Our overall objective is to gain empirical evidence on the genetic stability of transgenic diploid organisms with respect to their genetic background and to investigate potential long-term risks associated with accidental or intentional release f transgenic fish. Our research is expected to have general implications to other diploid organisms as well. Specific objectives are to: 1) Use the Japanese medaka (Oryzias latipes) as a model organism to determine how insertion of a growth hormone gene affects overall fitness and its components (viability, mating success, and fecundity) by comparing transgenics with non-transgenics, derived from the same heterogeneous base population, under controlled experimental conditions. 2) Determine if eight generations of natural selection will modify expression of the transgene such that transgenic individuals become a greater environmental risk than initial screening would suggest. Here we outline the rationale, methodology, and experimental design for the study.

Key Words: Performance testing, fish, transgenic, risk, fitness components

Transgenic technology has attracted commercial and scientific interest because fast growing fish are desirable in both aquaculture and sport fishery and DNA sequences for growth hormone (GH) genes and cDNA's are well characterized. GH sequences have already been transferred to several game and commercial fish species: channel catfish (Dunham et al., 1988), northern pike (Schneider et al., 1989), rainbow trout (Chourrout et al., 1986; Maclean et al., 1987, Guyomard et al., 1990; Penman et al., 1990; Rokkones et al., 1989) and Atlantic salmon (Rokkones et al., 1989). Outdoor tests of transgenic common carp have begun in the United States and China (Hallerman and Kapuscinski, 1990). Because larger body size often confers higher fitness in fish due to enhanced competitive ability (Magnuson, 1962) and fecundity (Gall and Gross, 1978; Huang and Gall, 1990), transgenic individuals could eliminate their wild-type conspecifics in nature. Thus, assessing potential risks of transgenic fish to the environment is imperative.

The environmental risk posed by transgenic organisms is similar to that of introduced species. Mechanisms for domination of native species by introduced fish include competition via interference or exploitation, predation, inhibition of reproduction, and hybridization (Kapuscinski and Hallerman, 1990). Other types of undesirable outcomes of transgenic animals include: creation of new pests, enhancement of effects of existing pests, harm to non-target species, disruptive effects on biotic communities through competition or interference (Tiedje et al., 1989). These mechanisms provide a useful framework for anticipating the impact of released transgenic fishes on natural aquatic communities and can be addressed by prior collection of appropriate performance data from transgenic fish reared in confinement (Levin et al., 1987). However, emphasis should be placed on environmental conditions under which transgenic organisms are likely to remain viable or increase in number (Tiedje et al., 1989).

Common suggestions (Kapuscinski and Hallerman, 1990; Hallerman and Kapuscinski, 1990; Tiedje et al., 1989; Regal, 1987) regarding development of risk assessment models, methods of ascertaining degree of hazard, and arguments for or against the release of transgenic organisms include: 1) the magnitude of the problem is such that a team effort involving biotechnologists, population biologists, and ecologists is needed, 2) no overall conclusion can be made as to the general safety or danger of engineered organisms, rather a case-by-case review is needed for each transgenic organism, 3) characterization of transgenic stocks through performance evaluation and risk assessment modeling is needed, 4) evaluations should be conducted first in closed laboratory systems, then in experimental mesocosms, and finally in more extensively controlled systems that simulate natural systems, with the result of each step used in the design of the next experiment. These suggestions were incorporated into the methodology devised for our research on Japanese medaka.

The test environment and duration. Because transgenic organisms cannot be released into the wild, the potential environmental impact of such organisms must first be inferred from inspection of causal components in a laboratory environment. If a transgenic organism exhibits superior fitness in an optimal laboratory environment in terms of growth, fecundity, competitive ability etc., the organism could be a hazard in the wild despite nature's less than optimal conditions and greater complexity. On the other hand, if a transgenic organism has low fitness in an optimum laboratory environment, the common conclusion is that it will not thrive or compete effectively in the wild where conditions are usually suboptimal. However, this argument has one flaw: evolution. Even if a transgenic organism initially displays low fitness, the cumulative action of natural selection over several generations could modify the expression of the transgene and make the organism more successful than native species. The net result could be an ecological catastrophe. Thus, multigenerational studies are needed to assess risk of transgenic release.

