CATDOLL : CATDOLL: Breeding Programs in Aquaculture - 10 Breeding Strategies

10.1 Introduction

The primary goal of a breeding program is to shift the mean of a trait in the desired direction for a normally distributed trait or a trait with two (or more) discrete classes (e.g., survival) to increase the frequency of the desired trait class. The change in the overall mean or class frequency from one generation to the next is called the selection response. The selection response for a normally distributed trait is shown in Figure 7.4.

Applying different breeding strategies can result in either selection response or genetic gain. For a long-term breeding goal, the only strategy appropriate for breeding nuclei is some type of purebred breeding for additive genetic improvement. Breeding strategies that can be used to produce commercial fry are less restrictive than breeding done in the nucleus. Any kind of hybridization, ploidy manipulation, and sex manipulation can be used if it can further improve the productivity of commercial fry. Any breeding strategy for commercial fry production that limits the progress of additive genetic performance in the nucleus should be avoided (e.g., use of highly inbred lines in a hybridization program, see below)

All breeding programs should begin with the collection, comparison, and selection of the best genetic material (see Chapter 16 for more details). The value of experimenting with strains and selecting the strains best suited for agricultural production may be equivalent to several generations of within-strain selection, as shown by Bentsen et al. (1998) for tilapia. Another example of the importance of selecting a base population is shown in Figure 10.1, which illustrates the differences in weight at slaughter between strains of Atlantic salmon and between full-sib families within a strain (Gjedrem, 1979b). The large differences between strains and between full-sib families within a strain illustrate the importance of starting a breeding program with the best genetic material. The differences between full-sib families within a strain also illustrate the potential for further improvement through selection.

Inbreeding in an infinitely large population is defined as the mating of individuals that are more closely related to each other than individuals mating randomly in a population. The populations actually used in most aquaculture programs are finite populations because they have a finite number of members. All finite populations experience some degree of inbreeding, which is determined by the number of individuals that contribute offspring to each offspring. The inbreeding of a population is measured by the inbreeding coefficient (F) as defined and described in Chapter 6. The inbreeding coefficient represents the amount of inbreeding that has accumulated since a specific point in the ancestry of the population. The inbreeding coefficient is meaningful only after a specific time in the past has been selected, beyond which ancestry is not considered and at which time all alleles are considered independent.

Knowing the effective population size (Ne) for any breeding structure, the inbreeding rate per generation (ΔF) can be obtained:

where Ne is a function of the number of sires and dams used as parents in each new generation (see Chapter 6 for other factors affecting effective population size):

Nm and Nf are the number of males and females, respectively. Assuming there is no genetic relationship between males and females, when using 50 males and 50 females, Ne = 100 and ΔF = 0.005 or 0.5%. This is probably an acceptable inbreeding rate for most traits.

The rate of inbreeding is strongly dependent on the smaller sex. If the number of sires is reduced to 30, the number of dams must be increased to 150 to make Ne = 100. If the number of sires is 20, then Ne cannot exceed 80, no matter how many dams are used. The variance between strain means increases and the variance within strains decreases, in other words, strain differentiation and genetic homogeneity within strains. In a closed population under selection, it is impossible to prevent the increase of inbreeding generations. Pante et al. (2001b) concluded that pedigree information is necessary to accurately estimate the rate and level of inbreeding because the effective population size (Ne) is a poor estimate of the rate and level of inbreeding. If the degree of inbreeding is too high, some unrelated animals should be used as parents to reduce the problem of inbreeding. For more information, see Chapter 6.

Biological Unity

Inbreeding can be a powerful technique for developing strains for research purposes. Highly inbred strains are genetically stable, which is important for using "standard" inbred strains for experimental purposes, especially for laboratory animals to be used in bioassays and other experiments (Komen, 1990).

Inbred lines and hybrids

In actual breeding, inbreeding is done intentionally only to produce inbred lines for crosses that exploit non-additive genetic variation. Inbreeding is almost always detrimental, and breeders usually try to avoid it as much as possible.

Inbreeding depression is the result of inbreeding. It mainly leads to a decrease in the average phenotypic value of traits related to reproductive capacity (fertility, egg size, hatchability) or physiological efficiency (fry deformity, growth rate, survival rate).

Inbreeding depression is the difference in average performance between an inbred population and a base population. Because traits related to reproductive and physiological efficiency often exhibit inbreeding depression, it is important to keep inbreeding rates low in breeding programs.

Gjerde et al. (1983) studied the effects of three levels of inbreeding (F = 0.25, 0.375 and 0.5) on the survival and growth rate of rainbow trout, Table 10.1. The average inbreeding depression (all levels of inbreeding) was 10% for eyed eggs, 5.3% for hatched eggs and 11.1% for fry. There was no linear relationship between inbreeding and inbreeding depression. The growth of fingerlings did not show significant inbreeding depression, while the growth of adults showed increasing growth depression with increasing inbreeding. For each 10% increase in the inbreeding coefficient, the inbreeding depression coefficient was 4.5%, 5.3% and 6.1%, respectively (for inbreeding levels of 0.25%, 0.375% and 0.5%, respectively).

