The great diversity of sex determination mechanisms in animals and plants ranges from genetic sex determination (GSD, e.g. mammals, birds, and most dioecious plants) to environmental sex determination (ESD, e.g. many reptiles) and includes a mixture of both, for example when an individual’s genetically determined sex is environmentally reversed during ontogeny (ESR, environmental sex reversal, e.g. many fish and amphibia). ESD and ESR can lead to widely varying and unstable population sex ratios. Populations exposed to conditions such as endocrine-active substances or temperature shifts may decline over time due to skewed sex ratios, a scenario that may become increasingly relevant with greater anthropogenic interference on watercourses. Continuous exposure of populations to factors causing ESR could lead to the extinction of genetic sex factors and may render a population dependent on the environmental factors that induce the sex change. However, ESR also presents opportunities for population management, especially if the Y or W chromosome is not, or not severely, degenerated. This seems to be the case in many amphibians and fish. Population growth or decline in such species can potentially be controlled through the introduction of so-called Trojan sex genes carriers, individuals that possess sex chromosomes or genes opposite from what their phenotype predicts. Here, we review the conditions for ESR, its prevalence in natural populations, the resulting physiological and reproductive consequences, and how these may become instrumental for population management.
Sex determination is the process that directs differentiation down the male or female pathway of ontogeny. Sex differentiation is the development of testes and ovaries from undifferentiated gonads. Organisms either possess sex-determining genes on sex chromosomes or on autosomes (from which their sexual differentiation can be predicted at conception), or they lack any consistent heritable genetic difference between the sexes and their sexual identity and fate is established only some time after fertilization by one or several environmental factors. Some organisms are under the influence of both genotypic and environmental effects, and their phenotypic sex is determined by the interaction of these two factors (Bull 1983; Valenzuela et al. 2003; Sarre et al. 2004; Baroiller et al. 2009). The diversity of existing sex determination mechanisms thus reflects a continuum ranging from being strictly genotypic to being fully under the control of environmental effects with intermediate forms between the two extremes (Bull 1983; Valenzuela et al. 2003; Wedekind & Stelkens 2010).
Because many plants and animals have their sex influenced by environmental conditions, environmental variation in time and space can lead to biases in population sex ratios. Indeed, skewed brood or family sex ratios are often reported (Werren 1987; Antolin 1993; Palmer 2000; Wedekind 2002; Dyson & Hurst 2004; Brykov et al. 2008). Some of these cases can be explained by sex allocation theory, e.g. when the fitness of male and female offspring varies with the environment (Charnov 1982) or with parental condition (Trivers & Willard 1973). Other cases may not be beneficial to parental fitness. Because sex ratio is a demographic parameter that is crucial for population viability, there is an obvious need for a better understanding of how the environment can influence sex ratio and how it interacts with genotypic sex determination.
This review has several goals. First, it gives an overview of sex determination mechanisms and how mismatches between genetic and phenotypic sex can arise. Second, it summarizes the causes of unbalanced sex ratios with regard to environmental sex reversal (ESR). Third, it describes how our knowledge of sex determination, sex allocation, and sex reversal can be instrumental for the conservation and management of wild and captive populations. Fourth, it outlines the potential long-term genetic effects of ESR. Finally, it summarizes which data and analyses are necessary to better understand the significance and potential of ESR for population management. We put an emphasis on reviewing the conditions and population consequences of sex ratio adjustment and sex reversal in teleost fish because all reproduction types relevant to the present context exist in this taxon.
Sex determination, evolutionary transitions, and mismatches between sex genotype and phenotype
Genetic factors influencing sex
If one plans on using the naturally existing spectrum of sex determination mechanisms for management purposes, such as controlling population growth and viability using ESR or other forms of sex ratio management, it is essential to understand the underlying factors causing organisms to differentiate into males or females. In the following four subsections (2.1–2.4), we give a short summary of what is known about the different factors that determine sex.
The inheritance of sex can be based on mono- or polyfactorial genetic effects. In many vertebrates, sex is determined by major sex factors located on sex chromosomes that either differ morphologically from each other (heteromorphic) or not (homomorphic) (Bull 1983). The XX/XY sex chromosome system with male heterogamety is the major sex factor inherent to most mammals (except for some voles and spiny rats; Just et al. 1995; Soullier et al. 1998) with a gene on the Y chromosome (SRY/Sry) acting as the master switch determining maleness (Goodfellow & Lovell-Badge 1993). Birds possess a ZW/ZZ system with females being the heterogametic sex and a Z-linked DMRT1 gene that seems to be required for male sex determination (Smith et al. 2009). In amphibians, reptiles, and fish, the XX/XY system, the ZW/ZZ system, and a variety of other sex determination mechanisms exist (Devlin & Nagahama 2002). Primary sex determining genes (equivalent of Sry in mammals) have been identified in a few fish species (Schartl 2004). In fact, DMRT1 has been recognized as a conserved module in the male sex determination pathway of some heterogametic fish including the rainbow trout (Oncorhynchus mykiss;Marchand et al. 2000), two species of the genus Oryzias (Matsuda et al. 2002; Kondo et al. 2003) and two hermaphroditic fish species (Huang et al. 2002; He et al. 2003). Duplicate copies of DMRT1 may also assume the role of an apical initiator in sex determination in some species, where DMY (a recent duplicate of the DMRT1 gene; Lutfalla et al. 2003) is thought to be the starting switch in male sex determination, e.g. in Oryzias latipes (Nanda et al. 2003; Matsuda 2005). Other mechanisms for genetic sex determination are ‘sex modifiers’ or ‘sex ratio distorters’, i.e. alleles often located on autosomal loci that can influence or override the master gene (Kallman 1984; Nanda et al. 2003; Horth 2006). Some consider that the feminizing effects of the bacterial insect parasite Wolbachia fall into the category of these selfish genetic elements (Rigaud 1997; Hurst & Jiggins 2000; Werren et al. 2008; Mercot & Poinsot 2009).