The study organism. To test the hypothesis of selective modification within a reasonable amount of time and expense, the study organism must be suitable for laboratory experimentation and have a relatively short generation interval. As a result, bacterial species are often used as model organisms. However, to make inferences concerning vertebrate transgenics, the model organism should mimic the behavior and physiology of the target species as close as possible. Selective mechanisms associated with several fitness components, such as sexual selection, are linked to changes in mating behavior and fecundity, which are impacted by changes in physiology. Considerable interest exists for the production of transgenic vertebrates, especially fish. Unfortunately a multi-generational study with fish of commercial interest would require more than a decade to complete. However, a non-commercial fish, Japanese medaka (Oryzias latipes), meets the above criteria very well. Medaka are small (< 4 cm), have a short generation interval (10 weeks), and are highly conducive to laboratory experimentation. For these reasons, we are using medaka as a model organism to test for selective modification of transgenes in a vertebrate organism. Because evolutionary mechanisms are universal, inference is directly applicable to commercial species; i.e. if one species of transgenic organism can increase in fitness due to selective modification across generations, then other species can as well.

We are using the Purdue Orange-Red (POR) strain of medaka to produce transgenic founder lines. This strain was established from a large heterogeneous, randomly mating population of medaka homozygous for the mutant color Orange-Red.

Several attributes of medaka also contribute to its value as a model study organism: 1) medaka are small (25 to 50 mm in total length), easy to study in aquaria, and easy to rear and maintain in the laboratory (Yamamoto, 1975). 2) They have a short generation interval (2-3 months); thus, several generations can be studied within a relatively short time. 3) Females are highly fecund: they can lay an average of 25 eggs/day for up to 4 months under proper conditions of photoperiod, temperature and food supply (Hirshfield, 1980); thus, an adequate supply of eggs for production of transgenic fish is readily available. In addition, the potential for rapid evolution is greater in species with high fecundities and short generation intervals (Kapuscinski and Hallerman, 1990). 4) Zygotes are easily collected: after fertilization, eggs remain attached to the female's vent by filaments and are normally brushed off onto aquatic vegetation. 5) The development, and endocrinology of medaka are well known; for example, Yamamoto (1976) lists over 1000 references on medaka. 6) Several features of medaka behavior facilitate study of ecological and mating competition: when food is spatially clumped, access to food depends on an individual's position in a dominance hierarchy, and larger fish are socially dominant to smaller ones (Magnuson, 1962). The courtship ritual is complex and highly predictable in time (usually within 1 hr of first daylight, Hirshfield, 1980). Thus, mate competition and mating success of transgenic and non-transgenic males can be directly observed. Because both ecological and mating competition are likely to be influenced by relative body size, our protocol includes experiments that evaluate both of these phenomena in transgenic individuals. 7) We will be able to produce a large founder population which will provide sufficient genetic variation for selection experiments to test theory related to the evolution of transgenic organisms. Any selective disadvantage associated with novel gene function may be quite labile; as a result, the magnitude of the disadvantage may depend on the genetic background which can be modified by subsequent selection. Natural selection will continue to increase the fitness of the engineered organism either by eliminating the gene or by modifying its expression (Lenski and Nguyen, 1988; Tiedje et al., 1989).

Founders. Founders are being produced using DNA microinjection of the POR strain. Initially several founder lines will be produced. Each founder is a unique individual with different transgene integration site(s), gene copy number and level of transgene expression. Founders will be chosen based on phenotypic expression and will be progeny tested for germline transmission of the transgene. Each founder will be backcrossed to the POR strain for 2 generations before homozygous individuals are produced. Backcrossing will allow the transgene to be integrated into a heterogeneous genetic background and reduce the rate of inbreeding to less than 3% per generation.

The transgene. We are inserting using both the chicken ß-actin (CBA) promoter and the Atlantic salmon growth hormone promoter (sGH) driving the expression of the human growth hormone (hGH) gene. Growth hormone is a major regulator of cell growth and proliferation as well as a regulator of metabolism. Introduction of GH transgenes is expected to have a major impact on the way fish grow through out their life cycle. Growth hormone, also referred to as somatotropin, is normally synthesized in the anterior pituitary and has a wide range of physiological effects on growth and metabolism (for review, see Boyd and Bauman, 1988). GH can directly or indirectly stimulate proliferation of several cell types such as muscle and bone, and influence protein, lipid, carbohydrate, and mineral metabolism. The effects of GH are to some degree mediated by somatomedins. The regulation of metabolism by GH is chronic in nature causing an increase in protein synthesis and carbohydrate and mineral metabolism.

