* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Evolution of the Y Sex Chromosome in AnimalsY chromosomes
Gene expression profiling wikipedia , lookup
Medical genetics wikipedia , lookup
Polycomb Group Proteins and Cancer wikipedia , lookup
No-SCAR (Scarless Cas9 Assisted Recombineering) Genome Editing wikipedia , lookup
Ridge (biology) wikipedia , lookup
Adaptive evolution in the human genome wikipedia , lookup
Genetic engineering wikipedia , lookup
Public health genomics wikipedia , lookup
Dominance (genetics) wikipedia , lookup
Biology and consumer behaviour wikipedia , lookup
Minimal genome wikipedia , lookup
Human genetic variation wikipedia , lookup
Quantitative trait locus wikipedia , lookup
Artificial gene synthesis wikipedia , lookup
Oncogenomics wikipedia , lookup
History of genetic engineering wikipedia , lookup
Frameshift mutation wikipedia , lookup
Genomic imprinting wikipedia , lookup
Site-specific recombinase technology wikipedia , lookup
Designer baby wikipedia , lookup
Epigenetics of human development wikipedia , lookup
Genetic drift wikipedia , lookup
Polymorphism (biology) wikipedia , lookup
Skewed X-inactivation wikipedia , lookup
Genome evolution wikipedia , lookup
Gene expression programming wikipedia , lookup
Koinophilia wikipedia , lookup
Point mutation wikipedia , lookup
Neocentromere wikipedia , lookup
Y chromosome wikipedia , lookup
X-inactivation wikipedia , lookup
Population genetics wikipedia , lookup
Evolution of the Y Sex Chromosome in Animals Y chromosomes evolve through the degeneration of autosomes William R. Rice S ex chromosomes are part of the gender determination system of many organisms. They are common in animals and rare in plants. There are many forms of sex determination. Gender may be determined genetically or in response to environmental cues such as temperature or social circumstance. Genetic sex determination is typically expressed in the developing zygote but can also be mediated maternally, with some females producing only daughters and others only sons. Zygotic sex determination has ~ many forms. In bees, ants, and wasps fertilized eggs develop into females and unfertilized eggs into males. In the guppy, and many other fish, invertebrates, and plants, gender is determined by the presence or absence of a sex determining gene (G, or a cluster of tightly linked genes); Gg is male and gg female. The heterozygous (i.e., heterogametic) sex can be female or male, depending on the species. Gender can also be determined by specialized sex chromosomes (X and Y). In mammals, females are XX and males are XY; in birds, females are XY (sometimes designated ZW) and males are XX (sometimes designated ZZ); and in nematodes, females are XX and males are XO, in which denotes the absence of a homologous chromosome. More than two sex chromosomes are found in some species, ° William R. Rice is a professor in the Biology Department, University of California, Santa Cruz, CA 95064. © 1996 American Institute of Biological Sciences. May 1996 Y chromosome evolution is a model system for the adaptive sigificance of sexual recombination in whole organisms but these are complex derivatives of the familiar XX/XY system. Sex determination and sex chromosomes are sometimes, but not always, associated with a dosage compensation mechanism, such as X inactivation, which ensures equal gene expression in the XX and XY sex. In this article I focus on the Y sex chromosome of animals. I first describe the major distinguishing characteristics of the Y and the steps in its evolution from a nonsex chromosome (autosome). Next, I present the genetic theory developed to account for the Y's evolution. I then offer an overview of recent experiments that test the theory and discuss two experimental approaches for studying Y chromosome evolution. In the first approach, used in my own laboratory, a combination of artificial selection and visible genetic markers is used to create synthetic sex-determining genes and Y chromosomes, which are in turn used to experimentally simulate the early stages of Y chromosome evolution. The second approach is taken by Manfred Steinemann's and Sigrid Steinemann's group, whose molecular studies of a naturally occurring Y chromosome permit certain molecular aspects of its evolution to be directly observed. Finally, I place the work on the Y chromosome in a broader context by considering the Y chromosome as a model system for the study of the adaptive significance of sexual recombination in whole orgarusrns. The Y sex chromosome Two properties distinguish the Y chromosome from an ordinary autosome: Recombination with its homolog, the X chromosome, is absent or restricted to a small region, and it has highly reduced genetic activity within the nonrecombining region. Consider humans as an example. Both sexes have two copies of each of the 22 autosomal chromosome types plus a pair of sex chromosomes. Males ha ve an X and a Y sex chromosome (XY), whereas females have two X chromosomes (XX). In most respects the human X is organized genetically like an autosome. Its major distinction is that it undergoes dosage compensation, a phenomenon whereby, in females, most genes on one of the X chromosomes are inactivated, whereas these same genes on the single active X chromosome are transcribed at twice the rate than they would be on an autosome. In males, X chromosomes are also hyperactivated, in this case not to make up for an inactivated X but to offset the genetic inertness of the Y 331 over most of its length. Molecular studies (summarized in Wachtel 1994)suggestthatthe human Y chromosome is nearly devoid of functional genes; in contrast, many hundreds of genes have been found on the X chromosome. Some genes found on the tip of the Y chromosome (the pseudoautosomal region) recombine with homologous genes on the X chromosome. The remaining genes are in the nonrecombining portion of the Y (the "differential segment," which constitutes most of the Y chromosome). To date, the male-limited differential segment of the Y chromosome has been demonstrated to contain the Sry gene, which causes development to follow the male rather than the female pathway; one or a few genes necessary for male fertility; and several other genes that appear to be unrelated to sex determination and fertility. New Y-linked genes may be found in the future, but it is clear that most genes located on the X chromosome are absent from the nonrecombining portion of the Y chromosome. What evolutionary processes led to dimorphic X and Y sex chromosomes? Stages ofY chromosome evolution Comparative studies across many taxa indicate that the Y has evolved independently from an ordinary au tosome many times in many different taxonomic lineages (see for review Bull 1983, Mitwoch 1967). I focus on the evolutionary processes that contributed to the transition from an ordinary autosome into the degenerate Y chromosome in animals, although most of the same principles also apply to plants (Charlesworth [1991] reviews additional information on plants). Because Y chromosomes appear to have evolved slowly over millions of years, direct observation of the steps by which the Y chromosome evolved from an ordinary autosome is impossible . A working hypothesis can be developed, however, from animals such as fish, in which the Y chromosome has evolved independently many times. The Y chromosome began to evolve at different times in different lineages, so Y chromosome evolution is at different 332 Ordinaryautosomes y X y y X X y X Polygeni c sex determination I Her ma phr odite ~ Genic sex determination Semlchro mo so mal sex determination Ch I romosoma . . eterm lnatt on ~ex XO sex determination Environmental sex det ermination Figure 1. Proposed pathway for the evolution of the Y sex chromosome in animals. The model is based on observations of different fish species. The most primitive Y chromosome carries a dominant gender-determining allele (G) that is associated with genic sex determination . A nonrecombining differential segment then arises surrounding the G locus (shading indicates nonrecomb ining region), giving rise to semichromosomal sex determination . Genetic deterioration of the differential segment of the Y chromosome then selects for dosage compensation (striped region-these genes are expressed at double-speed) on the corresponding portion of the X chromosome. Dosage compensation occurs in mammals and many other groups but appears to be absent in birds and ha s not been studied in fish. Expansion of the differential segment leads to chromosomal sex determination. Extensive deterioration of the nonrecombining Y chromosome may ultimately lead to its complete loss, in XO sex determ ination. stages in different extant species. By ordering the variety of Y chromosome types found in extant fish species, Kirpichnikov (1981) was able to construct a trajectory, shown in Figure 1, going from primitive (most similar to a typical autosome) to advanced (nonrecombining and lack ing most or all genetic funct ion) stages. The major steps in Y chromosome evolution in other groups of animals are thought to be