In polyfactorial systems sex is determined by the combination of several genes distributed throughout the genome each with minor additive and/or epistatic effects. Here, the final sex of the organism depends on whether the sum of the effects at all sex loci surpasses a threshold value (e.g. as shown in swordtails (Xiphophorus helleri); reviewed in Volff & Schartl 2001). Polyfactorial systems were thought to be rare and they were long regarded as a transient state or intermediate step in the evolution of genotypic sex determination (Rice 1986). However, the sex of several fish species/populations has recently been shown to be under the control of many genes of minor effect [e.g. poecilid fish (Volff & Schartl 2001), sea bass (Vandeputte et al. 2007), and tilapia (Baroiller et al. 2008)]. One of the best studied cases is the ‘dosage compensation’ system in Drosophila, where sex depends on the ratio between female-determining factors on the X chromosome and male-determining factors on the autosomes (X:A system; reviewed in Marin et al. 2000).
Polymorphisms for sex-determining genes within or among populations have been reported in many fish taxa (Conover & Heins 1987; Baroiller et al. 1995; Shine et al. 2002; Davidson et al. 2009) and reverse systems of heterogamety can be found among closely related species (e.g. in Oryzias;Takehana et al. 2007). The tilapia model is a particularly good example of the evolutionary lability of GSD systems. Baroiller et al. (2008) suggested a gradation of these, with at one extreme XX/XY or ZW/ZZ systems and the absence of modifiers, intermediate systems with minor to moderate effects of autosomal modifiers, and at the other extreme an increasing number of loci with increasing impact representing polygenic sex determination systems. Comparative research between tilapiine species has revealed that major male determining loci can be located on different linkage groups in different species (e.g. in both Oreochromis karongae and Tilapia mariae the sex determining locus is on linkage group 3, whereas in O. niloticus and T. zillii it is found on linkage group 1; Cnaani et al. 2008).
Penman and Pifferer (2008) predict that genomics techniques will soon lead to a better understanding of the evolution of sex determination in fish. For species with mono-factorial sex determination, single sex-specific markers will become available while for species with more complex systems (polygenic sex determination or GSD systems susceptible to ESR) genome-wide approaches, such as QTL linkage mapping, may bring the most benefit.
Environmental factors influencing sex
Species with true environmental sex determination (ESD) do not possess a primary sex fixed at conception, and the first ontogenetic difference between sexes is environmentally induced. For example, shift incubation experiments in fish and reptiles have revealed that temperature can affect sex determination during a specific window of time during embryonic development (the thermosensitive period; Conover & Kynard 1981; Valenzuela & Lance 2004). ESD is widespread in reptiles (Janzen & Paukstis 1991; Valenzuela & Lance 2004) and it was assumed that ESD and especially temperature-dependent sex determination (TSD) was also common in fish. However, Ospina-Álvarez and Pifferer (2008) recently suggested that for any given species to have true ESD, it should fulfil both of two conditions: (i) not having sex-determining chromosomes, and (ii) have sex ratio response to an environmental factor that is within the range normally experienced by the organism. Applying these exclusion criteria resulted in the assignment of species from only four teleost orders to have true TSD suggesting that under ecologically relevant conditions, temperature plays less of a role in sex determination in fish than previously thought.
In fish species with TSD, three types of reaction norms have been proposed (Baroiller et al. 1999; Devlin & Nagahama 2002): (i) the frequency of males decreases with temperature, (ii) the frequency of males increases with temperature, and (iii) the frequency of males is higher at extreme (high and low) temperatures. Ospina-Álvarez & Piferrer’s (2008) analysis suggests that sex ratio response to increasing temperatures typically results in more male-biased sex ratios, regardless of whether a species has TSD or GSD with environmental influences.
It should be noted here that even in species with true ESD, selection is ultimately acting on genes that create a link between offspring sex and a certain environmental factor (e.g. incubation temperature). A recent study on a classic model organism for investigating TDS, the jacky dragon (Amphibolurus muricatus), has lead to a revision of the earlier paradigm that genes do not contribute to variation in sex ratio in species with TSD (Warner et al. 2008). The results demonstrate a significant genetic contribution to variation in offspring phenotypes, including sex ratio. These findings challenge the traditional view that sex determination is not influenced by genetic factors in species with ESD (Valenzuela et al. 2003) and demonstrate that sex is more likely the outcome of interactions among maternal, hormonal, genetic, and environmental factors. Similar conclusions are emerging from studies on a diverse range of organisms including lizards (Radder & Shine 2007; Warner et al. 2007), turtles (Freedberg et al. 2006), fish (Conover & Heins 1987; Baroiller et al. 1995; Magerhans et al. 2009), and invertebrates (Kozielska et al. 2006).
Environmental sex reversal
Consistent with the above, Ospina-Álvarez & Pifferer (2008) found evidence that many cases of observed sex ratio shifts in response to temperature reveal thermal alterations of an otherwise predominately GSD mechanism rather than the presence of true TSD. This indicates that environmental factors can directly or indirectly override sex chromosomes or sex genes (Baroiller et al. 2009). In such systems, the primary sex can be modified during ontogeny such that individuals develop into the sex opposite to that coded by their genotype (Bull 1983; Valenzuela et al. 2003). This process has been termed environmental sex reversal (ESR). In a narrow sense, ESR can only occur in species possessing a primary (genotypic) sex. Once a GSD individual has been sex-reversed, its phenotypic sex typically remains fixed for the rest of its life. If we allow for a broader definition of ESR, sex reversal can also occur in ESD species if the sexual identity that an individual first adopts (due to the environmental conditions prevailing at egg, larval or juvenile stage) later becomes reversed by new or changing environmental conditions. Here we focus on ESR in gonochoristic species with GSD, where sex reversal is not known as a typical life history strategy (as is the case in sequential hermaphrodites) but can occur as a response to fluctuating environmental conditions.