The effect of administration of GH to fish has been investigated since the 1950's (for review see Donaldson et al., 1979; Weatherley and Gill, 1987). Injection of GH or implantation of GH releasing pellets have been shown to improve growth rate and feed conversion in several species of fish. Early studies used bovine GH (bGH) purified from pituitaries and later Gill et al. (1985) showed that bGH, and chicken GH purified from the enterobacterium Escherichia coli expressing the recombinant GH fusion genes was able to increase growth rate and feed conversion. Recombinant salmon GH (sGH) administration to hypophysectomized killifish (Fundulus heteroclitus) causes an increase in plasma steroids, and in vitro incubation of gonadal tissues with sGH stimulates gonadal steroidogenesis in both the ovary and testis (Singh et al., 1988). This is an indication that GH may have a direct effect on reproductive traits. These results demonstrate that recombinant sources of GH retain their biological activity in fish. The success of exogenous treatment of fish with GH indicates the dominant effect GH has on growth; however, injection or implantation of pellets is not practical for use in large scale aquaculture. Since the production of the first GH transgenic mouse (Palmiter et al., 1982), there have been major efforts in nearly all species of agricultural importance to produce transgenic animals that express GH fusion genes.

The CBA promoter and sGH promoter should differ in their effects. The CBA promoter is not tissue-specific and expression in a variety of tissues may cause unexpected side effects. The ectopic expression of GH due to a constitutive promoter like CBA may result in unusual phenotypes due to altered organ or tissue physiology. The sGH promoter may increase growth through more normal physiological regulation. The sGH promoter was PCR amplified using sequence information from Johansen et al. (1989) and a sGH/hGH construct was generously supplied to us by Dr. Geoffrey Waldbieser. If the promoter does confer pituitary tissue specificity and is normally regulated, these transgenic fish may have more consistent genetic and pleiotropic relationships associated with enhanced growth.

Conservative model. Test of environmental risk should be conservative. That is, the design should be such that risks are accentuated. Because medaka propagate prolifically in the wild and under domestication, and can survive under diverse environmental conditions; i.e., in free flowing, stagnate, and brackish water, if any fish can adapt to a new environment or setting, medaka can. The organism thus provides a more stringent test of environmental risk than most species of commercial interest.

The aquaria and large fish tanks that we are using as habitats in our study also provide a conservative assessment because relatively simple or small aquatic ecosystems are considered to be more vulnerable to de-stabilization by transgenic fishes than relatively complex or large ecosystems (Kapuscinski and Hallerman, 1990). This conclusion is supported by the observation that introduced fish are often more successful in habitats perturbed by human activities than in pristine habitats. However, insights that we obtain, regarding evolutionary adaptation in transgenic medaka in artificial habitats, will be applicable to natural environments because the evolutionary mechanisms involved in selective modification are general mechanisms.

Effect of transgene on fitness. Overall fitness of an organism depends on all factors which impact offspring production. These can be divided into three forms of selection: sexual, fecundity, and viability. Sexual selection is due to differential mating success and depends on an individual's ability to attract mates and copulate successfully. Fecundity selection is due to differential numbers of progeny produced by a female and is influenced by such factors as body size and ovulation rate. Viability selection results from differential survival of individuals and depends on such traits as competitive ability, predator avoidance, foraging ability, energy demands, and age at sexual maturity. It is important to differentiate these modes of selection because they involve different mechanisms. In addition, viability and sexual selection may oppose each other because traits which enhance the ability to attract a mate may also attract predators. Fecundity and viability selection may also oppose each other because energy devoted to offspring production limit energy that can be channeled into growth and maintenance thereby reducing viability. The effects of a transgene on these modes of selection must therefore be evaluated separately. Changes in individual growth rates will probably affect a number of life history traits such as age and size at maturity, maximum body size, longevity, fecundity, mating success, and offspring size. However, the direction of change is uncertain. Larger body size might provide a competitive advantage against conspecifics and heterospecifics in controlling areas that contain a high density of preferred prey. Alternatively, increased body size might reduce foraging efficiency by increasing handling time required for preferred prey. Larger body size could also either reduce predation intensity of normal predator species or increase susceptibility to novel predator species, thus producing changes in food web dynamics in the community (Kapuscinski and Hallerman 1990). Similarly, larger body size might significantly affect aspects of social behavior, in particular mating success. Studies on a variety of fish species have shown size-dependent success in both intra- and inter-sexual interactions (e. g., Beacham, 1988; Bisazza et al., 1989; Collins et al., 1967; Contantz, 1975; Hoelzer, 1990; Rowland, 1989). Alternatively, large males might be rejected as mates by females because their species identity is ambiguous.

