© R.A. Dunham 2004. Aquaculture and Fisheries Biotechnology: Genetic Approaches
104 (R.A. Dunham)
Genetic Variation, Population Structure and Biodiversity
Genetics is in reality a relatively new field.
For centuries, humans moved fish among countries and stocked conspecifics in new watersheds without any knowledge or con-cern regarding genetic principles, impact or consequences. During the first 70 years of the 20th century, fish movement and stocking were rampant. Beginning in the 1970s, con-servation genetics has become recognized and is a burgeoning issue, as well as biodi-versity and genetic biodibiodi-versity. In general, individual countries and natural-resource agencies now take a much more conservative approach to stocking programmes, genetic conservation and biodiversity. However, many decisions and policies are made and implemented without data on population genetics and the genetic interactions of fish populations. There is a need for much more research in this area.
When data are available, lack of accurate stocking histories complicates data interpreta-tion. Another void is a lack of data demon-strating the relationship between performance and biochemical and molecular markers, which was introduced in the previous chapter.
These are very difficult data to generate, as it is not easy to replicate the natural environ-ment, rivers, reservoirs, lakes and oceans in a realistic manner. Geneticists sometimes give natural-resource managers conflicting advice regarding the desirability of increased or decreased genetic variation and the policies and mechanisms to achieve various goals.
The first question that needs to be consid-ered is the importance of genetic variation in natural populations. Is it better to have more genetic variation or less genetic variation?
These are difficult questions to answer, and the answer may be different depending upon the individual circumstances. Do population structures dictate the need for the quantity and type of genetic variation? In the circum-stances leading to different population struc-tures, have the selective pressures led to the optimum genotypes in a particular environ-ment or have limitations on gene flow in that environment limited the development of the optimum genetic structure of a population?
Theoretically, genetic variation is benefi-cial and important. Genetic variation is important for the long-term survival of a species. Genetic variation can ensure the fitness of a species or population by giving the species or population the ability to adapt to changing environments.
Obviously, a lack of genetic variation or too much homozygosity can be detrimental to an individual’s or a population’s survival traits and fitness. The cheetah is a prime example of the potential detrimental effects of excess homozygosity. This highly homozygous species has severe reproductive problems. Homozygosity has also been cor-related with bilateral asymmetry (fluctuating asymmetry) – unbalanced meristic counts on the right and left halves of the body – in fishes. Additionally, highly or totally homozygous individuals and populations actually exhibit greater phenotypic variation than outbred controls because they are more
greatly affected by environmental or microenvironmental change and have reduced homoeostatic ability compared with more heterozygous individuals and tions. Inbreeding in small, natural popula-tions increases extinction rate (Doyle, 2003).
Inbreeding depression resulting from increased homozygosity is well documented in fish (Dunham et al., 2001). Field crops have been endangered when they did not have the genetic variation to respond to new pathogens or plagues. Clearly, the existence of genetic variation is important to the long-term survival and fitness of a species. Many natural populations respond to different forms of selection, such as directional, bidirec-tional, cyclical and stabilizing selection, which help to ensure the maintenance of the genetic variability and/or fitness of populations.
Levels of homozygosity and inbreeding can be important not only in domestic or aquaculture populations, but in wild popula-tions as well. Inbreeding does adversely affect reproductive success in wild deer (Slate et al., 2000). Microsatellite heterozygos-ity was utilized as an indicator of individual inbreeding coefficients among unmanaged deer on the island of Rhum, Scotland.
Heterozygosity was correlated with lifetime breeding success (total offspring) in both males and females.
The majority of inbreeding experiments on fish (Dunham et al., 2001) and other organisms have been done in aquaculture and laboratory-type environments. Some have hypothesized that inbreeding depres-sion would be more severe and affect fitness more adversely in the harsher natural ronment compared with the laboratory envi-ronment or aquaculture envienvi-ronment where animals are well taken care of. However, the fitness of mosquito populations declined to the same extent in natural tree holes as under favourable laboratory conditions for mosquitoes (Armbruster et al., 2000).