There are many different triggers for sex reversal. Most of them are abiotic (e.g. temperature, pH, endocrine-disrupting hormones, photoperiod, hypoxia) but a few biotic factors are also known to induce sex reversal (crowding, pathogens like Wolbachia, population size). Many reptiles, amphibians, fish, and insects are capable of undergoing sex reversal at early life stages (Vance 1996; Wallace et al. 1999; Devlin & Nagahama 2002; Sarre et al. 2004; Freedberg et al. 2006; Janzen & Phillips 2006; Orlando & Guillette 2007; Rempel & Schlenk 2008), but not every organism is equipped with the genetic and physiological capacity to change sex. In humans (and presumably in most other mammals), for instance, Sry and related sex-determining genes (DAX1, SF1) have pleiotropic effects on metabolic, nervous and skeletal functions (Kent et al. 1996; Ramkissoon & Goodfellow 1996), which likely links sexual differentiation too closely to other developmental processes to allow for functional reversal of the primary sex in these species.
Artificially induced sex reversal is often used in aquaculture by exposing fish to exogenous steroids (Mair et al. 1997; Kitano et al. 1999; Piferrer 2001; Muller-Belecke & Horstgen-Schwark 2007) or different rearing temperatures (Tessema et al. 2006) in order to produce the preferred sex (Hunter & Donaldson 1983; Nakamura et al. 1998; Piferrer 2001). More than 50 fish species have been studied in such steroid- or temperature-mediated sex reversal trials (Devlin & Nagahama 2002; Cnaani & Levavi-Sivan 2009). Trials have included androgens, oestrogens, and precursor steroids administered by immersion or inclusion in the diet at various stages of development, in different dosages, and for different durations. For instance, 17α-methyltestosterone has been found to be very effective in inducing masculinization, and treatment with the estrogens oestradiol-17β, ethynylestradiol, and diethylstilbestrol has in most cases resulted in the feminization of genotypic males (Devlin & Nagahama 2002; Cnaani & Levavi-Sivan 2009). Even though these experiments have shown that the sex of many fish can be manipulated relatively easily (especially with endocrine disrupting chemicals) it is important to note that the sex ratio response observed after administration of an unnaturally high dosage of certain chemicals, or after exposure to temperatures not usually encountered in the wild, does not necessarily reflect the species’ or the population’s natural capacity to change sex in response to an environmental trigger. These studies merely demonstrate that exposure to sufficiently extreme conditions during sensitive ontogentic periods will probably affect the sex ratio of many fish populations. This can be of economical importance in aquaculture but may be of little or no ecological or evolutionary significance. Figure 1 lists major orders within teleosts with examples of species where ESR has been observed under artificial or natural conditions.
Although thermal or chemical sensitivity of gonadal sex differentiation has been documented for an increasing number of fish species since the first report of Conover & Kynard (1981), the conditions and consequences of ESR are not yet well understood (Devlin & Nagahama 2002; Godwin et al. 2003; Baroiller et al. 2009). One important insight from recent work on gonad differentiation is that aromatase inhibition at masculinizing temperatures seems to be a well conserved and central feature of environmental sex reversal in reptiles (Ramsey & Crews 2009) and fish (Guiguen et al. 2010). However, few empirical studies have addressed the mechanism by which genotype and environment interact to control sex (Crews et al. 1994; Warner et al. 2008; Baroiller et al. 2009). The emergent pattern suggests that the susceptibility of a sex determination system to alteration by environmental effects depends on the underlying genetic architecture (Magerhans et al. 2009). It is conceivable that when many genes with small additive effects are involved, the impact the environment has on sex determination is larger, as suggested by the results of a comparative analysis of Antlantic silverside (Menidia menidia) populations that differ in the number of genetic factors determining sex (Lagomarsino & Conover 1993).
Transitions between genetic and environmental sex determination systems
A theoretical treatment known as the Charnov and Bull model predicts that evolutionary transitions between GSD and ESD are possible (Charnov & Bull 1977). Following from sex allocation theory (Charnov 1982), selection should favour ESD mechanisms over genetic mechanisms when the environment affects male and female fitness differently. Hence, GSD and ESD can be regarded as two peaks in an adaptive landscape. If, on the other hand, the environment fluctuates, frequency-dependent selection is predicted to favor GSD (Bull 1983). Although the significance of this concept is widely accepted, it only recently received first empirical support from a study on lizards which demonstrated that the fitness of each sex was maximized by the incubation temperature producing that sex (Warner & Shine 2008).
One of the best-studied examples for adaptive transitions between ESD and GSD is the sex determination system of the Atlantic silverside (M. menidia). In this species of fish, sex determination is thought to be controlled by an interaction between major genetic factors, polygenic factors, and temperature. The relative contribution of each of these factors to the final sex of an individual varies with latitude (Lagomarsino & Conover 1993). Populations at high latitudes contain major sex determining gene(s) that override temperature effects while at low latitudes such major genetic factors are absent. It has been suggested that the evolutionary mechanism behind this pattern may be that a more thermo-sensitive sex determination system gives higher fitness returns where the breeding and growing season is longer, as is the case in low latitude populations (Conover & Heins 1987). In low latitudes more females are produced at the beginning (at lower average temperatures) and more males at the end of the breeding season (at higher average temperatures) (Conover & Kynard 1981). An early birth affords females a longer growing season and female offspring seem to benefit more than male offspring from large body size at reproductive age in the following year (Conover 1984).
These findings, together with accumulating phylogenetic evidence for frequent transitions between sex-determining mechanisms (Sarre et al. 2004; Mank et al. 2006), support the view that GSD and ESD represent opposite endpoints on a continuum rather than discrete categories. Populations may shift in one direction or the other in response to changing ecological conditions (Barske & Capel 2008), this leaving considerable scope for interactions between genetic and environmental influences on sex determination. Bull (2008) outlined that perhaps the most interesting question in this context is whether the coexistence of the genotypic and thermosensitive mechanisms of sex determination (as recently shown to be present in two species of lizards; Quinn et al. 2007; Radder et al. 2008) is adaptive or whether it should rather be interpreted as accidental (e.g. if the temperature effect is a relic from an ancestry without sex chromosomes).