similar, although this extrapolation is tentative because in most other groups of animals (e.g., mammals), most of the intermediate steps are no longer pre sent in extant species. Kirpichnikov's (1981) overview of the genetics of sex determination in fish provides comparative evidence that hermaphroditism, in which each individual has both male and female function, is the primitive state. The first stage in the evolution of a Y chromosome is the evolution of genic sex determination, in which a single dominant factor (G) determines gender. This dominant factor may be a single dominant allele or a collection of tightly linked genes, but for simplicity let us assume that G is a single dominant allele. We will also assume for simplicity that males are the heterogametic sex (Gg, produc- ing two types of haploid gamete, G or g) and females the homogametic sex (gg, producing only g gametes). Although males are most commonly the heterogametic sex, in those cases in which females are the heterogametic sex the same principles apply. The next stage in Y chromosome evolution is semichromosomal sex determination. Here the Y chromosome stops recombining with the X in a limited region (the differential segment) surrounding the G allele. Because recombination no longer homogenizes the genetic information in this region, for reasons described below, the Y-linked differential segment is free to diverge in sequence from the homologous region on the X. As th e differential segment becomes larger, for reasons described below, the X and Y chromosomes diverge to a greater extent. Once the differential segment constitutes all, or virtually all , of the Y chromosome, the system is referred to as chromosomal sex determination. The inviability of experimentally produced YY males in a number of fish specieswith small differential segments suggests that the Y begins to degenerate soon after a differential segment is formed (reviewed in Kirpichnikov 1981). Once the chromosomal stage BioScience Vol. 46 No.5 is reached, continued degeneration of the Y ultimately erodes virtually all of its genetic activity, thereby permitting it to be lost altogether, as occurs in XO sex determination. Reversals in the trajectory depicted in Figure 1 can occur. For example, chromosomal sex determination may be converted back to semichromosomal sex determination by the translocation of a large piece of an autosome onto the Y chromosome. The trajectory may also arrest at certain stages for long periods of evolutionary time. For example, the conversion from XY to XO sex determination has not been observed in some lineages. (Other, more complex forms of sex chromosomes, such as multiple sex chromosomes [XXY or XYY], also can evolve, but these are not reviewed here; Bull 1983.) From the deduced sequence of events in Figure 1, two factors appear to be critical in the evolution of the Y chromosome: elimination of recombination between the X and Y chromosomes due to continual expansion of the differential segment, followed by gradual degeneration of this region. What factors lead to suppressed recombination and degeneration of the Y chromosome? Theory for the evolution of suppressed recombination There are two major hypotheses for the evolution of suppressed recombination between the X and Y chromosomes. One is Nei's (1969) extension of Haldane's (1922) earlier suggestion that if there were multiple sexdetermining loci on a single chromosome and if some intermediate genotypes were prone to becoming infertile intersexes, then natural selection would favor the elimination of recombination between the contributing loci. Recombination is selected against because it is likely to continually generate the genotypes producing infertile intersexes. This hypothesis is actually a special case of the second hypothesis, which is based on the accumulation of sexually antagonistic alleles at loci that are tightly linked to the gender-determining locus (sexually antagonistic genes hypothesis; Bull 1983, Fisher 1931, Rice 1987a). Sexually antagonistic genes are loci with allelic variMay 1996 ants that are favored in one sex but ally antagorusnc allele, no matter disfavored in the other. how detrimental to females, is likely The sexually antagonistic genes to accumulate on a primitive Y chrohypothesis was motivated by early mosome. To see why, consider a genetic mapping studies of the guppy, male-benefit sexually antagonistic a common aquarium fish with genic allele introduced by mutation just sex determination (or an undetected one map unit (1 cM) away from the small differential segment). Males G allele. In this case the new mutant are highly ornamented, with a vari- allele will co segregate with the G ety of traits, such as bright body allele 99% of the time and therefore color, long tails, and flashy spots of will almost always be transmitted to varying color, size, and position. sons, where it is favored. Only 1 % of When the pioneering geneticist Winge the time will it cross over to the (1927) mapped 18 major genes pro- primitive X chromosome and be ducing these ornamental character- transmitted to daughters, where it is istics, he was surprised to find that disfavored. 17 were located on the sex chromoTight linkage to the G allele theresomes. Moreover, all of these were fore generates sex-biased gene translocated within two recombinational mission, which greatly facilitates the map units (centimorgans [cM]) of accumulation of male-benefit sexually the male-determining factor. This antagonistic mutations, even when they clustering of genes controlling unre- are highly deleterious to females and la ted traits is unlikely to occur by trivially beneficial to males (see Bull chance because the guppy, like hu- 1983, Rice 1987a, for quantitative demans, has 23 pairs of small chromo- tails). As a consequence, the constraints somes. Winge concluded that some- for the accumulation of male-benefit thing about the region near the sexually antagonistic mutations are far gender-determining locus makes it a more permissive in the vicinity of a "hot spot" for ornamentation genes. gender-determining locus than at ordiWhat is the evolutionary basis of nary locations within the genome. The the hot spot? Fisher (1931) was fa- result is the disproportionate accumumiliar with Winge's results and pro- lation of sexually antagonistic alleles posed a simple model based on sexu- near a major sex-determining locus. ally antagonistic genes. Fisher Hence the chromosomal region immecorrectly assumed that genes caus- diately adjacent to a gender-detering ornate characters, when ex- mining locus should be a hot spot for pressed in both sexes, as is likely to sexually antagonistic genes. be the case early in their evolution, Once sexually antagonistic alleles would be favored in males but disfa- begin to accumulate on the primitive vored in females (see Bischoff et al. Y chromosome near a gender-deter1985, Endler 1980). In both sexes, mining locus, they are likely to conthe ornamental characteristics would tinually cross over to the X chromake the fish more conspicuous to mosome (at a low rate for each predators and in that respect would generation but with substantial cube disadvantageous. However, or- mulative impact). This crossing over nate males, but not the ornate fe- creates a genetic load in females (or, males, have a substantial mating more generally, the heterogametic advantage, which can outweigh the sex). There are two evolutionary predation disadvantage. Alleles pro- ways to counter this load: sex-limducing ornamental characteristics ited gene expression, which prevents are therefore termed male-benefit the trait from being expressed in sexually antagonistic alleles. females, and suppressed recombinaGenetic theory predicts that at tion near the G allele, which keeps most locations within the genome, the sexually antagonistic allele on sexually antagonistic alleles will the Y chromosome and out of feaccumulate in the gene pool only males. Because both outcomes elimiwhen they proffer a net selective nate the genetic load in females, the advantage, that is, when the advan- accumulation of sexually antagonistage to one sex is larger than the tic alleles ini tiates a race between disadvantage to the other. At loci these two alternative evolutionary tightly linked to the G allele, how- pathways. Sex-limited gene expression is a ever, virtually any male-benefit sexu333 a b c d e f the observed reduction in recombination rate. No reduction in recombination rate was found in the chromosomal region flanking that selected for reduced recombination, demonstrating a highly localized response to selection. This study, along with other studies using D. melanogaster, suggests that suppressed recombination near a sex locus can evolve rapidly and that the evolution of reduced recombination will frequently win the race against sex-limited gene expression. If suppressed recombination does evolve near the G allele on the primitive Y chromosome, then, at least in theory, this suppression can in itiate a genetic cha in reaction that gradually eliminates recombination along all or most of the