- Email: [email protected]
Application of breeding schemes
Aquaculture. 85 (1990) 163-169 Elsevier Science Publishers B.V., Amsterdam -
163 Printed in The Netherlands
Application of Breeding Schemes T. REFSTIE Institute of Aquaculture Research (AKVAFORSK), N-6600 Sunndalsara (Norway)
ABSTRACT Refstie, T.. 1990. Application of breeding schemes. Aquaculture, 85: 163-169. Application of animal breeding theory to fish is discussed. Most of the traits investigated in fish show a high degree of phenotypic variation with possibilities of producing progeny groups that have a high number of individuals in each group. Therefore, good genetic progress can be expected even for traits with low heritabilities when the breeding schemes are based on results from fulland half-sibs. However, the control of reproduction and mating represents a limiting factor for many species. The Norwegian selection program for Atlantic salmon is described. The present breeding program utilizes a combination of between-family and within-family selection for traits such as growth rate, age at maturity and survival. This system can easily include new traits such as meat quality and disease resistance in the selection scheme.
INTRODUCTION
Animal breeding theory has only recently been applied to fish culture and is not widely practised on aquatic species. Compared with other farmed animals, fish have the advantage of a high reproductive capacity which allows for the possibility of a high selection intensity. This high fecundity makes it possible to produce progeny groups with many individuals in each group. The possibilities of testing different traits on full- or half-sib groups are better in fish than in most other species. Most of the traits investigated in fish have shown a high degree of phenotypic variation (Gjedrem, 1983). Even for traits with low heritabilities, good genetic progress can be expected when the breeding schemes are based on results from full- and half-sibs. If, however, breeding schemes could be used in practice, control of reproduction and mating is necessary. At present we have this control in only a few commercially farmed species, and breeding schemes are applied to only a few species, of which salmonids seem to be the most important. In this paper, the Norwegian breeding program for Atlantic salmon will be used as an example of the application of a breeding scheme.
0 1990 Elsevier Science Publishers B.V.
164
T. REFSTIE
Before possibilities of applying breeding schemes can be discussed, breeding goals have to be defined. Breeding goals should be established using the input of both industry and consumers. All traits of economic importance which show genetic variation should be defined as breeding goals. Even if a trait is economically important, the problems and costs associated with measuring it should be considered before including such information in a breeding scheme. Gjedrem (1983,1985) identified growth rate, food conversion efficiency and survival rate as traits to be considered in formulating breeding goals in most production systems involving fish species. Meat quality, age at maturity and fecundity are traits which might be of economic importance for some species in some production systems. Because some traits cannot be measured on live fish, a statement of breeding goals will influence selection strategy. SELECTION OF STRAIN
If different strains or populations are available, all breeding programs should start with the collection, comparison and selection of the best genetic material available. The value of testing strains and selecting the best for farming can circumvent several years of within-strain selection. Differences in performance between strains within species have been well documented (for review, see Kinghorn ( 1983 ) ). At AKVAFORSK, eggs from about 40 Norwegian Atlantic salmon strains were collected and compared during the years 1972-1976. At the same time progeny groups were produced within strains. This allowed the estimation of genetic parameters for important traits and the testing of strains to be done simultaneously. Fig. 1 gives an example of differences in growth rate, measured as weight at slaughter, found between strains and between progeny groups within strains 70
560 -u -C o\ p 50 3 $
40
famify no Fig. 1. Ranking of strains of Atlantic salmon and full-sib families within a strain for body weight after 2 years in sea cages (year-class 1973) (Gjerde, 1986).
