tables 17 and 18 record harvest data. These data tables can be used to record data at any time, from stocking to harvest. In the examples, Tables 17 and 18 were used to record phenotypic values from the sample of fish that were measured to determine the cut-off value(s). The phenotypic values in Tables 17 and 18 would be transferred to the data table illustrated in Figure 20 to determine the phenotypic value(s) that corresponds to the desired cut-off percentage.
Table 19 is a table that takes the means from Tables 17 and 18 at harvest (or at any other time if selection is done prior to harvest) and records the progress that is being made by the selective breeding programme. Mean lengths for the select population and control population, as well as the genetic gain can be recorded in Table 19.
Tables 19–23 are examples of data tables that can be used to record data from selective breeding programmes that use family selection. Some of the data tables can be used for both individual and family selection, but those illustrated in Tables 20, 21, and 23 are specific for family selection.
Table 19. Example of a data table that can be used to record the mean lengths for each generation, and the progress that is made as a result of selection. This data table can be used for both individual and family selection. In this example, only the first two years data are recorded.
|Selective breeding programme to improve growth rate in: Any fish species|
|Date||Generation||Original mean||Select population mean||Control population mean||Genetic gain|
|1995||F1||330 mm||321 mm||9 mm|
Table 20. Example of a data table that can be used to record spawning data for a selective breeding programme that will improve growth rate by within-family selection. The table also includes information on the pond where each family was stocked, the number of fry stocked in the pond, and when the pond was stocked. Ponds 13 and 15 are smaller than the others, so they were stocked with fewer fish.
|Dates: May 1–May 15, 1995|
Pond Nos.: 1,2, and 3
|Species: Any fish species|
Generation: P1 generation.Fish grown will be selected to become F1 select brood fish.
|Stocked on:April 30,1995|
|Date||Spawn||Pond||Weight egg mass(g)||Number of eggs||Number hatched||Number of fry||Family||Number stocked||Pond stocked||Dated stocked|
Tables 20 and 21 are data tables that can be used to record spawning information for each generation of family selection. Table 20 is for within-family selection, and Table 21 is for between-family selection. The information recorded on these data tables are similar; however, they are organized differently.
The data tables presented in Tables 17 and 18 and Figure 20 can be used to record phenotypic values for selective breeding programmes that use family selection. The only difference would be that the data tables would record the family that is being measured.
The data table presented in Table 19 can also be used to record yearly means from a breeding programme that uses family selection. If desired, a separate table can be used to record data about every family.
Finally, the data tables illustrated in Tables 22 and 23 can be used to record the number of fish that spawned each generation. Table 22 is for individual and within-family selection, and Table 23 is for between-family selection.
Table 21. Example of a data table that can be used to record spawning data for a selective breeding programme that will improve growth rate by between-family selection. The table also includes information on the ponds where each family was stocked, the number of fry stocked in the pond, and when the pond was stocked. The table is only partially completed.
|Dates: May 5 – May 15, 1995|
Pond Nos.: 4,5, and 6
|Species: Any fish species|
Generation: P1 generation. Fish grown will be selected to become F1 select brood fish.
|Stocked on: May 1,1995|
|Date||Spawn||Pond||Weight egg mass (g)||Number eggs||Number hatched||Number of fry||Family||Number stocked||Pond stocked||Date stocked|
Table 22. Example of a data table that can be used to record the number of fish that are spawned each generation for a selective breeding programme that uses individual and within-family selection. In this example, only two years' data are recorded.
|Selective breeding programme to improve growth rate in: Any fish species|
|Date||Number of fish spawned||Number of families used||Number of brood fish Contributing offspring|
Table 23. Example of a data table that can be used to record the number of fish that are spawned each generation, the number of families that are used, and the number of families that are saved for a selective breeding programme that uses between-family selection. In this example, only two years' data are recorded.
|Selective breeding programme to improve growth rate in: Any fish species|
|Date||Number of fish spawned||Number of families used||Number of families saved||Number of brood fish contributing offspring|
The selective breeding programmes outlined in this chapter demonstrate that growth rate and other quantitative phenotypes can be improved by relatively simple and inexpensive selective breeding programmes. If possible, individual selection should be used because it is the easiest and least expensive. If fish can be spawned synchronously, a selective breeding programme to improve growth rate can be conducted in only one or two ponds, and it will have little effect on normal farming operations. Even if the species cannot be spawned synchronously, if several age cohorts can be produced, the selective breeding programme will still be relatively simple and inexpensive, and it will have minimal impact on normal farming operations.