Variation in sex ratio not caused by ESR
As soon as environmental or social factors can affect sex determination, widely varying family sex ratios are to be expected and are indeed often reported (Bull 1983; Penman & Piferrer 2008). In the following, we give a brief overview of the different adaptive responses to changing environmental conditions with respect to sex ratio. These responses often vary with the underlying sex determination system. We focus here on cases that do not involve ESR (they will be discussed in the next chapter).
GSD species with major sex factors typically produce a balanced 1:1 offspring sex ratio due to random segregation of sex chromosomes. Hence this genetically based mechanism fits with predictions from sex ratio theory (Fisher 1930). However, problems occur if the prevailing environmental conditions render one sex more successful than the other, i.e. if the production of a balanced primary sex ratio is not maximizing parental fitness. To avoid fitness loss, GSD species that invest heavily into individual offspring (as many mammals or birds) have evolved several adaptive responses to actively modify family sex ratio, such as temperature-induced differential mortality of male and female offspring genotypes (Burger & Zappalorti 1988), or embryo abortion or resorption (Krackow 1992; Blackburn 1998). It is important to note that these scenarios are not ESD because temperature or other environmental factors do not affect sex determination itself but simply prevent the development of offspring of a given, genetically determined sex or modify parental investment (e.g. in the case of embryo abortion or resorption).
Adaptive sex ratio adjustment is often encountered in polygynous species, which experience large variations in male reproductive success, where females in better condition can maximize their fitness by producing more offspring of the more costly sex. For example, sons in good condition often have greater reproductive success relative to daughters in good condition. Indeed, females in poor condition often produce more female offspring as the capacity to produce high quality offspring declines (e.g. Komdeur et al. 1997; Nager et al. 1999; Whittingham & Dunn 2000; Proffitt et al. 2008). Facultative adjustment of sex ratio in response to maternal condition has recently been suggested to also occur in some human populations (Pollet et al. 2009). In mammals, the effects could be produced by high levels of circulating glucose that sometimes seem to favour the survival of male blastocysts (Larson et al. 2001; Mathews et al. 2008).
Some species with sex determination systems susceptible to environmental effects have evolved facultative behavioural responses that allow the parents to actively influence the sex of their offspring, for example by temperature-dependent selection of oviposition or rearing sites in patchy environments as shown in many reptiles (Janzen 1994; Roosenburg 1996; Warner et al. 2008). Here, the sex-determining mechanism itself can be interpreted as a possible adjustment to sex ratio selection.
Sex ratio biased by ESR
The increasing presence of anthropogenic impacts on natural populations has been proposed to cause abnormal skewing of sex ratio in animal and plant populations (Hayes et al. 2002, 2004; Wang 2005; Cotton & Wedekind 2008; Grubler et al. 2008; Lee & Sydeman 2009). Global temperature change has received particular attention in this context, as it is expected to alter the sex ratios of species with TSD (Nelson et al. 2002; Schwanz & Janzen 2008; Hulin et al. 2009; Lee & Sydeman 2009; Zhang et al. 2009) or with temperature-induced sex reversal (Wedekind et al. unpublished). A recent analysis indicates that even small water temperature changes of 1–2 °C could significantly alter the sex ratio from 1:1 (males:females) to 3:1 in both freshwater and marine fish species (Ospina-Álvarez & Piferrer 2008). Also the increasing prevalence of sex hormones, hormone mimics, and endocrine-disrupting chemicals in natural watercourses has been suggested to create biased sex ratios (Orlando & Guillette 2007; Rempel & Schlenk 2008). A particularly intriguing case of possible ESR was discussed for a natural population of fall chinook salmon (Oncorhynchus tshawytcha) in the Columbia River. Nagler et al. (2001) reported a high frequency (84%) of a genetic marker for the Y chromosome detected in wild phenotypic females which was not observed in hatchery-raised females. A possible explanation is that genotypic males (XY) had been sex-reversed to phenotypic females by unknown environmental factors. If XY females reproduce with normal XY males, this creates the potential for an abnormal YY genotypes produced in the wild. YY genotypes would produce all-male offspring and alter sex ratios significantly which, on the long run, could affect population viability.
Population management using parental sex ratio adjustment
The insight we have gained into the potential factors controlling sex allocation and sex reversal can now become instrumental for the management of populations. In the following we give an overview of and how we may apply this knowledge to both endangered and invasive species. Before we explore various possible applications, a note of caution: At the beginning of breeding programs manipulating sex-ratios, populations are often purposefully driven through genetic bottlenecks. Artificially changing the sex ratio can lead to increased demographic stochasticity, higher inbreeding rates, loss of genetic variance, and fixation of deleterious mutations due to genetic drift. Simulation studies, however, predict that these negative effects can sometimes be outweighed by accelerating population growth and increasing reproduction rates (Wedekind 2002; Cotton & Wedekind 2008).
Generally, manipulating family sex ratio can be achieved through non-invasive methods like changing specific environmental or social conditions, or through more invasive methods including hormone treatment of embryos or sperm sexing (reviewed in Wedekind 2002). These methods may be particularly rewarding in captive populations where assisted reproductive technologies can be used. Wedekind (2002), Lenz et al. (2007) and Cotton & Wedekind (2007b, 2008) showed in theoretical treatments how the induced ‘overproduction’ of daughters can, under certain conditions, lead to increased population growth.
The Trivers–Willard hypothesis holds interesting applications for the conservation of populations whose viability is compromised or endangered by unbalanced sex ratios. Females are predicted to manipulate the sex of their offspring in response to characteristics of the rearing environment which may also depend on parental (and particular maternal) effects (Trivers & Willard 1973). If parents can affect family sex-ratio, population management may sometimes benefit from influencing such parental decisions (Wedekind 2002). This is especially so if number of females or the number of available eggs limits population growth; as seems to be the case in several species [for example in the European eel (Anguilla anguilla); Cnaani & Levavi-Sivan 2009; ]. The Kakapo (Strigops habroptilus), a nocturnal flightless parrot of New Zealand, is critically endangered and has been subject to supplementary feeding programs since 1989. Ironically, an analysis of sex allocation revealed that supplementary feeding has led to highly skewed sex ratios (Tella 2001) because females in good condition produce more sons than daughters (70% of offspring were sons; Clout et al. 2002). Robertson et al. (2006) recently showed a way out of this conservation dilemma. They removed the male bias by altering the condition of females slightly to the point where brood sex ratio shifted back towards parity.