Y (sexually antagonistic genes chain reaction model; Figure 2) . Whether this genetic model explains the breakdown in recombination between the X and Y chromosomes depends cr itically on the availability of two types of genetic variation: that for localized suppression of recombination and that for sexuaIIy antagonistic alleles . The requisite va riation for localized suppression of recombination has been found to be abundant in the only model system (D. melanogaster) in which it has been systematically sought. But are sexually antagonistic alleles also common enough to drive the genetic ch ain reaction? Figure 2. The sexually antagonistic genes chain reaction model leads to the gradual elimination of recombination between the X and Y chromosomes. (a) A primitive Y chromosome differentiated from the X chromosome only by the G allele at a maledetermining locus. The G locus is placed near the end of the chromosome to simplify the diagram. (b) The presence of the G allele creates a hot spot where male-benefit sexually antagonistic alleles accumulate more readily. Mutation produces malebenefit sexually antagon istic alleles at loci located throughout the chromosome. (c) A male-benefit sexually antagonistic allele (aSA ) accumulates within the hot spot, selecting for reduced recombination in the chromosomal region (G-a SA ) ' (d) Evolution of suppressed recombination in the shaded interval G-a SA causes linkage to the G allele to become tighter at downstream loci, extending the length of the hot spot . (e) An additional male-benefit sexually antagonistic allele (bSA ) accumulates in the newly generated hot spot region, and this allele selects for reduced recombination in the region a sA-b sA' (f) Evolution of suppressed recombination in the region a sA-b sA further extends the hot spot, continuing a chain reaction in which the accumulation Recent experiments on sexually of sexually antagonistic alleles reduces recombination and reduced recombination antagonistic genes enhances the accumulation of sexually antagonistic alleles. To directly test whether sexuaIIy antagonistic aIIele s would accumucomplex adaptation that, for ex- suppression of recombination has late in the vicinity of a major sex ample, requires the evolution of sex- been shown to be common in Droso- determin ing gene on a p rimitive Y specific regulatory sequences (e.g., phila melanogaster (Brooks and chromosome, I used sex-specific arthose that bind sex hormones) near Marks 1986, Charlesworth and tificial selection to convert an autothe sexually antagonistic aIIele. It Charlesworth 1985, Chinicci 1971). somal dominant eye color aIIele into therefore is expected to evolve For example, after 33 generations of a new, synthetic female-determining slowly. By contrast, localized sup- selecting for reduced recombination aIIele (Sfd aIIele, Figure 3; Rice 1992): pression of recombination can evolve within a 15.4-centimorgan region of Any fly heterozygous for th is domifar more rapidly, at least when the the X chromosome of D. melan- nant allele (Sfd/sfd) was female (berequisite genetic variation (i.e., ogaster, Chinicci (1971) found that cause males of this genotype were genes that suppress recombination the recombination rate in this region remo ved from the breeding populaalong specific chromosomal regions) was reduced so that the map distance tion)' and an y fly homozygous for is common in the gene pool. became only 8.5 cM (although the the recessive aIIelomorph (sfd/sfd) Although little is known about the physical distance of this region along was male (because females of this molecular mechanisms that determine the chromosome remained the same). genotype were removed from the the rate of recombination within local- Genetic analysis demonstrated that breeding population). This simple ized chromosomal regions of meta- one or more genes on each of the selection protocol converted a popuzoans, genetic variation for localized major chromosomes contributed to lation with XY -male/XX-female 334 BioScience Vol. 46 No.5 chromosomal sex determination into one with genic sex determination: When sfd/sfd (males) are crossed with Sfd/sfd (females), the progeny retained to breed in the next generation are 1/2 sfd/sfd sons and 1/2 Sfd/sfd daughters. To make a macroevolutionary phenomenon (sexually antagonistic allele accumulation near a new female-determining gene) occur on a microcvolutionary time scale (i.e., within one National Science Foundation [NSF] grant period), the experiment was enhanced in two ways. First, two Sfd genes on separate chromosomes were used simultaneously, doubling experimental power because sexually antagonistic genes could accumulate in two hot spots simultaneously, instead of in just one. Second, the two Sfd genes were placed in the euchromatin near the centromere, an area where genes are approximately 7.5 times as dense per map unit than at typical chromosomal locations. Consequently, ex perimental power was further increased by a factor of 7.5, so that the accumulation of female-benefit sexually antagonistic genes in this 29generation experiment was comparable with what would be expected in a 29 x 2 x 7.5 = 435-generation experiment with a single gender-determining gene in a typical chromosomal region. In the replicated control populations, the same Sfd alleles and protocols were used, but the gender of the homozygotes and heterozygotes was switched for each generation. In these controls, sexually antagonistic female -benefit alleles should not accumulate beyond the normal level at a typical autosomal region. After 29 generations the two chromosomal regions containing the new female-determining alleles (and the chromosomal regions tightly linked to these alleles) were put into males, using standard genetic crossing procedures, and the lifetime fitness of these males was measured. Fitness of males carrying the chromosomal regions adjacent to the two Sfd alleles was reduced by more than 50% relative to males receiving the same regions from the control populations. Most of the reduction in fitness was attributable to mating performance rather than survivorship. This result supports the May 1996 Adults Females Males Figure 3. Artificial selection protocol used to make an arbitrary dom inant allele (D) on an autosome act as a synthetic female-determining allele (S(d). Heterozygous females (Dd) are crossed to homozygous recessive males (dd) , producing both homozygous and heterozygou s offspring. Only heterozygous daughters and homozygous sons are retained and bred each generation. idea that sexually antagonistic alleles are common in the gene pool of D. melanogaster and that a sex-determining G allele produces a hot spot for the accumulation of sexually antagonistic alleles that are deleterious to the homogametic sex . These experiments did not prove that the genes responsible for reducing male fitness accumulated because they increased fitness in females . It is possible, although less likely, that these male-detriment genes accumulated via genetic drift or genetic hitchhiking (see below) because of their tight linkage to the femaledetermining genes. In either case, these experiments confirmed that the G allele produces a hot spot for the accumulation of alleles deleterious to the homogametic sex. Moreover, the unexpectedly large magnitude of the reduction in male fitness suggests that a major battle of the sexes may be going on at the level of the genome. Each sex may be restricted in its adaptive evolution because both sexes must share a common gene pool and alleles beneficial to one sex are sometimes maladaptive to the other (or are tightly linked to such genes) . The two lines of research with Drosophila (i.e., genetic variation for sexually antagonistic genes and localized suppression of recombination) provide direct experimental evidence for the operation of the genetic chain reaction lead ing to the evolutionary elimination of recombination between the X and Y chromosomes. But what are the consequences of a lack of recombining on the Y? Theory for the degeneration of a nonrecombining Y Once the Y chromosome stops recombining with the X chromosome, which continues to recombine in XX females, it becomes a clonal (i.e., asexual) component in an otherwise sexually recombining genome . Nonrecombining Y-linked genes degenerate, whereas recombining Xlinked genes persist over geological time. In an interesting parallel, it 335 has been repeatedly observed that lineages of asexual multicellular species are almost always short lived on a geological time scale, whereas lineages of their sexually recombining relatives persist (Bell 1982). Much of the theory developed for the adaptive significance of recombination in sexually recombining versus asexual organisms can be applied to the X and Y chromosomes. In both contexts, the major question is: Why should the absence of recombination ca use lineages of organisms (chromosomes) to go extinct (deteriorate)? In the case of the Y chromosome, degeneration via Muller's shelteredlethal model (Muller 1918, 1932) was the first proposed explanation. This model posits that because