APPLICATION OF BREEDING SCHEMES
165
for one year-class. The great differences found between strains and between progeny groups within strains indicate the importance of starting a selection scheme on the best available genetic material. The differences between fullsib families within a strain also illustrate the possibilities for further progress through selection. ESTIMATION OF GENETIC PARAMETERS
To apply a simple selection scheme it is only necessary to know if the trait or a genotype of traits shows genetic variation. In order to optimize a breeding scheme or predict responses to selection, reliable phenotypic and genetic parameters are required (for review, see Gjerde ( 1986) ). It might also be necessary to record traits not considered in the primary breeding to be sure that they do not change in an unfavorable way due to genetic correlation with traits included in the selection scheme. It is possible that genotype-environment interaction may be important when applying a breeding program to fish kept under different environmental conditions. Gunnes and Gjedrem (1978, 1981)) however, concluded that genotype-environment interaction could be ignored for Atlantic salmon and rainbow trout under Norwegian conditions. Cross-breeding schemes should be considered if non-additive genetic variance exists for traits which are included in the breeding goals. Gjedrem (1985) describes a cross-breeding scheme for fish farming. If a breeding scheme is based on cross-breeding it is also possible to keep the parent line secret and thereby protect the investment. Little heterosis is found, however, when crossing strains in salmonids (for review, see Chevassus, 1979 and Gjedrem, 1985 ). Gjerde (1988) crossed inbred lines of rainbow trout and concluded that the cost and time delay in developing and test-crossing inbred lines could only be justified by larger heterotic effects than found in his experiment. Most of the reports dealing with inbreeding in fish have shown significant inbreeding depression (Gjerde et al., 1983). Therefore it is important to keep the rate of inbreeding at a low level in a breeding scheme. TESTING AND SELECTION
Comparison of strains, progeny groups and individuals in a breeding program is dependent on a good system for the testing and recording of traits. Very often it is also necessary to record traits that are not selected. This is done ensure that important traits do not change in an undesirable way due to genetic correlation with traits included in the selection scheme. I believe that all testing, if possible, should be under commercial production conditions. If selected fish are tested under the same conditions as those which their offspring will encounter, the possibilities for mistakes are minimized.
166
T. REFSTIE
NORWEGIAN
SELECTION
PROGRAM FOR ATLANTIC SALMON
Salmon farming in Norway is more or less divided between smolt producers and sea farmers who buy smolt for growth until slaughter size. Therefore, the breeding program has to be a cooperative effort between the breeding organization and the egg producers selling eggs to smolt producers, while the testing has to be a cooperation between the breeding organization and sea farmers (Fig. 2). Selection is based on a combination of between-family and within-family selection. At present the breeding organization has two breeding centers (NFA, Kyrksaeterera and AKVAFORSK) with possibilities of testing a total of 670 progeny groups. About 350 progeny groups of Atlantic salmon are tested each year while the rest of the capacity is used for rainbow trout. The progeny groups are kept in separate units until they are big enough to be marked. This occurs about 10 months after first feeding. All recording of production traits in freshwater before smoltification is done at the breeding centers. The marking system presently used (a combination of fin-clipping and freeze-branding) has some limitations and it is possible to tag only 120 progeny groups at each of the breeding centers. Therefore, a preliminary be-
-.-
-----5molt
.... .. . . ... &*&
fish
-
- Registrations Egg /mdt
Fig. 2. Organization of the Norwegian selection program for Atlantic salmon.
APPLICATION OF BREEDIXG SCHEMES
167
tween-family selection based on growth and survival in the freshwater phase is applied. For the remaining groups a sample of fish from each progeny group (about 150 fish) is marked and placed together for further rearing in floating net cages at the breeding centers. These fish are potential brood fish for the next generation. Another sample from each progeny group is marked and divided between test stations (i.e. cooperating commercial fish farmers). All together the two breeding stations have 10 test stations for Atlantic salmon. The test stations are located on the coast from Finmark in the north to Rogaland in the south. Any genetic-environment interaction influencing the ranking should be taken into account. All measurements are taken at the test stations during the ongrowing seawater phase and all fish are slaughtered at normal marketing size. Records at the test stations, in combination with the records from the freshwater phase, are used to produce the preliminary breeding value for progeny groups. These records are also the basis for a preliminary selection of brood fish at the breeding centers. In connection with this preliminary selection of progeny groups, recording of production traits is done on all fish at the breeding centers. The fish which are initially selected are individually tagged. The final breeding value for each fish includes information on the individual itself, together with information from full- and half-sibs. The brood fish are then transferred back to the freshwater unit before the final selection of parents for the next generation. The known relationships between all individuals in the system permit the calculation of inbreeding coefficients for all matings. Matings which result in inbreeding coefficients above a stated limit can thereby be avoided. If inbreeding results in too many mating restrictions within a year-class, the system gives possibilities for mating between unrelated fish from different year-classes. Eggs from selected fish not used for progeny groups are sold to commercial smolt producers.