Another example of the genetic and demographic consequences of such management actions comes from the lesser kestrel (Falco naumanni). Lenz et al. (2007) worked out the specific parameter space of a Spanish population, i.e. its population size, the range of naturally occurring family sex ratios, and the constraints of an already existing captive breeding program. Their study indicates that manipulating both female condition and male attractiveness of the captive breeders (in order to obtain more female offspring, which can then be reintroduced) may lead to a successful long-term conservation of the population. Figure 2 indicates the possible consequences of sex ratio manipulation, i.e. the initial loss and the potential long-term gain in population size.
Increasing population growth with Trojan sex gene carriers
A Trojan gene was originally defined as a gene that drives a local population extinct as it spreads due to destructive self-reinforcing cycles of natural selection (Muir 2004). Carriers of Trojan sex genes are individuals that possess sex chromosomes or genes opposite from what their phenotype predicts, e.g. XX individuals that have been artificially changed to phenotypic males, or XY individuals that have been changed to phenotypic females, for example by ESR. Figure 3 illustrates how different types of carriers of Trojan sex genes can be produced, and how a stock of Trojan individuals can be maintained. Gutierrez & Teem (2006), Cotton & Wedekind (2007b, 2008), and Gutierrez (2009) modelled the repeated introduction of sex-reversed individuals (from here on ‘Trojans’) into wild populations. Cotton & Wedekind (2007b) demonstrated how the introduction of Trojans can boost population growth and potentially rescue populations with skewed sex ratio from extinction. They showed that the release of WW females or WW males in a ZZ/ZW sex determination system can cause a female-biased sex ratio in later generations and result in an increase in both the absolute and the genetically effective population size. The impact on population growth was much higher compared to what would be achieved if the same number of wild type (not sex-reversed) females was introduced.
Obviously, the viability of YY or WW individuals is crucial for the maintenance of a population that was stock-enhanced with Trojans or is under the influence of naturally occurring ESR (Cotton & Wedekind 2008) and first efforts to measure the fitness of Trojan sex chromosome carriers and their offspring are underway. Breeding programs involving sex reversal or chromosome set manipulations in teleost fish have revealed that YY or WW genotypes are often viable and fertile although there is large variation among species (reviewed in Penman & Piferrer 2008; Cnaani & Levavi-Sivan 2009). The performance of YY or WW individuals may be related to the degree of sex chromosome differentiation (Frolov 1991). For instance, YY male coho salmon (Oncorhynchus kisutch;Hunter et al. 1982), YY male rainbow trout (O. mykiss;Parsons & Thorgaard 1985), and YY males of the common carp (Cyprinus carpio;Bongers et al. 1997) are often viable and fertile and they are all species with undifferentiated sex chromosomes. In the guppy (Poecilia reticulata) on the other hand, a species with some degree of sex chromosome differentiation, YY males are only viable if the two copies of the Y chromosome came from different strains (Volff & Schartl 2001). In rainbow trout populations with heteromorphic sex chromosomes, the evidence for viability is mixed. Also mammalian YY genotypes seem to not be viable which is thought to be due to the small number of genes residing on the mammalian Y-chromosome (Goodfellow & Lovell-Badge 1993).
The method of releasing large numbers of sex-reversed individuals into the wild in order to boost population growth needs to be applied with care because it could eventually produce negative long-term effects. Two theoretical treatments have shown that the release of large numbers of sex-reversed individuals can ultimately result in the extinction of sex-determining genes (Kanaiwa & Harada 2002; Cotton & Wedekind 2008). Kanaiwa & Harada (2002) investigated the influence of stock enhancement on the sex-determining system of the Japanese flounder (Paralichthys olivaceus). They showed that releasing hatchery-bred, sex-reversed males (i.e. genotypic females that were masculinized by ESR) could drive genetic males to extinction in the wild population. Once the hatchery release of Trojans is terminated, population viability may become seriously threatened if few to no males are produced in the wild without sex reversal.
A similar problem occurs if ESR affects natural populations to such an extent that one of the sexes is mainly the product of the sex-reversing environmental effects. If reproduction mostly takes place between homogametic individuals (of which one is sex-reversed), sex-determining genes can be driven to extinction. A cessation or modification of the environmental conditions triggering sex determination could then result in a breakdown of the population because the population would consist of individuals of only one sex. Cotton & Wedekind (2008) studied the population effects of various degrees of ESR in computer simulations. Their parameter settings assumed that (i) ESR has no negative effect on viability and fertility, (ii) that populations do not adapt to ESR, (iii) that males can mate multiply, and (iv) that the number of females is limiting population growth. Under these settings, environmental feminization quickly increased the proportion of females and therefore population size (Fig. 4a). However, populations declined again when X-chromosomes become rare as a consequence of feminization (Fig. 4a). Analogously, environmental masculinization first reduced the number of females and hence population growth (Fig. 4b). With strong ESR, population may quickly go extinct while at low degrees of ESR, Y-chromosomes become rare, and the population consequences of ESR are mitigated (Fig. 4b). Figure 4 also illustrates possible population consequences of a cessation of ESR (here at generation 60). Populations subjected to strong ERS all crashed either before or shortly after the cessation of ESR. Only those exposed to weak ESR recovered.
Past experiments have shown that fish populations may have the genetic resources to evolve rapidly as a response to a sustained environmental pressure skewing sex ratios (see e.g. the removal of ESD in less than 12 generations in M. menidia; Conover et al. 1992). The development of more elaborate models capable of simulating the evolution of ESD or ESR over a few generations seems therefore timely. It remains to be seen if allowing sex determining mechanisms to evolve avoids the dramatic consequences current models have demonstrated (Cotton & Wedekind 2008).