Depending upon population structure, inbreeding can be prevented in natural pop-ulations via migration. However, migration rates may need to be larger than previously expected to prevent inbreeding. A simulation by Vucetich and Waite (2001) indicates that in real populations the number of
immi-grants needed to prevent inbreeding is actu-ally much greater than one individual per generation, which is the theoretical require-ment in idealized Fisher–Wright populations (Doyle, 2003). In random-mating popula-tions, where reproductive variance follows a Poisson distribution, one immigrant per gen-eration will theoretically prevent inbreeding if the numerical population size is larger than about 20. However, variation in mating success caused by spawning frequency, fecundity and mortality differences in real populations increases reproductive variance and causes the effective population size, Ne, to be considerably less than the numerical (census) size. The reproductive variance of immigrants is highly variable, as well exacer-bating the problem. Therefore, more than one immigrant individual is needed to pre-vent inbreeding and the situation becomes worse the greater the discrepancy between actual and effective population sizes. The required number of immigrants increases with the census size of population, which is not the case in idealized, theoretical popula-tions, in which the census and effective sizes are equal (Doyle, 2003).
Migration can counteract the deleterious effects of inbreeding. Experimental popula-tions of mustard, Brassica campestris, were maintained at a census population size of N = 5 for five generations, with three levels of migration (0, 20 and 50%) per generation, and the result was that several measures of fitness were lower in populations that had experienced no in-migration (Newman and Tallmon, 2001). Similarly, in a natural popu-lation of warbler on an island off the west coast of Canada, 98% of the population died in the winter of 1989, with the resulting inbred birds suffering higher mortalities (Keller et al., 2001). All measures of genetic diversity dropped dramatically, and then quickly recovered to pre-crash values due to the immigration of only one animal per year.
This immigration made the bottleneck unde-tectable. The increased mortality of inbred individuals during the crash also eliminated many deleterious recessive alleles.
However, immigration can also have nega-tive effects. A single immigrant warbler imme-diately introduced deleterious alleles back into
Population Genetics 105
the population (Doyle, 2003) and, because inbreeding was high, the overall level of inbreeding and inbreeding depression (Keller et al., 2001) rapidly increased to their high pre-crash levels. The purging of deleterious alleles in the bottleneck by one set of immigrations was counteracted by immigration from a sin-gle individual. Wright called this phenomenon immigration load (Doyle, 2003).
Data from houseflies support this popula-tion phenomenon of recovery from bottle-necks and purging of deleterious alleles.
Housefly lines were inbred either rapidly and severely followed by population expan-sion, or by chronic low population size over a long period (Reed, 2001), both resulting in the same inbreeding. As expected, inbred populations have consistently lower fitness than outbred populations across a range of environments. However, the bottlenecked populations had lower inbreeding depres-sion for a given level of inbreeding in all environments than populations kept at a constant small size. Populations initiated from a small number of founders are able to recover fitness and survive novel environ-mental challenges, provided that habitat is available for rapid population growth (Doyle, 2003).
Doyle (2003) also explores how a bottle-neck – a brief period of very low numbers – affects the ability of a population to meet new selective challenges and evolve new adapta-tions. In the most extreme case, with one pair of breeders passing through the bottleneck, only a maximum of four alleles per locus will be available for population regrowth and evolutionary adaptation, and some of those alleles may be identical. Possibly the effects of individual genes simply add together, genetic variance will decrease when genes are lost by drift and the rate at which the population can evolve will decrease because of the bottleneck. Generally, only additive genetic variance allows an evolutionary response to selection. Theoretically, when some alleles are lost by drift from a non-addi-tive genetic system, the remaining genes will sometimes make a contribution to additive genetic variance (Doyle, 2003), and in this case the capacity to respond to selection will be enhanced by the bottleneck.
Experiments with butterflies support this hypothesis that additive variance will increase in traits closely related to fitness, such as fecundity and survival. However, quantitative traits that are under weak selec-tion pressure will lose additive genetic vari-ance in the bottleneck, as do neutral marker alleles (Doyle, 2003). The traits related to fit-ness have an especially high proportion of dominance variance, complicating the inter-pretation of the data related to the theories presented here.
Additive genetic variance of wing size (supposedly a neutral fitness trait) decreased during bottlenecks, while additive variance and heritability of egg-hatching rate – a fit-ness component – increased (Saccheri et al., 2001). The change of wing-size variance, but not hatching rate, followed random-sampling expectation exactly. Dominant genes can be selected for and therefore, especially in early generations of experiments, this response could increase the narrow-sense heritability as reflected by the increase response to selec-tion from the dominant alleles.