the Y is permanently heterozygous, lossof-function mutations (which the model assumes are completely recessive) on the Y chromosome would virtually never be expressed. Hence, natural selection would not prevent loss-of-function mutations from accumulating and gradually leading to deterioration of the Y. However, Fisher's (1935) quantification of Muller's sheltered-lethal model showed that in large populations, before the Y chromosome has deteriorated, there are sufficient Xlinked loss-of-function mutations that are allelic to those on the Y chromosome to prevent complete sheltering. As a result, loss-of-function mutations cannot accumulate on the Y chromosome. Studies at the whole organism level by Kitagawa (1967) and others (summarized in Simmons and Crow 1977) were even more damaging to Muller's sheltered-lethal hypothesis because they showed that so-called recessive lethal mutations actually reduce fitness by a small percentage when heterozygous. Subcellular level studies of enzyme flux rates and pool sizes of intermediate substrates, carried out with heterozygotes for lossof-function alleles (reviewed in Kaeser and Burns 1981), also indicate that sheltering is incomplete. Lack of complete recessiveness is likely to prevent loss-of-function alleles from accumulating on the Y chromosome. If sheltering ofY-linked mutations cannot explain deterioration of the Y chromosome, then why does it decay once recombination is halted? Popu- 336 lation genetic theory predicts decay via a variety of mechanisms, most of which only operate in nonrecombining genomes or chromosomes. These include sampling drift, genetic hitchhiking, background-trapping, Muller's ratchet, and mutational overload. These can be descri bed most easily in their original context of a sexual species that splits into sexual and asexual lineages, so I describe these decay mechanisms in the more general context of asexual organisms. The simplest form of decay is due to sampling drift (random genetic drift) and is expected to occur in both sexual and asexual lineages. To understand this form of genetic decay, consider a new deleterious mutation in a finite diploid population of size N. Natural selection is a deterministic process that acts to decrease the frequency of the new mutation. Sampling error will randomly increase or decrease its frequency each generation. When the change in frequency due to selection is small relative to that due to sampling error, natural selection is overwhelmed by sampling error, and the new deleterious mutation can become fixed in the gene pool by chance accumulation. Using diffusion analysis originally developed in physics and chemistry, population genetic theory (reviewed in Crow and Kimura 1970; see Wright 1931 for an alternative approach) predicts that deleterious mutations will accumulate despite selection whenever s < 1/(4N ), in which s is the decrement to fitness associated with a mutation in the homozygous state and N; is the effective population size (approximately equal to the number of breeding adults in a population). Even when the population is large, mutations of small effect (i.e., those with small s values) are likely to accumulate, and therefore all finite populations are chronically decaying. Consequently, finite populations must perpetually evolve to counter such decay. The other decay mechanisms operate only in asexual lineages and are a consequence of variation in the genetic background of mildly deleterious mutations (i.e., mutations reducing fitness by a few percent or less when in the heterozygous state). Most new mu- tations fall into this category (Crow and Simmons 1983). To introduce these additional decay mechanisms, consider the fate over thousands of generations of a hypothetical, new population founded by a single asexual female-the progenitor. (Note that the hypothetical bottlenecked population is a simplification used to illustrate the mechanisms by which asexual populations decay over time. The underlying principles, however, are completely general because all individuals in any asexual population can trace their ancestry back to a single common ancestor or progenitor.) We can describe the population at any point in time by plotting the distribution of individuals carrying 0, 1, 2, ...,n mildly deleterious mutations. Only mutations that are mildly deleterious in the heterozygous state (i.e., that reduce fitness by no more than a few percent) are considered, because natural selection rapidly removes all mutations that are dominant and highly deleterious. Because the population is bottlenecked down to one individual, the original population is composed entirely of a single individual with P deleterious mutations (P, which represents the progenitor class, is more than because the original female was unlikely to be mutation-free; Figure 4a). In each subsequent generation, three processes affect the distribution of mutations per individual. First, mutation adds an average ofUD new mildly deleterious mutations per genome (top arrow, Figure 4b). Mutation studies on multicellular organisms (reviewed in Crow 1993) are still inconclusive but indicate that UD is probably greater than one per diploid genome, and perhaps five or more. Mutation also adds an average of UB new beneficial mutations, although such mutations appear to be rare (i.e., UB « U D ) . Second, as new mutations begin to accumulate and produce genetic variation' natural selection reduces the mean number of deleterious mutations in each generation because individuals with fewer deleterious mutations tend to produce more surviving offspring (middle arrow, Figure 4b). Third, sampling error either increases or decreases the number of individuals in each mutational class randomly (bottom, double-headed arrow, Figure 4b). Initially there is no variation in ° BioScience Vol. 46 No.5 a C a tally for the P + 1 class in the next gen eration (Figure 4d). Some of the recruits to thi s class will be derived Natur al selection % from progen y from the P class that % received one new mildly deleterious Sampling drift mutation, and the rest will be produced fr om the P + 1 cla ss itself. oo' 0 p o Because at steady state th e number Number of deleterious Numberof deleterious mutations in the P + 1 class does not change mutations from generation to generation, the P + 1 class is generated only in part from its own reproduction , with the remainder being genera ted from mutated recruits from the P class. More generally, all mutational classes except the P class are not selfP P+1 sustaining because some recruits in Surviving Offspring the next generation come from a Number of deleterious cascading down of newly mutated mutations offspring from less mutated classes Figure 4. The distribution of mutational classes of a hypothetical pop ulation (Figure Sa). founded by a single female. (a) The founding female has a genotype with P mildly This simple characteristic of asexual deleterious muta tion s (the progenitor class). (b) During each generation, mutation populations has an important implicaadds an average of U o new mildly deleterious mutation s (top arrow ). At the same tion: Barring new beneficial mutations, time, selection removes such mutations (middle arrow ) and sampling error ran- individuals in all but the P class give domly adds or removes them (botto m double-headed arrow ). Mutation also add s beneficial mutations at a very low rate of UB « Uo' (c) Recurrent mutation and rise to lines of descent that are not selfselection ultimatel y lead to an equilibrium distribution of mutati ons per genome. sustaining and are th erefore declining Sampling erro r chan ges the realized form of this distribution each generat ion in a toward eventual extinction. Only indistochastic fashion . When the genome-wide murat ion rat e (U Il ) is large, only a small vidu als in the P class produce lin es of fraction of individual s remain in the progenitor class (P). (d) At equilibrium, only descent that persist over geological the progenitor class is self-sustai ning becau se recruitment for the replacement of time . Because individuals flow unidirecall other classes (e.g., P + 1) is made up in part by newly mutated offsprin g derived tionall y from less to more mutated from less mutated classes (stippled portion of P + 1 surviving offspring). classes, the population is sa id to be geneticall y polarized, which guarfitness, natural selection cannot op- into two unequ al parts: a tiny, self - antees that all individuals are ultierate, and deleterious mutations ac- perpetuating progenitor class, and mately derived from ancestors who cumu late (Figure 4a) . As mutations what is essentiall y the living dead resided in the P class. An asexu al accrue, heritable variation in fitness (i.e., the remaining majority of the population therefore can be d ichotoincreases, causing the strength of population, which produces lineages mized int o the progenitor (P) class natural selec tion to continually that are doomed to eventual extinc- and the living dead (Figure Sa) . build . Ultima tely, mutations stop tion). To appreciate the significance