Objective 1: Initial screening of the transgenic individuals. The first objective of this research is to evaluate the ecological and social performance of transgenic medaka relative to non-transgenic individuals by measuring relative growth, maximum body size, size and age at sexual maturity, fecundity, fertility, mating success, and predator avoidance ability as well as to determine environmental conditions under which transgenic fish are likely to remain viable or increase in size. However, because of the enormous combinations of possible environmental factors (e.g., water temperature, pH, salinity, oxygen level, prey types, etc.), we limited environmental variables to food availability because food is the most likely limiting factor in supplying energy for growth and reproduction. For what follows, it is assumed that multiple transgenic lines, based on different founders, will be established. Procedures for this objective will be completed twice, once immediately after the transgenic line has been isolated and again after eight generations of natural selection.

Competitive fitness. Because transgenics are not phenotypically distinguishable from others from the same base population of Purdue Orange Red's (POR), competitive ability will be measured relative to a common tester strain that is phenotypically distinct (Guanineless). Performance of the transgenic lines (tr), POA, and Gu will be determined in, 40 l aquaria, maintained at 27^oC, with experimental treatment combinations as shown in Table 1: The low feeding level used supplies a diet that averages 25% above maintenance levels; the high feeding level results in maximum growth rate as determined in a series of preliminary experiments.

The entire experiment will be replicated to provide adequate power to detect differences. For the treatment marked with an asterisk, we will only collect data on total weight at sexual maturity. For all other treatments, we will determine, age and size at sexual maturity (females only), fecundity (females only), and longevity (both sexes). Sexual maturity will be determined by noting the onset of spawning in females. Fecundity will be measured by removing eggs from females within one hour of the time lights come on. Egg production and body weight will also be taken 6 weeks and 9 weeks after sexual maturity. Longevity will be measured as the time to 50% mortality, and will be ascertained by counting all live fish daily and removing any dead fish observed. Body weights, total egg production, and longevity will then be plotted using DeWit diagrams, as shown in Figure 1. This procedure provides a particularly sensitive method of evaluating the relative importance of within and between strain competition (Ricklefs, 1990).

Mating success and fecundity. To determine reproductive differences between adult transgenic and POR individuals, the following fitness components will be measured: fecundity (females only), fertility (both sexes), and mating success (males only). Experimental treatments will contain either 10 transgenic or POR females with 5 POR males and 5 transgenic males. This protocol will allow us to determine how male mating success varies as a function of both male and female genotype. That is, larger transgenic males might be able to outcompete the smaller POR males for access to females; however, females might preferentially mate with POR males. Similarly, experiments will also be conducted using one of the transgenic lines in which the 10 POR or transgenic females are placed in aquaria with a 9:1 ratio of POR:transgenic males (and vice versa). This extension will test for the possibility of a "rare male" effect, a well known phenomenon in which male mating types that occur at low frequency in a population are disproportionately successful in obtaining mates (Ehrman, 1967; Farr, 1977; O'Donald, 1980; Speiss, 1982). Such a test is particularly suited for investigations of the potential fate of accidentally released transgenics in nature because released individuals will occur at a low frequency relative to conspecifics. The experimental design summary is given in Table 2.

This experiment requires individual identification of adults. To do this, a colored thread is inserted through the tail near the base of the caudal fin. The high level feeding regime will be used in this experiment. Average daily mating success of males will be obtained by direct observation of individually marked mating pairs.