Although genetic variation is usually con-sidered desirable, cases may exist where a lack of genetic variation may enhance an organism’s short-term fitness when a popu-lation is highly adapted for a particular envi-ronment. Theoretically, the introduction of inferior genotypes could reduce a popula-tion’s fitness, and some conservation geneti-cists have coined the term outbreeding depression for this population phenomenon, although it is not well documented.
Outbreeding depression is usually related to temporary relaxation of selection pressure.
For some critical developmental events and biochemical pathways, canalization and epistasis negate potentially detrimental genetic variation.
Organisms near the periphery of a recent range or habitat expansion are often geneti-cally less variable than those at the heart of the geographical range (Doyle, 2003). For example, mitochondrial DNA (mtDNA) variation is low in the current northernmost populations of marine snails, which have recently been recolo-nized from glacial refugia located further south (Hellberg et al., 2001). These homozygous pop-ulations at the edges of geographical ranges or
106 Chapter 7
in suboptimum habitats may be a result of founder effects and drift or they may be a result of selection. They could be strengthened, weakened or not affected by the introduction of new genetic variation, but this is little studied and needs to be addressed.
Various forms of stabilizing selection may lead to wild homozygous lines, and out-breeding depression may be a natural phe-nomenon required for the long-term fitness of a population or a natural product of a par-ticular genetic structure. In some cases, selec-tion acts to develop lines that then mate to produce fit offspring, although these off-spring do not have hybrid vigour in terms of ecological or reproductive fitness. Landry et al. (2001) found that wild Atlantic salmon chose mates with genotypes different from their own, maximizing the heterozygosity of offspring at the major histocompatibility complex (MHC). Microsatellite allele and MHC data indicated that enhancing the diversity of the peptide-binding region of the MHC appeared to be the mating objective, not solely the avoidance of inbreeding. Such an apparent genetic structure and process are counter to recent information that indi-cated that salmon mating was highly ran-dom (Doyle, 2003). This stabilizing selection is a major influence on MHC gene-frequency distributions only at the local population level, such as within rivers, but over larger geographical distances migration and ran-dom drift were the ran-dominant evolutionary process at the MHC locus, as indicated by the similar geographical pattern of MHC allele frequencies and neutral microsatellite variation (Landry and Bernatchez, 2001).
This same balancing selection exists at the MHC in the endangered chinook salmon of the Sacramento river, and has apparently maintained MHC diversity for millions of years in these fish and continues to counteract potential random loss of diversity via genetic drift caused by the recent, local population bottlenecks (Garrigan and Hedrick, 2001).
Mounting evidence indicates that this balancing selection to maintain high levels of genetic diversity via overdominance for disease resistance is common in salmon and has been documented in 31 popula-tions of a third species, sockeye salmon
(Miller et al., 2001). Again, balancing selec-tion took place locally, within sockeye salmon populations. However, directional selection also occurred at the MHC in sev-eral of the sockeye populations, illustrating that different forms of selection can be prevalent in different populations.
In the case of Atlantic salmon, genetic dis-tances between populations as measured at the MHC locus correlated well with genetic distances measured at neutral microsatellite loci and also with geographical distance (Landry and Bernatchez, 2001). Additionally, divergence of the Atlantic salmon popula-tions was essentially a random process.
Conversely, the apparent heterogeneity in selection at MHC loci in sockeye salmon resulted in strong genetic differentiation between geographically proximate popula-tions with and without detectable levels of balancing selection, in stark contrast to observations at neutral loci (Miller et al., 2001; Doyle, 2003). Miller et al. (2001) con-clude that, based on the distribution of MHC class II diversity throughout the Fraser drainage, conservation of sockeye salmon must be conducted on the basis of individual lake systems.
Fontaine and Dodson (1999) established the relatedness of salmon fry (in their first summer of life) and parr (in their second and third summers of life) captured in adjacent territories by examining microsatellites, and found that fish collected near each other were not full sibs, which possibly has impli-cations on how to collect brood stock for genetic conservation, assuming that a similar distribution of individuals is found for adults. However, the distribution of individ-uals appears to be different for other species.
Pouyaud et al. (1999) found that in mouth-brooding black-chin tilapia, Sarotherodon melanotheron, related individuals tended to aggregate in open water environments and that mating occurred preferentially within small groups of kin, based on heterozygote deficiencies and similarity indices at four microsatellite loci. However, this inbreeding did not take place in riverine populations. If similar breeding structures exist in other tilapias, inbreeding may be higher in aqua-culture populations than expected.