MULTIPLIER
STATIONS
The industry in Norway has grown to a level where it is impossible for the breeding centers to produce enough eggs from selected fish and therefore multiplier stations have been established. The multiplier stations obtain smolts from the breeding centers and apply individual selection before delivering eggs to the smolt producers. This organization is highly effective in transferring the genetic gain to the industry. In 1988, for example, 40 000 1 of Atlantic salmon eggs were produced in the system.
168 QUANTITATIVE
T. REFSTIE
MEASUREMENTS
At the breeding stations information on the following traits is recorded: On females used to produce progeny groups: egg number from each female; egg diameter. On progeny groups: mortality of eggs before eyed stage ( % ); mortality of eyed eggs ( % ); mortality of yolk-sac fry (% ) ; mortality during first feeding ( % ); mortality from end of first feeding to recording of weight in freshwater (%); growth in the freshwater phase recorded as average weight at a fixed time from first feeding; abnormal individuals; precociously mature males. Growth rate, which is recorded as weight at a fixed time from first feeding, is corrected for “tank effect” and differences in density. Registration at test stations and breeding centers: On progeny groups: early maturity ( % ); normal maturity (% ); late maturity (% ); growth as average weight at slaughter; mortality from transfer to sea until recording of early maturity (% ); mortality from recording of early maturity until slaughter. Data are corrected for differences between test stations and differences between cages within test stations. In addition, individual weight is recorded for preliminary selected fish at the breeding centers. The traits included in the selection at present are growth rate, age at maturity and survival. Selection for growth is based on growth in freshwater prior to smoltification where we have information from full- and half-sib groups, and growth in seawater where, in addition to full- and half-sib averages, we have information about the individual itself. Selection for age at first maturity is based on information from full- and half-sibs. All early maturing fish are excluded as breeders. For each trait all information is combined into a breeding value for the trait. The breeding value for each trait is given a weight in accordance with the economic importance of that trait and an overall breeding value is estimated for each fish.
APPLICATION OF BREEDING SCHEMES
FUTURE
169
POSSIBILITIES
The system allows for the estimation of genetic parameters for new traits or recalculations of genetic parameters for traits already used in selection_ It also permits new traits to be easily included in the selection scheme. Meat quality, which is the most promising new trait to be added, can easily be recorded at the test stations. Criteria for meat quality are, however, still under discussion. Selection for better disease resistance is difficult because of the difficulties encountered when measuring this trait. The freeze-branding cannot be read on dead fish and the reason for death of individual fish is not recorded. Therefore, the development of a better tagging system, at a reasonable cost, is important. Identification of dead fish, together with reasons for mortality, will give better selection criteria than percentage survival for each progeny group. Another possibility involves deliberate infection with diseases on parts of all progeny groups and then recording mortality during the outbreak. This system is costly, requiring a testing station that will not contaminate other fish farms or indigenous populations. The most promising possibility seems to be indirect selection on parameters genetically correlated with disease resistance. Measurable parameters in the immune system might allow a better selection criterion for improved disease resistance in the future.
REFERENCES Chevassus. B.. 1979. Hybridization in salmonids: results and perspectives. Aquaculture, 17: l-l 1. Cjedrem. T., 1983. Genetic variation in quantitative traits and selective breeding in fish and shellfish. Aquaculture, 33: 5 l-72. Gjedrem, T., 1985. Improvement of productivity through breeding schemes. Geo JournaI, 10 (3): 233-241. Cjerde. B., 1986. Growth and reproduction in fish and shellfish. Aquaculture, 57: 37-55. Gjerde, B., 1988. Complete diallele cross between six inbred groups of rainbow trout. Aquaculture, 75: 71-07. Gjerde, B.. Gunnes, K. and Gjedrem, T.. 1983. Effect of inbreeding on survival and growth in rainbow trout. Aquaculture, 34: 327-332. Gunnes. K. and Gjedrem. T., 1978. Selection experiments with salmon. IV. Growth of Atlantic salmon during two years in the sea. Aquaculture, 15: 19-33. Gunnes, K. and Gjedrem, T., 1981. A genetic analysis of body weight and length in rainbow trout reared in seawater for 18 months. Aquaculture. 24: 161-174. Kinghorn. B.P., 1983. A review of quantitative genetics in fish breeding. Aquaculture, 31: 283304.