If a farmer cannot control spawning behaviour and if he cannot develop age cohorts or if the phenotype that the farmer wants to improve has a small heritability and is strongly influenced by environmental variables, he must use family selection to improve the phenotype. If possible, within-family selection should be used because it is easier and less expensive than between-family selection. However, the most appropriate type of selective breeding programme is determined by the biology of the species, by the phenotype's heritability, and by the type of environmental factors that influence phenotypic expression, not by a farmer's desire to have an inexpensive breeding programme.
The complexity of a selective breeding programme depends on a number of factors: First and foremost is the number of phenotypes that will be improved. Virtually all selective breeding programmes should attempt to improve growth rate. This is the most important phenotype, because faster-growing fish take less time to reach market and faster-growing fish increase yields. Other phenotypes can be added, but a farmer must add only those traits that are truly important, because the rate of improvement for growth rate will be inversely related to the number of phenotypes that he incorporates into the selective breeding programme. At most, a farmer should add only one or two additional phenotypes.
Secondly, the complexity and cost of the selective breeding programme is determined by the way the fish are produced: a single-phase growing system, where fry are stocked in a pond and grown to market; a two-phase system, where fry are grown to fingerlings in one pond and where fingerlings are then grown to market size in a second pond. If fish are produced under a single-phase production system, selective breeding programmes will be easier and less expensive because fewer ponds will be needed. On the other hand, if fish are produced under a two-phase system, selection can occur twice, which means greater gains can be achieved.
Finally, the complexity of the selective breeding programme is partially determined by whether the species exhibits sexual dimorphism for body size. If it does, cut-off values will have to be created for each sex, which means each fish must be sexed in addition to being measured. This doubles the cost of measurement and increases record keeping slightly.
One of the end goals of a selective breeding programme should be to save 100–200 select brood fish each generation. This will ensure that a farmer will be able to spawn at least 25 males and 25 females each generation. If this is done, inbreeding-related problems should be minimized for 5 generations. A farmer also needs to save enough select brood fish in order to produce sufficient offspring for the next generation of selection. Although this can be a major concern on large farms, it should be of little concern on medium-sized farms.
If the programme is conducted properly, it can be integrated with and can complement normal farming operations. This is important because if there is a conflict, the selective breeding programme will usually be neglected or abandoned; a farmer's top priority is going to be food production.
If there are enough select brood fish, they can be used to produce offspring for both the selection programme and for the production ponds. This will enable a farmer to transfer the genetic gain from the selective breeding programme to the production ponds immediately.
If the select brood fish cannot produce enough offspring for both purposes, they must be used to create the next generation of select fish; any surplus offspring can be stocked in the production ponds. In this case, culls from the selective breeding programme or the previous generation's select brood fish (beginning with the F2 generation of selection) can be used to produce the production fish. This approach will transfer genetic gain from the selective breeding programme to the production ponds, but the transfer will be delayed; in the extreme, the transfer will be with a one-generation delay. This means a farmer must maintain two sets of brood fish: those used to produce the select generation, and those used to produce the fish stocked in the production ponds. Even if the transfer of genetic gain is delayed by one generation, the mean growth rate and yields that are obtained in the selective breeding programme will allow a farmer to predict the growth rates and yields that he will be able to achieve in his production ponds in the future. Additionally, these data will demonstrate that selective breeding is improving his fish and will also enable him to realize that his selective breeding programme is going to create better fish, larger harvests, and greater profits.