Population Genetics 107
Various population structures exist, such as panmixia, sympatry, disjunct and step-ping-stone (May and Krueger, 1990). When all of the fish constitute a single reproductive unit, panmixia exists and mating is random.
Whether a panmictic (a single stock) exists or several discrete, non-interbreeding stocks dictates the genetic management strategy.
The conclusion that individuals within a body of water represent one panmictic popu-lation can be considerably strengthened if a study uses genetic data in combination with fish movement data relative to spawning areas and observations of reproductive behaviour (May and Krueger, 1990). Based on allozyme data, most panmictic popula-tions are marine species, such as the milk-fish, Chanos chanos (Winans, 1980), or the southern African anchovy, Engraulis capensis (Grant, 1985). Low levels or lack of genetic differentiation may be observed in fresh-water species such as northern pike, Esox lucius (Seeb et al., 1987), even though geo-graphical isolation through lake and drainage boundaries prevent panmixia.
Populations that are genetically differenti-ated but apparently have free access to spawn with each other because they live in the same body of water are sympatric.
Reproductive isolation among sympatric populations is not due to geographical boundaries, such as waterfalls or lake shore-lines, but instead is due to processes such as olfactory homing to natal areas, assortative mating, behavioural selection of different spawning substrates or physiologically based differences in the timing of spawning (May and Krueger, 1990). Populations of Pacific salmon, such as the coho salmon, mix during part of their life cycle and then subse-quently assort to natal waters prior to spawning, representing one form of sympa-try (Wehrhahn and Powell, 1987). Temporal rather than spatial reproductive isolation, such as is the case with pink salmon, Oncorhynchus gorbuscha, which have a strict 2-year life cycle, resulting in genetic differen-tiation between odd- and even-year popula-tions that use the same spawning stream (Aspinwall, 1974; Beacham et al., 1985), rep-resents another form of sympatry. Lake Superior is large enough to have discrete
spawning areas for lake trout, and geneti-cally differentiated populations appear to occupy the same body of water for their entire life cycle (Dehring et al., 1981; Goodier, 1981; Krueger et al., 1989).
Genetically differentiated sympatric pop-ulations may exhibit differences in quantita-tive life-history traits that are important to fishery management (May and Krueger, 1990). Brown trout individuals sampled from a single body of water often demon-strate genetic differentiation between life-history categories, suggesting that the traits were population-specific. Morphologically dissimilar brown trout in Lake Brunnersjoma, Sweden, have fixed gene dif-ferences, indicating two isolated sympatric populations (Allendorf et al., 1976; Ryman et al., 1979) and sympatric populations of brown trout have different feeding habits (Ferguson and Mason, 1981) and migration traits (Krieg and Guyomard, 1985; Krueger and May, 1987).
Discrete or disjunct populations, such as those that live in separate ponds or lakes with no outlet or in headwater streams with inaccessible barriers to upstream migration, have no possibility of reproduc-tive contact between them, and these iso-lated breeding units tend to diverge genetically with time (May and Krueger, 1990). The extent of differentiation of these populations from their nearest neighbour will be directly proportional to the time of their separation and will be influenced by effective population size, selection, mating pattern, migration and mutation rates.
Allozyme differentiation for disjunct popu-lations has been measured in popupopu-lations of largemouth bass (Philipp et al., 1981, 1983a, 1985; Norgren et al., 1986), bluegill, Lepomis macrochirus (Felley and Avise, 1980), and brook trout (Stoneking et al., 1981; Dunham et al., 2002e).
Stepping-stone is a population structure where localized breeding populations are adjacent to one another such as in many trib-utary streams of a major river system, and the populations maintain reproductive isola-tion by homing to their hatching locaisola-tion but occasionally stray to neighbouring streams, leading to gene flow among populations
108 Chapter 7
(May and Krueger, 1990). Genetic differentia-tion among populadifferentia-tions is then directly pro-portional to geographical or stream distance and the intensity or frequency of homing to natal areas. This population structure will often lead to genetic similarity among stocks within a region or area and increasing levels of differentiation as geographical distances increase between sources, as observed in Alaskan chinook salmon (Gharrett et al., 1987) and sea lamprey, Petromyzon marinus, in the Great Lakes (Krueger and Spangler, 1981; Wright et al., 1985).
Continuous geographical changes in allele frequency within a population are
Continuous geographical changes in allele frequency within a population are