Obtained by full-sib mating of consecutive one-generation (F=0.25), two-generation (F=0.375) and three-generation (F=0.5)

Su et al. (1996) also found in rainbow trout that for every 10% increase in inbreeding coefficient, female spawning age was delayed by 0.53% and spawning amount was reduced by 6.1%. Inbreeding had no significant effect on male egg size and spawning age. For every 10% increase in body weight due to inbreeding, inbreeding depression was between 2.26% and 5.77%, and inbreeding depression tended to increase with increasing body weight.

In experiments with channel catfish, Bondari and Dunham (1987) reported that inbreeding (25%) increased the number of days required for eggs to hatch but had no significant effect on egg weight or hatchability.

Typical levels of inbreeding depression are shown in Table 10.2. At this point, it is important to emphasize that the inbreeding coefficient for a round of full-sib matings is 0.25 and the inbreeding coefficient for a round of half-sib matings is 0.125.

As mentioned previously, inbreeding should be done intentionally only to produce inbred lines for crosses that exploit non-additive genetic variation. In general, the resources and time required to produce, maintain, and replace inbred lines would be better utilized by improving additive genetic performance in purebred breeding (Gjedrem, 1985; Gjerde, 1988).

Breeding methods to avoid inbreeding can be divided into three main categories:

Using large random mating populations

Eliminating inbreeding using systematic hybridization protocols

Producing hybrids by crossing strains

Using a large, randomly mating population is the simplest approach and requires only that the breeder take steps to ensure that a large number of fish will provide offspring for the next generation.

Crossbreeding is a well-known genetic improvement method that is also used in aquaculture. Crossbreeding refers to the mating between species, varieties, populations, strains or inbred lines. The main goal of crossbreeding is to exploit non-additive genetic variance (heterosis). When an inbred line is unselected, the average of all its crosses should be equal to the average of the population of crosses derived from it. Therefore, crossbreeding after inbreeding will not produce any improvement, and selection must be applied at some stage if any improvement is to be achieved. Therefore, crossbreeding should be considered as a complement to additive genetic improvement programs.

Heterosis, also known as hybrid vigor, can be defined as the phenomenon whereby offspring exceed the average of their parents in one or more traits, a reversal of inbreeding depression, obtained by mating related individuals. Both phenomena are almost universally distributed in plants and animals, especially in relation to reproductive fitness. There are two methods commonly used to estimate heterosis. The first is to compare the hybrid offspring with the average of the parental lines/strains, and the second is to compare the hybrid offspring with the average of the best parental line/strain. If the parents are from different gene pools, the heterozygosity of the hybrid increases, and therefore heterosis is expected to increase. The extent to which heterosis for a given trait increases depends on the genetic distance between the parental populations.

The relative gains from crossing and selection depend on the size of the additive and nonadditive variance in the trait or traits in question. If the nonadditive variance is large, large gains may be obtained from crossing (see Section 10.3.4).

General combining ability refers to the average value of offspring traits of a series of hybrid combinations of a parent variety with many other varieties. For example, after hybridization of variety A with other varieties such as B, C, D, E, the offspring yield is relatively high, indicating that variety A has a high general combining ability. Special combining ability refers to the hybridization of variety A with other varieties such as B, C, D, E, where only one combination AB has a high average yield trait, while the offspring of other combinations such as AC, AD, AE are average or low. The ability of this AB combination is characterized by high special combining ability.

GCA = Additive Effect

SCA = epistasis and dominance effect

Differences in general combining ability are due to additive variance (A) and A×A interactions in the base populations. Differences in specific combining ability are attributable to non-additive genetic variance and epistasis. Table 10.3 shows how general and specific combining abilities are estimated when four populations are crossed.

The general combining ability (GCA) of population A and population B can be estimated as follows:

Special Compatibility (SCA) of A×B and B×A:

Generally, it is difficult to measure differences in specific combining ability and to exploit these effects in breeding programs. Creating and maintaining inbred lines is almost the only means of exploiting SCA commercially, although some uses of SCA can be achieved through hybridization. However, it is impossible to determine the specific combining ability of a hybrid combination without making and testing the specific hybrid combination.

(Measured by Teacher Deng Fei)

Repeated reciprocal selection is a hybridization scheme that exploits both general and specific combining abilities. The theoretical basis for RRS was given by Comstock et al. (1949) and Dickson (1952). RRS starts with two populations, line A and line B.