163 Printed in The Netherlands
Application of Breeding Schemes T. REFSTIE Institute of Aquaculture Research (AKVAFORSK), N-6600 Sunndalsara (Norway)
ABSTRACT Refstie, T.. 1990. Application of breeding schemes. Aquaculture, 85: 163-169. Application of animal breeding theory to fish is discussed. Most of the traits investigated in fish show a high degree of phenotypic variation with possibilities of producing progeny groups that have a high number of individuals in each group. Therefore, good genetic progress can be expected even for traits with low heritabilities when the breeding schemes are based on results from fulland half-sibs. However, the control of reproduction and mating represents a limiting factor for many species. The Norwegian selection program for Atlantic salmon is described. The present breeding program utilizes a combination of between-family and within-family selection for traits such as growth rate, age at maturity and survival. This system can easily include new traits such as meat quality and disease resistance in the selection scheme.
INTRODUCTION
Animal breeding theory has only recently been applied to fish culture and is not widely practised on aquatic species. Compared with other farmed animals, fish have the advantage of a high reproductive capacity which allows for the possibility of a high selection intensity. This high fecundity makes it possible to produce progeny groups with many individuals in each group. The possibilities of testing different traits on full- or half-sib groups are better in fish than in most other species. Most of the traits investigated in fish have shown a high degree of phenotypic variation (Gjedrem, 1983). Even for traits with low heritabilities, good genetic progress can be expected when the breeding schemes are based on results from full- and half-sibs. If, however, breeding schemes could be used in practice, control of reproduction and mating is necessary. At present we have this control in only a few commercially farmed species, and breeding schemes are applied to only a few species, of which salmonids seem to be the most important. In this paper, the Norwegian breeding program for Atlantic salmon will be used as an example of the application of a breeding scheme.
0 1990 Elsevier Science Publishers B.V.
164
T. REFSTIE
Before possibilities of applying breeding schemes can be discussed, breeding goals have to be defined. Breeding goals should be established using the input of both industry and consumers. All traits of economic importance which show genetic variation should be defined as breeding goals. Even if a trait is economically important, the problems and costs associated with measuring it should be considered before including such information in a breeding scheme. Gjedrem (1983,1985) identified growth rate, food conversion efficiency and survival rate as traits to be considered in formulating breeding goals in most production systems involving fish species. Meat quality, age at maturity and fecundity are traits which might be of economic importance for some species in some production systems. Because some traits cannot be measured on live fish, a statement of breeding goals will influence selection strategy. SELECTION OF STRAIN
If different strains or populations are available, all breeding programs should start with the collection, comparison and selection of the best genetic material available. The value of testing strains and selecting the best for farming can circumvent several years of within-strain selection. Differences in performance between strains within species have been well documented (for review, see Kinghorn ( 1983 ) ). At AKVAFORSK, eggs from about 40 Norwegian Atlantic salmon strains were collected and compared during the years 1972-1976. At the same time progeny groups were produced within strains. This allowed the estimation of genetic parameters for important traits and the testing of strains to be done simultaneously. Fig. 1 gives an example of differences in growth rate, measured as weight at slaughter, found between strains and between progeny groups within strains 70
560 -u -C o\ p 50 3 $
40
famify no Fig. 1. Ranking of strains of Atlantic salmon and full-sib families within a strain for body weight after 2 years in sea cages (year-class 1973) (Gjerde, 1986).