On a different note, the introduction of Trojan sex genes carriers for stock enhancement into natural populations may cause some of the same concerns as the introduction of genetically modified organisms. The potential for unintended, pleiotropic effects of the introduced sex genes or sex chromosomes on other traits of the sex reversed individuals and their offspring should prompt careful consideration about possible ecological consequences. For example, the introduction of Trojans may influence the environmental adaptability and geographical range of the population, or alter predator–prey relationships. Also, if Trojan males have mating advantages relative to wild type males, but if their offspring have reduced viability relative to wild type offspring, the population may decline and go extinct (Muir & Howard 1999, 2002). Population extinction has recently been demonstrated in a deterministic model for the introduction of genetically modified organisms (Hedrick 2001).
Managing invasive populations with Trojan sex gene carriers
Gutierrez & Teem (2006) and Gutierrez (2009) analysed the employment of females containing Trojan sex chromosomes as a method for eliminating harmful populations with male heterogamety. The models reveal that when phenotypic females containing two Y chromosomes (YY) are released and mate with wild type (XY) males, this can cause a male-biased sex ratio in subsequent generations owing to the disproportionate influx of Y chromosomes. As a result, true (XX) females become increasingly rare and the population may become extinct. To briefly outline this scenario in more detail: In fish species with male heterogamety, viable females that carry two YY chromosomes instead of two XX chromosomes can be created over two generations by, for example, oestrogen treatment during early development (Fig. 3). Such YY females can then be released to mate with normal XY males. The resulting progeny will consist of only XY and YY males. The latter males will again have only XY sons if they mate with normal XX females or only YY sons if mated with further introduced sex-changed YY females. Continuous introduction of sex-changed YY females will therefore lead to ever-increasing sex ratio biases towards males and diminishing numbers of true XX females which could eventually lead to population extinction. Analogous dynamics are expected in systems in which sex determination is influenced by one of a few major genes on autosomes (Cotton & Wedekind 2007a).
Probably the greatest advantage of using Trojans to eradicate exotic fish species is that it is a specific biocontrol method that targets only the exotic and not the native species. Thus, it has minimal impact on other members of the aquatic community. Unlike other biocontrol methods with little or no impact on non-target species (e.g. adding sterile males to the population; Knipling 1955), the efficiency of using Trojans does not depend on the elimination of all matings between wild type males and females in order to affect the total population. Instead, reduction in the numbers of wild types is achieved over multiple generations as a consequence of matings between Trojans and wild types. Further, the method can potentially be used on all exotic fish species with an XY sex determination system, offering a new tool for the elimination of invasive species.
There are, however, also potential pitfalls to using Trojan sex chromosomes for invasive species management. Probably the most critical parameter to be evaluated for testing the efficacy of this method is the fitness of sex-reversed individuals (Cotton & Wedekind 2008) as discussed above. Another factor affecting the rate at which wild types are eliminated from the population is the frequency dependence of mating probabilities, which is caused by the relative abundances of different genotypes in the population. Aikio et al. (2008) showed, albeit in a different context, that the lack of frequency dependence would result in a serious overestimation of both invasion and subsequent extinction efficacy. Additional factors could limit the application of the model. Tilapia, for example, are known to have minor sex-determination genes located on autosomes (Baroiller et al. 2008). These genes can cause deviations in the expected sex ratios of progeny fish and may hence influence the rate at which extinctions occur (Gutierrez & Teem 2006). Furthermore, the blue tilapia (Oreochromis aureus) has a ZW sex determination system whereas Nile tilapia (Oreochromis niloticus) has a XY system and hybrids between these species do occur in the wild. The use of YY fish to cause eradication of invasive tilapia in the wild would thus be restricted to O. niloticus and could not generally be applied to all tilapia species, or their hybrids. However, this may also be seen as an advantage of the Trojan method if it only works for the target species it was designed for.
There are, of course, practical limitations that first have to be overcome before Trojan gene carriers can be used for population management, whether it is to enhance or reduce population size. Generating controlled crosses and the development of hormonal masculinization or feminization can be challenging in some species. Also, to drive an invasive population to extinction, the production of large numbers of Trojan carriers and multigenerational designs would probably be required. The most urgent questions arising are: How many and how often do Trojan gene carriers have to be introduced into the system to achieve extinction, and what is the expected timeframe? Answers to these questions are crucial for determining the feasibility of this method. At present, suitable empirical data that would validate the model are not yet available.
Long-term genetic consequences of ESR
Environmental sex reversal usually changes the population sex ratio and hence the variation in reproductive success within a population. In small populations, increased variation in reproductive success can lead to a loss of genetic variation and hence to an increased rate of fixation of deleterious mutations. Apart from these potentially harmful, long-term genetic effects, there are also potential beneficial effects of ESR that may enhance the average viability of future generations. In this section, we briefly outline how ESR may counteract the degeneration of the Y chromosome as recently suggested by Perrin (2009).
In male heterogametic systems, the Y-chromosome is known to accumulate deleterious mutations due to the overall suppressed recombination rates often encountered in males (Burt et al. 1991; Lenormand & Dutheil 2005; Hedrick 2007). For example, the female:male ratio of recombination in Atlantic salmon (Salmo salar) was estimated to be 8.26 (Moen et al. 2004) or 3.92 (Gilbey et al. 2004), and in brown trout (S. trutta) to be 6.4 (Gharbi et al. 2006). An important consequence of reduced recombination is genetic degeneration, i.e. poorer gene function, loss of genes, and sometimes the evolution of dosage compensation (Bergero & Charlesworth 2009). Human-induced changes in life-history and reproductive traits (e.g. due to exploitation or supplementary breeding) often enhance the rate of genetic deterioration (Cotton & Wedekind unpublished). However, in sex-reversed XY genotypes differentiating into viable phenotypic females, meiosis presumably follows a female-typical protocol (Wallace et al. 1997; Matsuda et al. 1999; Kondo et al. 2001; Lynn et al. 2005; reviewed in Perrin 2009). If so, recombination rates on the Y-chromosome would become enhanced and large linkage groups carrying deleterious mutations may be broken up. As a consequence of ESR, followed by natural and sexual selection, the Y-chromosome might thus undergo a ‘rejuvenation’ process that it would otherwise not experience in an all-male background of phenotypic differentiation (Perrin 2009). Analogous arguments may be valid for the WZ/ZZ sex determination system. An important consequence of this observation is that YY or WW males or females produced from sex-reversed individuals are often viable and fertile (Cnaani & Levavi-Sivan 2009) and may hence be used to manipulate growth and decline in population management programs.