of accumulating when the strength of To understand why asexual pop u- genetic polarization, first consider a selection bui lds t o a point sufficient lations are genetically polarized, it new beneficial mutation with a minor to offset recurrent mutation, and a is necessary to carry out some simple, effect on fitne ss (Figure 5b). Such a characteristic equilibrium distribu- al beit tediou s, bookkeeping on the re- mutation has a high probability of bet ion is achieved (Figure 4c). The production of the P and P + 1 classes ing introduced into a genetic back exact shape of the steady-state distri- (Figure 4d) . Let N , and N p + ! be the ground carrying many mildly deleteribution of mutations va ries with the number of individuals in the P and ous mutation s. As a result, the form of multilocus select io n, but P + 1 classes re spectively . To persist, beneficial mutation is trapped in a unless UD is small, onl y a small the P class must produce more than genetic lineag e that is doom ed to fract ion of individuals remain in the N , surviving offspring because some eventual ext inction, and it cannot lea st mutated progenitor (P) class, (or most, when UD is large ) offspring persist in the population (background a nd mo st individuals carry many are likel y to carry one or more new trapping; Manning and Thompson mildly deleterious mutations . mildly del eterious mutations. If we 1984, Peck 1994, Rice 198 7b ). To When sexual recombination is ab- temporarily ignore new beneficial persist, the mutation mu st be introsent , the distribution of mutational mutations, all of the recruits to the P duced into the P class, a nd therefore clas ses becomes genetically polar- class are derived so lely from the most beneficial mutations a re lost ized . This polarization is respon- reproduction of thi s class (i.e., the P and progressive evolution is slowed sible for all of the additional decay class is self- sustaining) . considerably. In sexual populations, processes that operate in asexual The P + 1 cla ss and all other more background trapping is eliminate d populations . In short, genetic polar- mutated classes are, however, not because recombination freely moves ization splits an asexual population self-sustaining. To see why, consider new beneficial mutations among mu 100 May 1996 b New mutations 100 337 a eage. In this case, one or more mildly deleterious mutations are likely to " hitch a ride " with the new beneficial mutation as it accumulates in the popu lation. Therefore, progres% sive evolution at one locus is likely to be at th e expense of the accumulation of one or more mildly deleterio p' o p ous m utations at other loci (genetic Number of deleterious mutations Number of deleterious mutations hitch hiking decay; Manning and c Thompson 1984, Rice 1987b) . d % When the expected number of individuals in the P class is small (2) (fewer than approximately 100 individuals; Maynard Smith 1978), then 1 there is a nontrivial probability that all members of this class will fail to z leave descendants in any generation (Charlesworth 1978, Haigh 1978). If sexual recombination is present, the %1_(4_) eIC. P class is rapid ly regenerated and Surviving p offspring p Number01deleteriousmutations the loss of the P class is inconsequential (Figure Sci. But with asexual reproduction, reconstitution of the P Figure 5. Genetic polarization and its consequences for the genetic decay of an class is expected to take many genasexual population. (a) Genetic polarization is the unidirectional flow of offspring erations, because it requires a new from less to more mutated classes. As a consequence, only the progenitor class (P) beneficial mutation and DB is small. does not depend on recruitment from less muta ted classes for part of its reproduc- In the interim, loss of the P class tion. This dichotomizes the population into a small, self-sustaining progeni tor weakens selection on the remaining class, which gives rise to persistent lineages, and the non-self-sustaining living dead, classes so that the bal a nce between from which all lineages are marching toward extinction. (b) Background trapping: mutation and selection will move New beneficial mutations are inefficiently recruited into asexual pop ulations because most (dotted arrows) are trapped in the extinction-bound lineages of the the distribution to the right (Haigh living dead . Genetic hitchhiking occurs when a genome carrying a small number of 1978), making the P + 1 class the new mildly deleterious mutations receives a beneficial mutation of large effect and progenitor class (Figure Sc-2) . To thereby is converted into a new progenitor class (P'), with P' more than P. recover from this "turn of Muller's Progressive evolution at one locus is therefore at the expense of mutation accumu - ratchet" (Felsenstein 1974, Muller lation at one or more other loci. (c) Muller 's ratchet: (1) The P class is lost due to 1964) a reverse mutation, a compensampling error (arrow) . (2) Absence of the P class weakens selection on all satory mutation, or a new beneficial remaining classes, which moves the mean of the distr ibution to the right . (3) mutation must accumulate in the Regeneration of the P class via progressive evolution (a reverse mutation, a compensatory mutation, or an unrelated beneficial mutation) is slow due to population. Such progressive evo lubackground trapping, and in the interim the P + 1 class (best remaining class) is lost tion is slow due to background trap by sampling error (arrow) . (4) The best remaining class may be lost many times ping. Consequently, th e best remainbefore progressive evolution moves the distribution a single step back to the left. ing mutational class may be lost This asymmetry leads to a net accumulation of mildly deleterious mutations. (d) many times due to sampling error Mutational overload occurs in asexual organisms when the genome-wide mutation before the successful recruitment of rate (Do) is high relative to the fecundity of the progenitor class (R o p or the per capita a single reverse, compensatory, or net reproductive rate of the progenitor class). Model ing mutation as a Poisson new beneficial mutation. In this way, process, only a fraction e -uo of offspring from the P class are unrnutared and remain mildly deleterious mutations graduin this class. ally accumulate in an asexual lin eage due to samp ling error and getatio na l classes. Quantitative work organisms or chromosomes. netic p ol ar iza ti on. (Manning and Thompson 1984,Peck Next, consider a beneficia l m uta tion T he fina l decay mec hanism is mu1994) demonstra tes that background of large effect th at substantially in- tational overload (Figure Sd), which trapping can substantially slow the creases fitness (Figure 5b). In this occurs when the progenitor class does ra te of progressive evolution such situation, most beneficial mutations not produce enough unmutated surthat asexual populations recruit new are still likely to be lost due to viving offspring to replace itself. Per beneficial mutations tens to thou- backgro und trapping, but new ben- capita lifetime reproduction is gensands of times slower than sexually eficial mutations trapped in m uta- erally expressed as the net reproducrecombining popu lations. In sum- tional classes neighboring the P class tive rate (R p)' and this value for the mary, background trapping causes may cause the carrier individual to P class is denoted as R o p. If there progressive evo lution to be far slower exceed the fitness of the P class and were no new mutations, Ro p must be in asexual than sexual lineages of give rise to a new progenitor lin- 1.0 for the progenitor class to perProgenitor class I Living dead b Progenitor Living dead class % I 338 BioScience Vol. 46 No.5 sist. With R o P less than 1.0, a succes- some is also expected to accumulate sion of new' progenitor classes will beneficial mutations more rapidly than be lost recurrently. When new muta- the Y due to background trapping, tions occur, R o p must be increased which over time causes Y -linked alleabove unity to accommodate the loss les to become relatively inferior. of surviving, but newly mutated, offspring to more mutated classes. Recent experiments on Y If we model new mutations as a chromosome degeneration Poisson process, then some simple calculations demonstrate that a frac- We cannot hope to directly observe tion (e- UD) of the surviving offspring the genetic decay of the Y chromoare likely to receive no new mildly some in nature owing to its slow deleterious mutations. In this case, speed of operation. Two alternative approaches are possible. One apRo,p must be more than or equal to eUD for the progenitor class to persist. proach, taken by the Steinemanns As U D increases, the requisite mini- and their collaborators, is to look for mal number of surviving offspring the footprints of the decay process (measured by RO,P(min) on a per capita via a molecular dissection of the Y basis) increases exponentially. For chromosome in those species whose example, if an average of five new Y chromosome is presently degenermildly deleterious