APPLICATION OF BREEDING SCHEMES
165
for one year-class. The great differences found between strains and between progeny groups within strains indicate the importance of starting a selection scheme on the best available genetic material. The differences between fullsib families within a strain also illustrate the possibilities for further progress through selection. ESTIMATION OF GENETIC PARAMETERS
To apply a simple selection scheme it is only necessary to know if the trait or a genotype of traits shows genetic variation. In order to optimize a breeding scheme or predict responses to selection, reliable phenotypic and genetic parameters are required (for review, see Gjerde ( 1986) ). It might also be necessary to record traits not considered in the primary breeding to be sure that they do not change in an unfavorable way due to genetic correlation with traits included in the selection scheme. It is possible that genotype-environment interaction may be important when applying a breeding program to fish kept under different environmental conditions. Gunnes and Gjedrem (1978, 1981)) however, concluded that genotype-environment interaction could be ignored for Atlantic salmon and rainbow trout under Norwegian conditions. Cross-breeding schemes should be considered if non-additive genetic variance exists for traits which are included in the breeding goals. Gjedrem (1985) describes a cross-breeding scheme for fish farming. If a breeding scheme is based on cross-breeding it is also possible to keep the parent line secret and thereby protect the investment. Little heterosis is found, however, when crossing strains in salmonids (for review, see Chevassus, 1979 and Gjedrem, 1985 ). Gjerde (1988) crossed inbred lines of rainbow trout and concluded that the cost and time delay in developing and test-crossing inbred lines could only be justified by larger heterotic effects than found in his experiment. Most of the reports dealing with inbreeding in fish have shown significant inbreeding depression (Gjerde et al., 1983). Therefore it is important to keep the rate of inbreeding at a low level in a breeding scheme. TESTING AND SELECTION
Comparison of strains, progeny groups and individuals in a breeding program is dependent on a good system for the testing and recording of traits. Very often it is also necessary to record traits that are not selected. This is done ensure that important traits do not change in an undesirable way due to genetic correlation with traits included in the selection scheme. I believe that all testing, if possible, should be under commercial production conditions. If selected fish are tested under the same conditions as those which their offspring will encounter, the possibilities for mistakes are minimized.
166
T. REFSTIE
NORWEGIAN
SELECTION
PROGRAM FOR ATLANTIC SALMON
Salmon farming in Norway is more or less divided between smolt producers and sea farmers who buy smolt for growth until slaughter size. Therefore, the breeding program has to be a cooperative effort between the breeding organization and the egg producers selling eggs to smolt producers, while the testing has to be a cooperation between the breeding organization and sea farmers (Fig. 2). Selection is based on a combination of between-family and within-family selection. At present the breeding organization has two breeding centers (NFA, Kyrksaeterera and AKVAFORSK) with possibilities of testing a total of 670 progeny groups. About 350 progeny groups of Atlantic salmon are tested each year while the rest of the capacity is used for rainbow trout. The progeny groups are kept in separate units until they are big enough to be marked. This occurs about 10 months after first feeding. All recording of production traits in freshwater before smoltification is done at the breeding centers. The marking system presently used (a combination of fin-clipping and freeze-branding) has some limitations and it is possible to tag only 120 progeny groups at each of the breeding centers. Therefore, a preliminary be-
-.-
-----5molt
.... .. . . ... &*&
fish
-
- Registrations Egg /mdt
Fig. 2. Organization of the Norwegian selection program for Atlantic salmon.
APPLICATION OF BREEDIXG SCHEMES
167
tween-family selection based on growth and survival in the freshwater phase is applied. For the remaining groups a sample of fish from each progeny group (about 150 fish) is marked and placed together for further rearing in floating net cages at the breeding centers. These fish are potential brood fish for the next generation. Another sample from each progeny group is marked and divided between test stations (i.e. cooperating commercial fish farmers). All together the two breeding stations have 10 test stations for Atlantic salmon. The test stations are located on the coast from Finmark in the north to Rogaland in the south. Any genetic-environment interaction influencing the ranking should be taken into account. All measurements are taken at the test stations during the ongrowing seawater phase and all fish are slaughtered at normal marketing size. Records at the test stations, in combination with the records from the freshwater phase, are used to produce the preliminary breeding value for progeny groups. These records are also the basis for a preliminary selection of brood fish at the breeding centers. In connection with this preliminary selection of progeny groups, recording of production traits is done on all fish at the breeding centers. The fish which are initially selected are individually tagged. The final breeding value for each fish includes information on the individual itself, together with information from full- and half-sibs. The brood fish are then transferred back to the freshwater unit before the final selection of parents for the next generation. The known relationships between all individuals in the system permit the calculation of inbreeding coefficients for all matings. Matings which result in inbreeding coefficients above a stated limit can thereby be avoided. If inbreeding results in too many mating restrictions within a year-class, the system gives possibilities for mating between unrelated fish from different year-classes. Eggs from selected fish not used for progeny groups are sold to commercial smolt producers.