The crosses are reciprocal, with some A-line females mated with B-line males and some B-line females mated with A-line males. The traits of the crosses are then measured for improved traits and the parents are judged based on the performance of the offspring. Only the best parent is selected and the remaining parents, as well as all the crosses, are used only to test the combining ability of the parents. They are then discarded. The selected individuals must be mated again with their own parents to produce the next generation of parents that are tested. They are crossed again as before and the cycle repeats. According to Falconer and Mackay (1996), the RRS program is used by commercial breeders of poultry and has given good results in corn, but direct comparisons with other selection methods have not been encouraging.

When heterozygotes are superior to homozygotes, the phenomenon is called dominance, see Figure 10.2 (Falconer and Mackay, 1996). The only way to produce a population of heterozygous individuals is to have two lines fixed for different alleles and then have an F1 gene in which all individuals are heterozygous. In a non-inbred population, no more than 50% of the individuals can be heterozygous for a particular pair of alleles. Therefore, if heterozygotes for a particular pair of alleles are superior to homozygotes in terms of merit, then inbreeding and hybridization will be a better means of improvement than without inbreeding selection.

Furthermore, only when there is excess dominance for a desired trait or combination of traits can inbreeding and hybridization achieve effects that would not be achieved without selection for inbreeding. The existence and importance of excess dominance has been widely debated, but experimental evidence generally suggests that for most traits studied, the phenomenon of excess dominance is not important (Falconer and Mackay, 1996).

10.3.5 Diallel Cross

Diallel crosses are a commonly used experimental design for crossing inbred lines or different lines or populations, i.e. each line/population is crossed with one other line. For p lines, this process produces a maximum of p2 combinations. Diallel crosses are often used to establish a base population before starting a breeding program. Crossing is often used to introduce new genes from an unfamiliar population into a local line. This is often a simple and very cheap way to improve a local line. However, before introducing new varieties, the population should be tested under existing local conditions.

The GIFT project (Genetic Improvement of Farmed Tilapia) was based on diallel hybrids. Four Asian farmed strains and four African wild strains were completely diallel crossed (8 × 8 = 64 combinations) to study the heterosis for growth performance and survival. The results are shown in Table 10.4 (Bentsen et al., 1998).

In a three-way cross, the F1 of two lines (e.g., the line that needs high productivity) is crossed with a third line. In a four-way cross, the F1 of two different lines are crossed. Backcrossing involves only two lines, with the F1 mating with one of the lines used in the first cross.

Crossbreeding is widely used in livestock production, and most animals used for meat production are the offspring of three-way crosses or backcrosses. In aquaculture, these methods are rarely used.

Diallel crosses to test potential populations are often the starting point for building a synthetic population, as has been done with Atlantic salmon, rainbow trout, and tilapia.

Composite populations are made up of different numbers of parent populations, breeds, or lines. When breeding a composite population, breeders try to create some new population that combines the strengths of the parent populations. New populations are created by randomly mating the F1 and offspring from a series of selected inbred lines or different populations, or usually by planned mating.

Synthetic populations are expected to be more heterozygous than the parental lines, and they should show some heterosis. After the synthetic population is reduced in size, inbreeding can and often does reduce this heterosis. In addition, if losses due to recombination are important, these losses will be evident in the offspring of the synthetic population.

The first step in determining whether hybridization has a place in a breeding strategy for a particular species is to evaluate all possible crosses between different strains or species for the economic trait in question. If the number of available strains is large, it is necessary to select the hybrid combinations that are most likely to produce valuable results. It may be advantageous to use strains from very different sources and to combine strains with favorable traits. In Israel, a hybrid breeding program is currently underway using crosses of strains from the common carp (Wohlfarth et al., 1983).

Secondly, inbred lines should be developed and the crosses tested under natural conditions to find the most valuable hybrids for agriculture. This breeding system is particularly designed to exploit non-additive genetic variation. One of the practical difficulties here is that inbred lines are difficult to keep running because of high mortality (inbreeding depression). Bakos (1979; 1987) reported the results of using inbred lines of carp in a crossbreeding program.

Thirdly and finally, if possible, a reciprocating recursive selection (RRS) program should be evaluated to determine the relative importance of general and specific combination abilities. RRS can only be used for organisms that spawn multiple times, so it cannot be used in, for example, Pacific salmon. Fishing would also be very difficult in Atlantic salmon, as most males die after their first spawning, and a large number of females die. Other species, such as tilapia and rainbow trout, may be better suited to the application of an RRS program.

A significant advantage of using crosses between selected lines in a breeding program is that it enables breeders to protect their genetically improved material. Only crossbred animals are sold and purebred breeding stock is not released.