Predator avoidance. To determine predator avoidance ability of transgenic lines relative to POR, we will conduct the following experiment: For each treatment combination, a total of 40 medaka and one predator will be placed in a 73 l aquaria. Each treatment will consist of 20 Gu medaka and either 20 transgenics or POR fish. The predator will be a bluegill sunfish (Lepomis macrochirus) of appropriate body size. Each aquaria will be divided in half with plastic netting using a mesh size that permits free movement of medaka but restricts sunfish to one side of the tank; thus, supplying medaka with a predator refuge. The high level feeding regime will be used in all treatments, and food will be distributed uniformly on the predator and refuge side of the tank. Control tanks will set up in a similar fashion except without the predator. Data to be collected include growth rate and longevity.

Objective 2: Estimation of possible modification of the transgene, or its expression, by natural selection. The second objective of this investigation is to determine if natural selection will modify expression of the transgene such that transgenic organisms become a greater environmental risk than initial screening would suggest. Common arguments regarding generic safety of transgenics were reviewed by Regal (1987), some of which were: 1) The organism is a finely tuned product of evolution and any alteration of the genome will harm rather than help the individual or population that is left untended in nature. 2) engineered species will be so burdened with novel combinations of metabolic functions that they will not be able to survive in nature without the assistance of humans. In addition, there may be a selective disadvantage associated with a transgene due to the energetic burden of synthesizing additional macromolecules, such as nucleic acids and proteins (Lenski and Nguyen, 1988). However, those arguments have ignored the possibility that natural selection can modify correlated traits to readjust the physiology of an organism such that the transgene becomes more integrated (or even better integrated) resulting in an organism that is more fit than its wild-type counterparts. There are three ways in which expression of the transgene could be modified by selection: a) modification of genes with pleiotropic and/or epistatic effects which can mitigate effects of (the transgene, b) modification of correlated traits, such as foraging efficiency and fecundity to better adjust the energy balance of the organism, and c) hybridization followed by genetic recombination within the inserted region with resulting modifications of the copy number or expression of the transgene itself. A multi-generation study is essential to examine the possibilities. This aspect of the study will be initiated immediately after pure breeding lines of medaka have been established.

Evolution in a complex competitive environment. Survival of transgenic individuals in a community containing conspecifics should alter gene frequencies of the conspecific population. However, the amount of change in gene frequencies should depend on how transgenics differ from conspecifics in one or more fitness components. Thus, our approach will be to not only measure changes in gene frequency but also to identify ecological mechanisms underlying any such change and the particular fitness components affected.

To determine if modifications of transgene expression will occur in a complex, competitive environment due to changes in correlated traits, where transgenics must compete with wild type medaka for mates, we will create mesocosms in 900 l tanks. The mesocosm will be a simulated natural community that will contain macrophytes, phytoplankton, and zooplankton and two other fish species, one of which will be a predator of medaka. A separate mesocosm will be set up for each transgenic line. An equal number of transgenics and POR individuals will be placed in each mesocosm. After approximately eight generations, the population will be completely inventoried to determine changes in population structure and frequency. The descendants from each of the transgenic lines will be competed against POR medaka individuals that have not experienced "mesocosm selection", using the protocol outlined in Objective 1.

Evolution in a simple noncompetitive pure strain environment. To determine if modifications of transgene expression can occur in a simple non-competitive environment due to changes in correlated traits, we will maintain the frequency of the transgene at 100% for eight generations. Each transgenic line, and the POR control, will be expanded to approximately 100 mating pairs and placed in a 363 l tank. Each tank will have a gravel bottom and numerous plants to serve as substrata for egg deposition. To shorten the generation interval and avoid overlapping generations, all adults will be removed every 12 weeks. The next generation will be produced by eggs that remain in the tank.

After the final generation, all lines will be performance tested for growth, fecundity, fertility, longevity, age and size at maturity, maximum size, mating success, competitive ability, and predator avoidance as detailed in the initial objective. These performance data will be compared to those obtained after the transgene was first isolated. It is expected that natural selection will modify correlated traits to increase fitness of the organism. However, which traits will be modified and to what extent cannot be anticipated.

Evolution in a simple environment evolution through recombination. To determine if selection will modify the gene itself through recombination as a result of hybridization, we will produce hemizygous transgenic fish by crossing transgenic individuals with POR individuals. The crossbreds will be allowed to undergo eight generations of natural selection.