mutations were ating. A second approach, used by produced during the production of a my laboratory, uses population gediploid asexual egg, then RO,P(min) netic theory to solve for experimenwould be more than 148, which is tal conditions under which the Y beyond the reproductive capacity of chromosome will genetically decay many vertebrates. Asexual reproduc- so rapidly that degeneration can be tion in such species would lead to observed directly in the la boratory continual decay in the fitness of the on a microevolutionary time scale. population and, eventually, to its To create a synthetic Y chromoextinction. When sexual recombina- some whose decay could be meation occurs, however, the P class is sured directly, I used genetic markproduced not by its own reproduc- ers and artificial selection protocols tion but via reassortment of muta- similar to those described a bove for tions from the population as a whole. the sexually antagonistic genes exIn this case, epistatic selection and periments to endow ordinary autopositive assortative mating for fit- somes with the major characteristics ness can greatl y reduce the requisite of the Y chromosome (i.e., they net reproductive rate for the P class, lacked genetic recombination and and its persistence is possible even were passed on exclusively from fawhen U D is large. ther to son; Rice 1994). The experiThe importance of mutational ments were designed so that the overload depends on the magnitude major mechanism by which the Y of U D , which is determined primarily chromosome was expected to decay by genome size of coding genes. was Muller's ratchet. When the nonrecombining genome Theory (Haigh 1978) indicated (or genomic component) is small, as that the speed of the ratchet depends in the case of bacteria, viruses, or on the generation time of the organsmall Y chromosomes, this decay ism, the chromosome-wide mutation process is likely to be unimportant. rate to mildly deleterious mutations, But when the nonrecombining ge- the effective population size (Ne ) , nome or genomic subunit is large, and the mean decrement to fitness of mutational overload may be the pre- individual, mildly deleterious mutadominant decay process. tions. The rate of decay increases Summarizing with respect to the X with decreasing generation time and and Y chromosomes, a recombining X effective population size and with chromosome is expected to decay increasing mutation rate and mean slowly due to sampling drift, but a clonal decrement of a mutation on fitness. Y chromosome is expected to decay far The generation time was made more rapidly, due to the additional small by using the fast cycling speoperation of Muller's ratchet, genetic cies D. melanogaster, which has a hitchhiking, and possibly mutational generation time of less than two overload. In addition, the X chromo- weeks. I made the mutation rate of May 1996 the synthetic Y chromosome high by virtue of its large size; sex-specific artificial selection on visible marker genes was used so that both of the major autosomes of D. melanogaster, collectively containing 80% of the genome, co segregated like a giant nascent Y chromosome. Effective population size was made small by permitting only 32 synthetic Y chromosomes to be transmitted across each generation. The mean effect of a mildly deleterious mutation was not experimentally manipulated, but this value was presumed to be relatively small based on prior studies (reviewed in Simmons and Crow 1977). There were a total of five synthetic Y chromosome lines, each with a paired control in which recombination was permitted and the autosomes were not restricted to males. Computer simulations spanning the range of likely values for the effects of mildly deleterious mutations and the chromosome-wide mutation rate indicated that the above combination of parameter values should speed the ratchet process to the point where it could be observed directly over a period of only one or two years and that recombination should rescue the controls from most of this decay. Deterioration of the synthetic Y chromosome, and the recombining chromosomes in the controls, was assayed not by directly sequencing specific genes (because thousands of gene loci were contained on the synthetic Y chromosome, this approach was not feasible) but by using genetic crosses to place the chromosomes into a standardized genetic background and measuring fitness. After only 35 generations, there was a strong consensus among the synthetic Y lines demonstrating statistically significant decay in fitness relative to their recombining controls (p < 0.0006). This decay was corroborated by a more extensive set of fitness assays at generations 48, 49, and 50. 1 Such rapid decay in the Y is not expected in natural populations, with their smaller Y chromosomes and larger population sizes; nevertheless, these experiments demonstrate that decay occurs in response to lack of recombination. Fac- -w. R. Rice, 1994, unpublished data. 339 tor s other th an M uller's ratch et may ha ve contributed to th e o bse rve d decay, but M uller's ratchet is likel y to have been th e major co ntr ibuto r. Molecular analysis of a degenerating Y chromosome The ex periments just descr ib ed demonstrate decay of a Y chro mos ome in respo nse to th e lack of recombinaMale Y X.., Xnew ti on, but th ey say nothing a bout th e m ol ecula r mech ani sm s by w hic h Figure 6. A schematic dra wing of th e sex decay ac tua lly occ urs . This aspect of chro mosomes of Drosop hila m iranda. Y chromoso me evo lution ha s been T he X XY/X XXX sex det erminati on sysaddressed by th e Ste ine ma nn gro up tem develop ed from an ances tra l XY/XX (Steine ma nn 19 82, Ste ine mann and system du e to a fusio n (i.e., t ran slocaStein em ann 1990, 1991, 19 92, tion ) of one member of a hom olo gou s Steinemann et al. 1993). Steinemann pair o f autoso mes to th e orig ina l Y ch roet a l. (1 99 3 ) used Dr osophila mosome, w hereas th e ot her hom ol omiranda as a model sys te m beca use go us a utoso me re mai ne d free . T he ex a recent (less th an 5 million yea rs tant Y ch ro moso me has an o ld sectio n ago) tr an slocation fused o ne mem- (Yold)' co nsis ting of th e origina l Y chromosome, and a new section (Y ), con ber of a pai r of hom ol ogou s auto - tain in g the translocated a u tos;;'~e . so mes onto thi s species' ex ist ing, alrea dy degenerat ed Y ch romosome, producin g an X o !dX new Yco mpos ite . sex teri zed , th er e are approxima tely 5 0 chromosom e system (Figure 6). Dur- differ ent kinds of T E, with an av ering meiosis th e ori gin al portion of age of approximately 50 members of the Y chro moso me (Yo!) pairs with each T E per genome (Lindsley and the o rigina l X chro moso me (X old)' Zimm 1992 ). Insertion of a T E into a while th e newl y acqui re d, previou sly new locati on frequentl y disrupts geautosom al portion of th e Y chromo- netic functio n in th e chromos oma l so me (Y ) pair s w it h it s fr ee, region neighb oring th e inserti on site, nontransl~~ated hom ol og (X ne) . Be- and an unu suall y high acc um ulation cau se of t he tran slocation , a lar ge of T Es on a Y ch ro mosom e could chro moso ma l seg me nt conta ining th erefor e potentiall y lead to its gethousand s of genes was introduced netic in acti vation. onto th e Y a nd became a new, Using in sit u hybridi zation technonrecombining differ ential segment niques (i.e., microscopic examina(bec au se th ere is no intrachrom- tion of ch romosomes ex pose d to ra osom al r ecombin at ion in m al e dioactively labeled pieces of DNA D rosophila) . Becau se th e transloca- th at spec ifica lly attach to th e sites tion is recent on a geo log ical time where ex ta nt T Es reside) to map th e sca le, degen eration of Y is ex- location of two fam ilies of T Es, it pected to be presently oc2~Wrring at was fo und th at both T E t ypes had hu nd red s of loci sim ulta neo us ly; accumu late d on Ynewto a far greate r th er efore, a mol ecul ar ana lysis of degre e th an on the ot he r chromoth ese genes sho uld catc h the deterio- so mes . Each TE had a den sit y at ration pro cess in pro gr ess. least fivefold high er o n th e nonreThe Ste inemanns ' lab oratory has combining Y as co mpare d with focu sed on deterioration caused by the X new or c~~parabl e autosomal tr ansposable elements (TEs ), which are regions of th e geno me . small seque nces of DNA, man y of In add iti on to th is wh ole chromo which sprea d lik e par asite s throu gh - so me app ro ach , Steine ma nn et a l. out th e geno me via a uto no mo us rep- look ed in det ail at a sma ll clu st er of licatio n and inse rt ion into new chro - genes on th e new X and Y chro mo mo som al lo cati on s. M o st TEs so mes (Xn.)Ynew-linked genes) th at accu mulate to on ly a limited exte nt co de fo r a su bse t of th e larval cuticle within th e genome . For exa mple, in pr otein s (LC P 1-4 genes). A comD. m elan ogast er, the Drosophila pa rison of hom olo gou s 7-ki lo base species wh ose T Es ar e best ch arac- segments co nta ining either X new- or 340 Ynew-linked LCP genes revealed