MULTIPLIER
STATIONS
The industry in Norway has grown to a level where it is impossible for the breeding centers to produce enough eggs from selected fish and therefore multiplier stations have been established. The multiplier stations obtain smolts from the breeding centers and apply individual selection before delivering eggs to the smolt producers. This organization is highly effective in transferring the genetic gain to the industry. In 1988, for example, 40 000 1 of Atlantic salmon eggs were produced in the system.
168 QUANTITATIVE
T. REFSTIE
MEASUREMENTS
At the breeding stations information on the following traits is recorded: On females used to produce progeny groups: egg number from each female; egg diameter. On progeny groups: mortality of eggs before eyed stage ( % ); mortality of eyed eggs ( % ); mortality of yolk-sac fry (% ) ; mortality during first feeding ( % ); mortality from end of first feeding to recording of weight in freshwater (%); growth in the freshwater phase recorded as average weight at a fixed time from first feeding; abnormal individuals; precociously mature males. Growth rate, which is recorded as weight at a fixed time from first feeding, is corrected for “tank effect” and differences in density. Registration at test stations and breeding centers: On progeny groups: early maturity ( % ); normal maturity (% ); late maturity (% ); growth as average weight at slaughter; mortality from transfer to sea until recording of early maturity (% ); mortality from recording of early maturity until slaughter. Data are corrected for differences between test stations and differences between cages within test stations. In addition, individual weight is recorded for preliminary selected fish at the breeding centers. The traits included in the selection at present are growth rate, age at maturity and survival. Selection for growth is based on growth in freshwater prior to smoltification where we have information from full- and half-sib groups, and growth in seawater where, in addition to full- and half-sib averages, we have information about the individual itself. Selection for age at first maturity is based on information from full- and half-sibs. All early maturing fish are excluded as breeders. For each trait all information is combined into a breeding value for the trait. The breeding value for each trait is given a weight in accordance with the economic importance of that trait and an overall breeding value is estimated for each fish.
APPLICATION OF BREEDING SCHEMES
FUTURE
169
POSSIBILITIES
The system allows for the estimation of genetic parameters for new traits or recalculations of genetic parameters for traits already used in selection_ It also permits new traits to be easily included in the selection scheme. Meat quality, which is the most promising new trait to be added, can easily be recorded at the test stations. Criteria for meat quality are, however, still under discussion. Selection for better disease resistance is difficult because of the difficulties encountered when measuring this trait. The freeze-branding cannot be read on dead fish and the reason for death of individual fish is not recorded. Therefore, the development of a better tagging system, at a reasonable cost, is important. Identification of dead fish, together with reasons for mortality, will give better selection criteria than percentage survival for each progeny group. Another possibility involves deliberate infection with diseases on parts of all progeny groups and then recording mortality during the outbreak. This system is costly, requiring a testing station that will not contaminate other fish farms or indigenous populations. The most promising possibility seems to be indirect selection on parameters genetically correlated with disease resistance. Measurable parameters in the immune system might allow a better selection criterion for improved disease resistance in the future.
REFERENCES Chevassus. B.. 1979. Hybridization in salmonids: results and perspectives. Aquaculture, 17: l-l 1. Cjedrem. T., 1983. Genetic variation in quantitative traits and selective breeding in fish and shellfish. Aquaculture, 33: 5 l-72. Gjedrem, T., 1985. Improvement of productivity through breeding schemes. Geo JournaI, 10 (3): 233-241. Cjerde. B., 1986. Growth and reproduction in fish and shellfish. Aquaculture, 57: 37-55. Gjerde, B., 1988. Complete diallele cross between six inbred groups of rainbow trout. Aquaculture, 75: 71-07. Gjerde, B.. Gunnes, K. and Gjedrem, T.. 1983. Effect of inbreeding on survival and growth in rainbow trout. Aquaculture, 34: 327-332. Gunnes. K. and Gjedrem. T., 1978. Selection experiments with salmon. IV. Growth of Atlantic salmon during two years in the sea. Aquaculture, 15: 19-33. Gunnes, K. and Gjedrem, T., 1981. A genetic analysis of body weight and length in rainbow trout reared in seawater for 18 months. Aquaculture. 24: 161-174. Kinghorn. B.P., 1983. A review of quantitative genetics in fish breeding. Aquaculture, 31: 283304.