Crossbreds will be produced by reciprocal matings of transgenic males to POR females, and vice versa. Hemizygous offspring from these matings will be placed in a 363 l tank with a gravel bottom and numerous plants. Approximately every 12 weeks, all adults will be removed. Eggs remaining in the tank at that time will produce each subsequent generation. Two hundred fish removed each generation will be tested for presence of the transgene by PCR test. We anticipate that homozygotes will be distinguishable from hemizygotes by using quantitative PCR or slot blot analysis.

Because the transgene will be segregating, and starting at a known gene frequency, data from the elimination (accumulation) rate of the transgene can be used to estimate total lifetime fitness of the transgene, which incorporates all stages in the life cycle including mating success, fecundity, fertility, viability, and competitive ability, through a recursive mathematical model. (See Fitness Parameters)

After the eighth generation, homozygous transgenic medaka will be re-isolated and their progeny performance tested for growth, fecundity, fertility, longevity, age and size at maturity, maximum size, mating success, competitive ability, and predator avoidance as detailed in Objective 1. These performance data will be compared to those obtained in first experiment of Objective 2 to determine if modifications of these traits by hybridization followed by natural selection is different from that due to natural selection without recombination.

It is expected that hybridization prior to natural selection will increase opportunities for selection to mitigate undesirable effects of the transgene and thereby increasing the probability that the transgene will remain in the environment. To verify those conclusions and prediction equations, reisolated transgenic fish will be reintroduced at a gene frequency of 0.5, using the same procedures as outlined at the beginning of this objective. Elimination rate will be monitored for an additional eight generations after the first set of eight generations.

Fitness parameters. Data from elimination rate of the transgene in experiments of the second objective will be fit to a model incorporating differential viability, fecundity-fertility, and mating success, as shown by Muir and Bell (1986) to determine lifetime fitness parameters. If selection modifies fitness, the model can be adjusted to incorporate a changing selection coefficient and thereby determine the rate at which natural selection is itself changing the selection coefficient. This procedure will allow computation of overall reduction in fitness and estimates of how long the expressing transgenic fish will remain in a population. Fitness parameters are as shown in Table 3.

The recurrence equation for the expected frequencies of the genotypes in each generation is developed as follows:

Let p_kt = frequency of kth genotype in the tth generation (i= 1, +/+; 2, +/gH; 3, gH/gH), and

W_t = overall fitness of the populations in the tth generation= _ijk (p_itf_i)(p_jtm_j)v_ijk;

then the expected frequencies of genotypes in the next generation are: p_k(t+1) = _ij (p_itf_i)(p_jtm_j)v_ijk/W_t for k=1,2,3

with expected gene frequency of the transgene

z_(t+1) = p_3(t+1) + p_2(t+1)/2.

Fitness estimates and standard errors are found using generalized weighted least squares, which, for a binomial variable such as gene frequency, are the same as a minimum chi-square estimators, and asymptotically the same as maximum likelihood estimators. For a binomial variable with mean p_t and variance p_t(1-p_t), this result can be seen as follows:

Let, a_t = observed frequency of the transgene in the tth generation,

n_t = sample size in the tth generation,

g = number of generations, and

N = _tn_t;

then X² = _t {[(a_t-z_t)n_t]²/[n_tz_t(1-z_t)]}.

Let be the vector of estimated fitness parameters, then the estimates are distributed as

_~ N(, 1/N(B'B)^-1).

That is, the estimates have a multivariate normal distribution with expected value and variance-covariance of (1/N)(B'B)^-1, where B is a matrix whose elements are

b_jk = [1/E_j()]E_j()/_k.

Since the elements of _k are normally distributed, tests of hypothesis can be computed using the t-statistic. These will be estimated for each experiment in the second objective and again after reisolation to determine if natural selection has changed the fitness parameters.

ACKNOWLEDGEMENTS

We wish to express appreciation to Rob Martens and Mark Smith for their assistance in setting up the experimental systems. This research was supported by USDA grant 93-33120-9468.

Arnold SJ, Wade MJ (1984). On the measurement of natural and sexual selection:application. Evolution 38:720-734.

Beacham JL (1988). The relative importance of body size and aggressive experience as determinants of dominance in pumpkinseed fish, Lepomis gibbosus. Anim. Behav. 36: 621-623.

Bisazza A, Marconato A, Marin G (1989) Male competition and female choice in Padogobius martensi (Pisces, Gobidae). Anim. Behav. 38: 406.