a stri king di ffer ence in express ion of th ese gen es. As pr ed ict ed from th e chromoso me-wi de an al ysis, th e Y seq uence (co nta ini ng th e LCP 1:'4' genes) was ridd led with T Es that were ab sent o n th e homologou s X link ed sequenc e. Hypothesizing th;t th ese T Es may have silenced th e Y -linke d a lleles, Steinema nn et a l. e~~wmi ned th e ex p ress io n of th e X and Ynew-linke d genes and found th~t, indee d, all of th e Ynew-linked a lleles we re either untranslat ed (LCP-1,2,4 ) or we re translated at a low level (LCP-3). By tr an sforming th e LCP-2 and LCP-3 genes from Y into D. melano gast er, Steineman~~t al. further showed th at removal of th e TE s from the up st ream regul at ory regions of th e LCP genes rest or ed gene ex pr ession . Less direct evidence indica ted th at T Es ma y have played a ro le in silencing th e ot her two LCP genes. Ano ther Y-linked LCP gene (LCP-5, located o uts ide th e stud ied gene clu ster ) was fo und to be active on Y . In ne;ddition to th e X versus Y difference in TE s, th e Y-link ed sequ ence s of th e LCP gen es were fo und to hav e diverged from th eir hom ologo us X-linked seq uences, wit h three o r mor e amino ac id substi tutio ns at eac h locu s, ma ny of whic h wo uld have alte re d th e charge, hydrophobicity, or pol arity of th e diverged am ino acids (when still ex presse d on Ynewl. Th e Ynew-linked LCP-4 gene also had accumulated a fra meshift mutation, and th e Y-link ed LCP-2 gene had lost its methionine tran siti onal sta rt site . What do th ese results reveal a bo ut th e mechan ism (s) by which th e Y chromosome degenerates? First, th ey indicate th at TEs may play an im po rtant rolein its degeneration. But why sho uld TEs be far mor e frequent on the Ychromoso mes ?W ha t is the fun cti on al significan ce of such a differ ence? Selectively neutral T Es would be expected to accumulate on th e Y chromosome for two reasons. First, consider a TE that was neutr al because it inserted into a region of no nfunctio nal DNA on th e nonrecombin ing Y chromosome. Lac k of reco mb ina tio n pr events un eq ua l hom ol ogous meiotic exc hange (a pr ocess where by two chro moso mes w ith a T E at th e sa me chro mosomal site rec ombine in such a way th at BioScience Vol. 46 No .5 one chromosome ends up with two tandem copies of the TE and the other with zero copies) from physically deleting TEs. This reduced removal rate alone could cause neutral TEs to accumulate on the Y chromosome. Second, if TE-induced loss-of-function mutations were completely recessive, they might also accumulate on the Y chromosome via Muller's sheltered-lethal model, although the operation of this model seems unlikely. Non-neutral TEs would accumulate on a nonrecombining Y chromosome for three reasons. First, they would accumulate because of mutation accumulation processes that operate only in the absence of recombination (Muller's ratchet, genetic hitchhiking, and mutational overload). Second, TEs would accumulate because lack of recombination will prevent ectopic exchange among homologous TEs at different (nonhomologous) genomic locations (ectopic exchange model; reviewed in Charlesworth et al. 1994). This model is based on the theoretical result that selection can stop the open-ended accumulation of parasitic TEs only when the strength of selection increases in a steeper-than-linear fashion as TEs accumulate in the genome. The importance of ectopic exchange is that it produces aneuploid gametes (i.e., gametes with unbalanced genomes containing lethal deletions or duplications of a chromosomal region) and is therefore thought to provide a strong selection against the openended accumulation ofTEs (reviewed in Charlesworth et al. 1994). Third, and counterintuitively, non-neutral TEs 'may accumulate because of positive Darwinian selection. Silencing would be selectively advantageous if the Y-linked genes had diverged from their X-linked alleles in a maladaptive way, either by mildly deleterious mutations accumulating and/or by the X-linked alleles evolving to a superior form more rapidly than their Y-linked homologs (due to background trapping on the Y chromosome). Either way, once the Y chromosome becomes fixed for alleles that are sufficiently less fit than their X-linked homologs (e.g., as might occur when the gene product of the inferior Ylinked allele competes for substrate May 1996 with the superior product from the X-linked allele), there will be active selection for any mutation that silences the Y-linked allele. TE insertion in the regulatory region of a gene is a simple way to evolve an inactive allele (or one with reduced activity). Thus the TEs that have accumulated on the Y chromosome and inactivated two or perhaps all of the Y-linked LCP genes may represent a case of natural selection turning off less fit Y-linked alleles. Another large group of experimental studies has shed additional light on the evolution of the Y chromosome. For more than a decade, molecular biologists have raced to identify and characterize the testisdetermining factor of mammals (Tdf-studied primarily in humans, mice, and several marsupials). Several candidate genes have been sequenced, and comparisons among mammalian species have provided information on the rate of divergence among Y-linked genes (see, for review, Hurst 1994, Wachtel 1994). These sequencing studies have produced two key results. First, at least two Y-linked genes (Sry [sex-determining region on the Y, now established to beTdf] and ZfY[a Y-linkedzincfinger protein locus that codes for a DNA-binding regulatory protein]) are evolving far more rapidly than typical autosomal genes. Hurst (1994) argues that selection, more specifically intragenomic conflict (i.e., antagonistic coevolution between two or more loci within the genome of the same species), rather than sampling drift or normal positive Darwinian selection, is driving this fast pace of evolution of Ylinked genes. Second, the sequencing studies demonstrate that the Y chromosome is continuing to degenerate, as evidenced by certain Y-linked genes being nonfunctional in one species and functional in another despite the small number of functional genes remaining on the mammalian Y chromosome (e.g., the Ylinked Sts [steroid sulfatase] gene is functional in mice, whereas it is a nonfunctional pseudogene in humans). The significance of this observation is that Muller's ratchet operates slowly if at all when the chromosome-wide mutation rate to mildly deleterious mutations is small, as it would be when few functional genes remain on the Y. So the observed continued decay of the Y even when it is almost completely degenerated suggests the operation of some other decay process, possibly genetic hitchhiking, active selection to silence inferior Y-linked alleles, or intragenomic conflict. Clonal chromosomes versus clonal genomes Although an understanding of how the Y chromosome evolves from an ordinary autosome is interesting in the context of chromosome evolution, it also has more widespread application as a model system for the study of the adaptive significance of sexual recombination. Phylogenetic studies indicate that most asexual lineages ultimately go extinct while their sexual relatives persist (Bell 1982). Similarly, once the Y chromosome stops recombining with the X chromosome it also becomes doomed to extinction (i.e., all, or virtually all, genetic activity is eliminated). The similar fates of clonally propagated species and chromosomes suggests that sexual recombination is a prerequisite for persistence over long periods of evolutionary time. Many hypotheses have been put forward for the adaptive significance of recombination (see,for review, Michod and Levin 1988). The fact that nonrecombining chromosomes as well as organisms are doomed to extinction suggests that any general theory for the adaptive significance of sex should apply to both of these entities. Therefore, many of the ecological hypotheses for the adaptive significance of sex, such as parasite-host coevolution (Jainike 1978, Levin 1975) or resource partitioning among siblings (Williams 1975), are unlikely to be complete explanations for the adaptive significance of sex because they do not explain the extinction of nonrecombining chromosomes. There are, however, important differences between clonally propagated Y chromosomes and species. First, the Y chromosome is permanently heterozygous, with redundant, homologous genes being present on the X chromosome. This genetic redundancy may reduce the fitness consequences of new deleterious mutations on the Y chro341 mosome, and hence speed genetic decay, because partially recessive deleterious mutations rarely become homozygous. This difference between nonrecombining Y chromosomes and the genomes of organisms applies only to haploid and not diploid asexual organisms. Second, as loss-of-function genes accumulate on the Y chromosome, the evolution of dosage compensation counterbalances the gene product imbalance that would otherwise arise. Dosage compensation ameliorates the deleterious effects of degeneration