Perrin’s hypothesis may also offer a solution to the puzzle of why extremely heteromorphic sex chromosomes are common in higher vertebrate species (e.g. in mammals), while sex chromosomes in lower vertebrates (e.g. amphibians and fish) are usually homomorphic (Devlin & Nagahama 2002; Schartl 2004). The high turnover rates of sex determination systems (e.g. from male to female heterogamety) found in some fish taxa may only partly explain this observation (Schartl 2004; Volff et al. 2007; Mank & Avise 2009). Purging deleterious mutations from the sex chromosomes of sex-reversed individuals following ESR as suggested by Perrin would help to explain this pattern, but the idea awaits experimental testing.
One way to test Perrin’s ‘fountain of youth’ hypothesis experimentally is to sex reverse XY genotypes, let them reproduce with wild type XY males, and test if their (non-sex-reversed) male XY offspring show increased variance in fitness-related traits compared to the male XY offspring from matings between two normal, wild type individuals (XX female × XY male). For this purpose, one could employ different fish species representative of the gradual steps of sex chromosome differentiation, i.e. (i) taxa with an all-autosome karyotype, (ii) taxa with largely homomorphic sex chromosomes, and (iii) taxa with heteromorphic sex chromosomes. One prediction would be that in species with all-autosome karyotypes or whose Y-chromosome is not degenerated and has not accumulated many deleterious mutations, XY offspring from a sex-reversed mother and a wild type male would not show significantly more variance in fitness traits than normal XY offspring. We would also predict that the variance in fitness would differ between the two types of XY offspring in species with hetermorphic sex chromosomes.
Temperature is obviously an important regulator of physiological processes in ectotherms, especially during embryogenesis. Global warming may therefore be predicted to have negative implications for geographically restricted populations of many different taxa, with freshwater habitats being particularly susceptible (Mote et al. 2003; Hari et al. 2006). The decline of many salmonid populations is usually attributed to habitat fragmentation or destruction, migration barriers, and overfishing (Reynolds et al. 2001; Quinn 2005). How much of this decline is actually caused by changes in temperature is still only marginally resolved. In light of this, the prevalence of ESD and ESR, and their significance for natural populations, remains largely unknown and will be an avenue of future study.
The ability to adapt depends, among others, on the speed of environmental change, the selection coefficient, the size of the population, its genetic diversity, and on reproductive strategy (e.g. the reproductive excess). Rapid evolution in response to environmental changes is possible (Carroll et al. 2007; Hendry et al. 2007) and has been demonstrated in various salmonid species (Koskinen et al. 2002; Heath et al. 2003; Edeline et al. 2007; Gregersen et al. 2008; Kinnison et al. 2008). Conover et al. (1992) found that TSD was lost in fewer than 12 generations from a population of M. menidia via frequency-dependent selection, and a population of O. niloticus was found to respond quickly to selection for increased temperature sensitivity (Wessels & Hostgen-Schwark 2006). Analyses on experimental broodstocks of rainbow trout (Magerhans et al. 2009) and sea bass (Saillant et al. 2002) demonstrated considerable genetic variation for temperature-induced sex determination, but it is unknown for the majority of natural populations with skewed sex ratios if they contain sufficient genetic variation to re-establish more balanced ratios. Some of the currently observed changes may be too rapid, especially for small and isolated populations, to allow them to adapt in time. If global temperature changes do emerge as a major threat to biodiversity, countermeasures such as increased shading of water bodies, the installation of heat exchangers, or temperature treatment of eggs and larvae for supportive breeding in hatcheries, may provide the additional time that populations need to adapt before they extinction. Careful implementation of these countermeasures will be necessary because, as mentioned above, any temperature manipulation may lead to the extinction of genetic sex-determining factors and may enhance the effects of demographic stochasticity. Sarre et al. (2004) suggested that in order to reduce the risk of losing the genetic component of sex determination and to enhance the effectiveness of stocking, it is crucial to select fish for hatchery brood stocks that continue feeding the sex-determining genes into the wild population. Also, the continuous monitoring of the sex ratio of the released fish is important (Kitano et al. 1999). Considering that the environmental contribution to sex differentiation has not been rigorously studied in many species, the precautionary principle demands keeping the contribution of stocking with sex-reversed individuals as low as possible if population growth is intended.
We urge researchers to report the sex ratios of the populations they are studying in order to obtain a more intimate understanding of the frequency of sex ratio distortion in the wild. Data on sex ratio can be easily obtained, often as a byproduct of other investigations. Evidently, adult sex ratios are not necessarily correlated to sex ratios at birth or conception (e.g. due to sex-specific mortality; Holtby & Healey 1990) and it is therefore difficult to draw conclusions on the potential cause of an observed skew in sex ratio without further investigations. If one was to design a study on sex ratio selection, ESR or sex allocation, the experimental design should clearly allow for the quantification of genetic versus environmental effects. The so-called full-factorial breeding design (also known as ‘nested half-sib’ or ‘North-Carolina II’ (NCII) design; Lynch & Walsh 1998), allows for direct quantification of additive and non-additive genetic variation, and maternal environmental effects on any measurable trait, including offspring viability and family sex ratio after incubation at various environmental conditions. Until recently, this powerful design has been used mainly in crop science, but many externally fertilizing amphibians (Welch et al. 1998) and fish (e.g. Wedekind et al. 2001, 2007; Pitcher & Neff 2006) are also fitted to the application of this design due to the availability of hundreds or thousands of eggs per female, with statistically independent sibships and large sample sizes even for within-family analyses (Wedekind et al. 2004). In such experiments, potentially relevant environmental factors (e.g. stable versus fluctuating temperatures, physical disturbance, pathogen challenges, hormone-active substances) can be tested for their interaction with genetic sex determination and compared to maternal effects. Such studies will shed light on some of most urgent questions of modern conservation biology, for instance the effects of temperature changes, pollution with hormone-active substances, and the impact of habitat fragmentation on the demography and genetics of populations. At the same time, such studies will provide information about the viability of sex-reversed individuals at various stages of their life cycle.