Boyd RD, Bauman DE (1988) Mechanisms of action for somatotropin in growth. In: Campion DR, Hausman GJ, Martin RJ (eds) Current concepts of Animal growth regulation. Plenum Publishing Co. New York.

Chourrout D, Guyomard R, Houdebine LM (1986) High efficiency gene transfer in rainbow trout (Salmo gairdneri) by microinjection into egg cytoplasm. Aquaculture 51,143-150.

Collins JM, Cramer J, Wecker S (1967) Nip frequency and dominance in male swordtails and platies. Amer. Zool. 7: 806.

Contantz GD (1975) Behavioral ecology of mating in the male Gila topminnow, Poeciliopsis occidentalis (Cyprinodontiformes: Poecilidae). Ecology 56: 966-973.

Donaldson EM, Fagerland UHM, Higgs HA, McBride JR (1979) Hormonal Enhancement of Growth. In: Hoar WS, Randall DJ, Brett JR (eds) Fish Physiology Vol. VIII. Bioenergetics and Growth. Academic Press New York. p. 455.

Dunham RA, Eash J, Askins J, Townes TM (1987) Transfer of metallothionein-human growth hormone fusion gene into channel catfish. Trans. Amer. Fish Soc. 116,87-91

Ehrman L (1967) Further studies on genotype frequency and mating success in Drosophila. Am. Nat. 101: 415-424.

Farr JA (1977) Male rarity or novelty, female choice behavior, and sexual selection in the guppy, Poecilia reticulata Peters (Pisces: Poeciliidae). Evolution 31: 162-168.

Gall GAE, Gross SJ (1978) Genetic studies of growth in domesticated rainbow trout. Aquaculture 13,225-234.

Gill JA, Sumpter JP, Donaldson EM, Dye HM, Souza L, Berg T, Wypych J, and Langley K (1985) Recombinant chicken and bovine growth hormones accelerate growth in aquacultured juvenile pacific salmon Onchoryncus kisutch. Bio/Technology 3,643-646.

Guyomard R, Chourrout D, Houdebine LM (1990) Gene transfer by microinjection into fertilized trout eggs: efficient integration and germline transmission. Aquaculture 85.

Hallerman EM, Kapuscinski A (1990). Transgenic fish and public policy: regulatory concerns. Fisheries 15:12-20

Hirshfield M (1980) An experimental analysis of reproductive effort and cost in the Japanese medaka Oryzias latipes. Ecology 61:282-292.

Hoelzer GA (1990) Male-male competition and female choice in the Cortez damselfish, Stegastes rectifraenum. Anim. Behav. 40:339-349

Huang N, Gall G (1990) Correlation of body weight and reproductive characteristics in rainbow trout. Aquaculture 86:191-200.

Johansen B, Johnson OC, Valla S (1989) The complete nucleotide sequence of the growth-hormone gene from Atlantic salmon (Salmo salar). Gene 77, 317-324

Kapuscinski AR, Hallerman EM (1990) Transgenic fish and public policy: anticipating environmental impacts of transgenic fish. Fisheries 15:2-11.

Kapuscinski AR, Jacobson LD (1987) Genetic guidelines for fisheries management. Sea Grant Research report 17. Minnesota Sea Grant, St. Paul MN.

Lenski RD, Nguyen TT (1988) Stability of recombinant DNA and its effects on fitness. TIBTECH 6:s18-s20.

Levin MA, Seidler R, Borquin AW, Fowle III JR, Barkay T (1987) EPA developing methods to assess environmental release. Bio-Technology (NY) 5:38-41

Lewin B (1994) Genes V, Oxford University Press

Lu JK, Andrisani OA, Dixon JE, Chrisman CL (1989) Integration and germ-line transmission of human growth hormone gene via microinjection into early medaka (Oryzias latipes), embryos. Second Symposium on Genetic Engineering of Animals. Cornell University, June 25-28, 1989.

Maclean N, Penman D, Zhu Z (1987) Introduction of novel genes into fish. Bio/Technology 5, 257-261.

Magnuson J (1962) An analysis of aggressive behavior, growth, and competition for food and space in medaka (Oryzias latipes (Pisces, Chprinodontidae). Can. J. Zoology. 40:313-363.

McEvoy T, Stack M, Keane B, Barry T, Sreenan J, Gannon F (1988) The expression of a foreign gene in salmon embryos. Aquaculture 68, 27-37.