of the Y chromosome, which should permit it to degenerate faster than genes in an asexual organism (Charlesworth 1978). Nonetheless, Y chromosomes have degenerated in groups like birds, in which dosage compensation has not been observed. The impact of genetic redundancy and dosage compensation on genetic decay can be measured directly in diploid organisms, such as D. melanogaster, that have a nonrecombining autosome. For example, comparison can be made with the small, nonrecombining so-called dot autosome (chromosome 4) of D. melanogaster, which is not permanently heterozygous or dosage compensated. Interestingly, chromosome 4 has maintained substantial genetic activity for millions of years, although there is indirect evidence that it may be decaying (Hochman 1976). Until more detailed molecular work is done on chromosome 4, however, we will not know the full extent of its deterioration. The third major difference between an asexual multicellular species and a Y chromosome is that the genome-wide deleterious mutation rate of a whole organism's genome is far larger than the chromosomewide rate of the Y chromosome. It is the genome- or chromosome-wide mutation rate, not the per-nucleotide rate, that determines the susceptibility of an asexual genome to the additional decay mechanisms (Muller's ratchet, genetic hitchhiking, and mutational overload) that operate only in nonrecombining genomes or genome segments. Consequently, the relative importance of the various decay agents may differ between the two. Despite the unique properties of the Y chromosome, it constitutes a 342 simple and experimentally tractable model system for deciphering the adaptive significance of sexual recombination. The experiments with the Y chromosome have focused on the qualitative question concerning the importance of the presence or absence of recombination. To make further progress we need to address the quantitative question of the adaptive significance of varying levels of recombination. In this context future progress will likely be achieved by experiments using different autosomal regions that have varying levels of recombination. Acknowledgments I thank Suzanne Kohin and Brett Holland for comments on the original manuscript. My research was supported by NSF grants BSR-8996268, DEB9118893, and DEG-9307735. References cited Bell G. 1982. The masterpiece of nature. Berkeley (CA): University of California Press. Bischoff R], Gould ]L, Rubenstein DI. 1985. Tail size and female choice in the guppy. Behavioral Ecology and Sociobiology 17: 253-255. Brooks LD, Marks W. 1986. The organization of genetic variation for recombination in Drosophila melanogaster. Genetics 114: 525-547. Bull ]J. 1983. Evolution of sex determining mechanisms. Menlo Park (CA): Benjamin Cummings. Charlesworth B. 1978. A model for the evolution of Y chromosomes and dosage compensation. Proceedings of the National Academy of Sciences of the United States of America 75: 5618-5622. . 1991. The evolution of sex chromosomes. Science 251: 1030-1032. Charlesworth B, Charlesworth D. 1985. Genetic variation in recombination in Drosophila. I. Response to selection and preliminary genetic analysis. Heredity 54: 71-83. Charlesworth B, .Sniegowski P, Stephan W. 1994. The evolutionary dynamics of repetitive DNA in eukaryotes. Nature 371: 215-220. Chinnici ]P. 1971. Modification of recombination frequency in Drosophila. I. Selection for increased and decreased recombination. Genetics 69: 71-83. Crow]F. 1993. How much do we know about spontaneous human mutation rates? Environmental and Molecular Mutagenesis 21: 122-129. Crow ]F, Kimura M. 1970. Introduction to population genetics theory. New York: Harper and Row. Crow]F, Simmons M]. 1983. The mutational load in Drosophila. Pages 1-26 in Ashburner M, Carson HL, Thompson ]N, eds. The genetics and biology of Drosophila. Vol. 3c. New York: Academic Press. Endler ]A. 1980. Natural selection on color patterns in Poecilia reticulata. Evolution 34: 76-91. Felsenstein J. The evolutionary advantage of recombination. Genetics 78: 737-756. Fisher·RA. 1931. The evolution of dominance. Biological Review 6: 345-368. _ _. 1935. The sheltering of lethals. American Naturalist 69: 446-455. Haigh J. 1978. The accumulation of deleterious genes in a population-Muller's ratchet. Theoretical Population Biology 14: 251-267. Haldane ]BS. 1922. Sex ratio and the unisexual sterility of hybrid animals. Journal of Genetics 12: 101-109. Hochman B. 1976. The fourth chromosome of Drosophila melanogaster. Pages 903-925 in Ashburner M, Carson HL, Thompson ]N, eds. The genetics and biology of Drosophila. Vol. 1b. New York: Academic Press. Hurst LD. 1994. Embryonic growth and the evolution of the mammalian Y chromosome II. Suppression of Y-linked growth factors may explain escape from X-inactivation and rapid evolution of Sry. Heredity 73: 233-242. ]ainike ]. 1978. An hypothesis to account for the maintenance of sex within populations. Evolutionary Theory 3: 191-194. Kaeser H, Burns ]A. 1981. The molecular basis of dominance. Genetics 97: 639-666. Kirpichnikov VS. 1981. Genetic basis of fish selection. New York: Springer-Verlag. Kitagawa O. 1967. Interactions in fitness between lethal genes in heterozygous condition in Drosophila melanogaster. Genetics 57: 809-820. Levin DA. 1975. Pest pressure and recombination systems in plants. American Naturalist 109: 437-451. Lindsley DL, Zimm GG. 1992. The genome of Drosophila melanogaster. San Diego (CA): Academic Press. Manning ]T, Thompson DJ. 1984. Muller's ratchet accumulation of favorable mutations. Acta Biotheoretica 33: 219-225. Maynard Smith]. 1978. The evolution of sex. Cambridge (UK): Cambridge University Press. Michod RE, Levin BR. 1988. The evolution of sex: an examination of current ideas. Sunderland (MA): Sinauer Associates. Mitwoch W. 1967. Sex chromosomes. New York: Academic Press. Muller HJ. 1918. Genetic variability, twin hybrids and constant hybrids, in a case of balanced lethal factors. Genetics 3: 422-499. . 1932. The differentiation of the sex chromosomes of Drosophila into genetically active and inert regions. Zietschrift fur Induktive Abstammungs und Vererbungslehre 62: 316-365 . 1964. The relation of recombination to mutational advance. Mutation Research 1: 2-9. Nei M. 1969. Linkage modification and sex differences in recombination. Genetics 63: 681-699. . 1970. Accumulation of nonfunctional genes on sheltered chromosomes. American Naturalist 104: 311-322. Peck ]R. 1994. A ruby in the rubbish: benefi- BioScience Vol. 46 No.5 cial mut ati on s, deleteriou s mut ation s and t he evo lut ion o f sex . Ge netics 13 7: 597-606 . Rice WR. 1987a. The accumu lat ion of sexually antagon istic genes as a selective agent promoting the evolution of reduce d recomb ina tion between p r im it ive sex chromosomes. Evolution 41: 91 1-914. _ _ . 1987b. Genetic hitch-hiking and the evolution of reduced genetic activity of the Y sex chromoso me. Genetics 116: 161-167. _ __. 1992. Sexu ally antagonistic genes: experi me n ta l evi dence. Scie nce 256: 1436-1439. _ _' 1994. Degenerat ion of a nonrecombining chro mosome. Science 263 : 230-232. Simmons MJ, Crow JF. 1977. Mutations affecting fitness in Drosoph ila populations. Annual Reviews of Genetics 11: 49-78. Steinemann M. 1982. Multipl e sex chro mosomes in Dros oph ila miranda: a system to study the degeneration of a chro mosome. Chromosoma 86: 59- 76. Steinemann M, Steinemann S. 1990. Evolutionary changes in the organization of the major LCP gene cluster during sex chro mosome differentiation in the sibling species Drosophila persim ilis, D. pseudoobscura and D. mi randa. Chromosoma 99: 424-431. . 1991. Preferential Y chro mosome 10~on of TRIM, a novel transposable element of Drosophi la mi randa, obscura group . Chromoso ma 101: 169-1 79. _ _ . 1992. Degenerating Y chromo some of Drosophila miranda: a trap for retroposons. Proceedings of the National Academy of Sciences of the United States of America 89: 7591-7595. Steinemann M, Steinemann 5, Lottspe ich F. 1993. How Y chro mosomes become inert. Proceedings of the Na tiona l Academy of Sciences of the United States of America 90: 5737-5741. Wachtel 55. 1994. Molecular genetics of sex determination . San Diego (CA): Academic Press. Williams Gc. 1975. Sex and evolution. Princeton (NJ): Princeton University Press. Winge O. 1927. The location of eighteen genes in Lebistes reticulata. Journ al of Genetics 18: 1-4 3. Wright S. 1931. Evolution in Mendelian popu lations. Genetics 16: 97-159. New Edition Scientific Style and For m at The ( BE Ma nua l fo r Au thors, Editors, and Publish ers Sixth Edition Edward J. Hu th From reviews ofthe Sixth Edition: This detailed and authoritative manual is completely reorganized with coverage expanded to all sciences and with a new focus on general and scientific publication style and formats for science papers, journals, and books. 1994 782 pp. 47154-0 Hardback Available in bookstores or from CAMBRIDGE UNIVERSITY PRESS May 1996 $34 .95 40 West 20th Street. NY, NY 10011-4211. Call toll-free 800-872-7423. MasterCardNISA accepted. Prices subject to change. 343