Another currently understudied line of research in the context of anthropogenic influences on natural fish populations are the effects of fishery-induced selection on sex ratio and sex determination mechanisms. Several life-history traits, for example size-at-age (Swain et al. 2007; Thomas & Eckmann 2007; Nussléet al. 2009) or age-at-maturity (Heino et al. 2002; Grift et al. 2003; Yoneda & Wright 2004) are known to be under the direct influence of size-selective fishing. Exploited fish populations often respond to size-selective harvest with slower growth rate or smaller size at maturity to the extent that some populations lose their sexual dimorphism entirely (Hard et al. 2008). At the same time, growth rate has been shown to influence the prevalence of temperature or socially induced sex reversal in some species of fish (Francis 1990; Koumoundouros et al. 2002). Other species undergo sex reversal in response to the overall density of conspecifics (Francis & Barlow 1993) or in response to population size (Brykov et al. 2008). It is therefore easy to imagine that fishery-induced selection can lead to unbalanced sex ratios in wild stocks. We may not yet be aware of the full dimensions of the impact that fisheries and harvesting in general can have on the evolution of sex determination and sex ratio selection in natural populations. Theoretical and empirical work in this field should be encouraged.
We thank Louis Bernatchez, Jean Paul Danko, Sarah G. Kenyon, Nicolas Perrin, and three anonymous referees for helpful comments and discussion, the participants of a CUSO workshop on ‘Evolution of sex-determination mechanisms’ in La Sage, Switzerland, in June 2009, for valuable input, and the Swiss National Science Foundation for funding.
•Sex determination: Process that directs differentiation to proceed down the male or female pathway of ontogeny.
•Primary sex differentiation: The transformation of undifferentiated gonads into testes or ovaries.
•Secondary sex differentiation: The transformation of differentiated gonads, i.e. testes into ovaries or ovaries into testes (sex reversal in sequential hermaphrodites).
•Genetic sex determination (GSD): Sex determination based on sex chromosomes or sex-linked factors on autosomes.
•Environmental sex determination (ESD): The first ontogenetic difference between sexes is environmentally induced, i.e. sex is not fixed at conception via sex-determining genes. The sex ratio response to environmental factors (e.g. temperature) happens within the range of conditions naturally encountered by the population.
•Temperature-dependent sex determination (TSD): ESD based on temperature.
•Environmental sex reversal (ERS): The sex of an individual (e.g. as genetically determined at conception) is reversed during ontogeny.
•Trojan gene carrier: An individual possessing sex chromosomes or sex-linked genes opposite from what their phenotype predicts.
Box 1: Molecular methods for the identification of genetic sex factors
Identifying sex chromosomes via karyotyping is often difficult even with high-resolution cytogenetic techniques (Ezaz et al. 2006; Kawai et al. 2007). Direct DNA analyses include (i) comparison of male and female DNA profiling/fingerprinting (Felip et al. 2005); (ii) subtractive techniques to look directly for sequences which differ between the male and female genomes (Hale et al. 2009); (iii) candidate gene approaches where genes or sequences that are sex-determining or sex-linked in a related species are analysed in the target species (McCormick et al. 2008); (iv) linkage mapping approaches where phenotypic sex is scored with many segregating markers (mostly DNA-based) to place the sex-determining gene(s) into one or more linkage groups, which may facilitate positional cloning of such genes later (Shapiro et al. 2009); and (v) molecular-cytogenetic approaches (Ross & Peichel 2008), e.g. to relate linkage and karyotypic data.
Less than 50% of all fish are currently known to have sex-linked genes, but even in these cases, sex determination is usually not purely genetic (Devlin & Nagahama 2002). The diversity of sex determination systems in teleosts might be partly attributed to the developmental plasticity of their gonads (Francis 1992). Testes and ovaries in teleosts derive from the same precursor tissue and can rather flexibly differentiate into male or female organs, i.e. gonadal development and sexual differentiation are somewhat decoupled (Mank et al. 2006). This may create more opportunity for environmental effects on sex determination, which in turn could influence the evolution of genetic sex factors. As a possible consequence, high turnover rates in sex determining genes may ultimately prevent the evolution of genetically hardwired sex chromosomes.
Teleosts possess genomic peculiarities that may be conducive to the diversification of sex determination systems (Mank et al. 2006). Genomic evolution seems to be more rapid in fish than in most other vertebrates. Various taxa within the teleost show a fast pace of genic and genomic duplications (Robinson-Rechavi & Laudet 2001), that have happened both in the recent (Vandepoele et al. 2004) and ancient past (Taylor et al. 2003). Duplications initially provide redundancy that may then create additional evolutionary flexibility (Ohno 1970). The evolution or the loss of master regulators of sex determination and their mode of inheritance can have profound consequences on the evolution of sex ratio, sexual selection, sexual dimorphism, sexual conflict, and ultimately on speciation and extinction as exemplified in fish (Lande et al. 2001; Kocher 2004; Kitano et al. 2009; Roberts et al. 2009).
Rike Stelkens uses freshwater fish species as model systems to investigate the evolutionary forces and ecological conditions that maintain biodiversity. She currently focusses on the population genetics and conservation of various endangered salmonid species and aims at disentangling additive genetic effects from maternal and environmental effects on various life history traits including family sex ratio. Claus Wedekind studies the selective forces that currently act on fish in the wild, including sexual selection, host-pathogen interactions, and the direct and indirect effects of human activities (e.g. non-random harvesting, supplementary breeding, species invasions, or climate change). His primary empirical systems are salmonid fishes. He also studies cooperation in conflict situations, exploring the link between cooperation theory and conservation issues.