Muir WM, Bell AE (1987) Multiple vital functions of the daughterless (da) gene in Drosophila melanogaster and factors influencing its expression. Genetica 72:43-54.

O'Donald P (1980) Genetic models of sexual selection. Cambridge Univ. Press.

Penman DJ, Beeching AJ, Penn S, MacLean N (1990) Factors affecting the integration of microinjected DNA into the rainbow trout genome. Aquaculture 85.

Penman DJ, Iyengar A, Beeching AJ, Rahman A, Sulaiman Z, Maclean N (1991). Patterns of transgene inheritance in rainbow trout (Onchoryncus mykiss). Mol. Reprod. Devel. 30, 201-206.

Regal PJ (1987) Models of genetically engineered organisms and their ecological impact. Recomb. DNA Tech. Bull. 10:67-85.

Regal PL (1988) The adaptive potential of genetically engineered organisms in nature. TIBTECH 6:s36-s38.

Ricklefs RE (1990) Ecology, Third Ed. W.H. Freeman.

Rokkones E, Alestrom P, Skjervold H, Gautvik KM (1989) Microinjection and expression of a mouse metallothionein human growth hormone fusion gene in fertilized salmonid eggs. J. Comp. Physiol. B158:751-758.

Rowland WJ (1989) The effects of body size, aggression and nuptial coloration on competition for territories in male threespined sticklebacks, Gasterosterus aculeatus. Anim. Behav. 37: 282-289.

Schneider JF, Hallerman EM, Yoon SJ, He L, Gross ML, Liu Z, Faras AJ, Hackett PB, Kapuscinski AR, Guise KS (1989) Transfer of the bovine growth hormone gene into northern pike, Esox lucius. J. Cell. Biochem. 13B:173.

Schneider JF, Hallerman EM, Yoon SJ, He L, Myster SH, Gross M, Liu Z, Zhu Z, Hackett PB, Guise KS, Kapuscinski AR, Faras AJ (1989) Microinjection and successful transfer of the bovine growth hormone gene into the northern pike, Esox lucius. Abstr. J. Cell Biochem. Supp 13B p. 173.

Singh H, Griffith RW, Takahashi A, Kawauchi H, Thomas P, Stegeman JJ (1988) Regulation of gonadal steroidogenesis in Fundulus heteroclitus by recombinant salmon growth hormone and purified salmon prolactin. Gen. Comp. Endocrinol. 72,144-153.

Speiss EB (1982) Do females choose their mates? Am. Nat. 119: 675-693.

Stuart GW, McMurray JV, Westerfield M (1988) Replication, integration and stable germ-line transmission of foreign sequences injected into early zebrafish embryos. Development 103, 403-412.

Tiedje JM, Colwel RK, Grossman YL, Hodson RE, Lenski RE, Mack RN, Regal PJ (1989) The planned introduction of genetically engineered organisms: ecological considerations and recommendations. Ecology 70:298-315.

Weatherley AH, Gill HS (1987) Influence of Hormones. In "The Biology of Fish Growth" Academic Press New York p.177

Yamamoto T (1975) Medaka (killifish):biology and strains. Keigaku Publishing, Yokya, Japan.

Yenikolopov GN, Benyumov AO (1989) Enhanced growth of transgenic fish containing somatotropin genes. Abstr. J. Cell Biochem. 13B. p. 175

Zhang P, Hayat M, Joyce C, Gonzalez-Villasenor LI, Lin CM, Dunham RA, Chen TT, Powers DA (1990) Gene transfer, expression and inheritance of pRSV-rainbow trout GH cDNA in the common carp (Cyprinus carpio). Mol. Reprod. Devel. 25, 3-13.

Zhang P, Lin CM, Hayat M, Joyce C, Gonzalez-Villasenor LI, Dunham RA, Chen TT, and Powers DA (1989) Gene transfer, expression and inheritance of pRSV-Trout-GH-cDNA in fish. Abstr. J. Cell Biochem. Supp 13B p. 166

Muir Table 1

Muir Table 2

Muir Table 3

Muir Figure 1

1. Correspondence: W.M. Muir, Dept. ANSC, Purdue University, W. Lafayette, IN 47907, Phone 317-494-8032, FAX 317-494-9346, E-mailbmuir@hub.ansc.purdue.edu