Chevassus (1979) reviewed the status of hybridization among salmonid species. He concluded that in most cases where hybrids were reared in the same environment, the offspring showed growth that was intermediate or at best equal to that of the better parent species. This is consistent with the results of Refstie (1983a) on hybrids of four salmonids (Atlantic salmon, brown trout, sea trout, and Arctic char). Neither the growth nor the survival of the hybrid fish exceeded the performance results of Atlantic salmon.

Some experiments have found good results in terms of survival rates, with hybrids often being similar to or even superior to the hardiest hybrid combinations.

Benzie et al. (1995) found no evidence of hybrid vigor in growth rate when crossing two tiger shrimp species, Penaeus Monodon and P. esculentus. The growth rate of the hybrids was similar to or lower than that of pure P. monodon.

Gjerde and Refstie (1984) investigated the effects of heterosis in crossbreeding five Norwegian strains of Atlantic salmon. They found no significant heterosis effects on either growth rate or survival (Table 10.5). Similarly, Friars et al. (1979) found no heterosis effects on growth rate of Atlantic salmon fry. However, in rainbow trout, Gall (1975) and Ayles and Baker (1983) reported significant heterosis on body weight in crossbreeding of rainbow trout strains.

Common carp species have been systematically hybridized. The significance of heterosis in growth rate, survival rate, and cold tolerance in crosses between wild and domesticated strains from Europe, Russia, China, and Japan has been repeatedly reported (Table 10.6) (Hulata, 1995).

Wohlfarth (1993) summarized experimental data collected over 20 years of research on carp in Israel and concluded that "heterosis in growth is a common but not universal phenomenon in carp." Typically, no heterosis was found when one of the parents was Dor-70, a long-term population selection experiment for faster growth. Gjerde et al. (1999) estimated heterosis for body weight and survival in Lochu carp and concluded that hybridization in Lochu carp populations in India appears to be of little practical interest.

The already mentioned GIFT hybridization experiment with tilapia (Bentsen et al., 1998) showed that of the 22 crosses showing clear heterosis, only 7 outperformed the best pure strain, with the maximum gain being about 11%. In general, non-additive genetic effects on body weight were modest compared with additive and inverse effects on body weight.

Overall, non-additive genetic effects on body weight were minor compared with additive and reciprocal effects.

Results reported by Wohlfarth (1993) and Bentsen et al. (1998) suggest that the expression of non-additive genetic effects may be more sensitive to environmental variation than additive effects. Because genotype-environment interactions affect non-additive genetic performance, heterosis may not be expressed well in certain farm environments. In such cases, it may be necessary to produce specialized hybrids for certain farm environments.

Knibb et al. (1997) found that hybridization between sea bream species produced little heterosis, due to a lack of inbreeding and genetic differentiation. Knibb (2000) reviewed the results of several hybridization trials and concluded that in all species, hybrids generally resembled the average of their parents. Given the large number of attempts to produce new hybrid fish, very few (significantly less than 1%) were able to be sustained in commercial production.

The search for sterile hybrids may become very important because such hybrids do not transfer food into the gonads and therefore have superior production traits. Since tilapia has obtained unisexual offspring in several interspecific crosses, such unisexual (male) culture is considered to be the best solution to the overpopulation caused by the high fecundity of tilapia in almost any pond conditions. In addition, males also grow faster than females. Pruginin et al. (1975) listed several cross combinations that produced 100% male offspring, while Hulata et al. (1983) recommended the use of progeny testing to ensure that 100% male offspring are obtained in tilapia. It has been reported that promising crosses between Nile tilapia and Oreochromis aureus in Israel produce almost all male offspring (Hulata, 1995).

The breeding method or strategy of additive genetic improvement within a population is called purebred breeding and is the method of selection for continuous genetic improvement over a long period of time. Inbreeding must be avoided and individuals with a majority of positive (desirable) genes are selected as parents for the next generation. Individuals with a majority of positive gene alleles generally show good production results. These "good genes" and characteristics are passed on to their offspring. Individuals with a majority of positive genes are considered to have high breeding values.

The reproductive value of an individual cannot be measured directly. Nor can it be measured with 100% accuracy. Therefore, the true breeding value will be unknown and largely obscured by effects caused by systematic and random environmental effects and by interactions between genes. See Chapter 14 for a discussion of genotype-environment interactions.

Breeding values ​​can be estimated primarily by recording the phenotypic values ​​of the product of a gene, the trait (or by using genetic markers linked to QTLs as described in Chapter 19). Phenotypic records can be obtained from the individual itself or from relatives as full and half sibs, offspring, or parents. Records of related individuals can be used because the individual and its relatives share genes in common. In general, information from close relatives is more valuable than information from more distant relatives. Thus, records of full sibs are more valuable than records of half sibs because an individual shares a greater proportion of common genes with its full sibs than with its half sibs. Records on progeny are particularly interesting because the breeding value of an individual is strictly defined as the value of the individual judged by the average of its progeny.