All animals present individual differences, and as man can modify his domesticated birds by selecting the individuals which appear to him the most beautiful, so the habitual or even occasional preference by the female of the more attractive males would almost certainly lead to their modification; and such modification might in the course of time be augmented to almost any extent, compatible with the existence of the species (pp. 750-751).       Traits that improved mating success, he argued, also stood a better chance of being passed on to offspring, and this process could account for the evolution of new species.  Darwin discussed two mechanisms of sexual selection: ?Contests? between males over females, which favor ?weapons? such as elk antlers, and ?mate choice? by females, which favors male ?ornaments? such as peacock tails.  More recently, biologists have identified additional sexual-selection mechanisms including endurance rivalry, scramble competition, and sperm competition (Andersson, 1994).       What makes a trait attractive to potential mates?  Darwin didn?t know, but subsequent evolutionary theorists have suggested several possibilities (which can act simultaneously).  Traits may become attractive because they advertise health, fertility, vigor, longevity, parenting ability, optimal genetic distance, good genes, and/or simply the prospect of passing on attractiveness itself (Andersson, 1994).       How can an ornament, such as a peacock?s tail, advertise genetic quality or fitness?  If healthier birds tend to grow brighter feathers, then the brightness of feathers would indicate fitness (Fisher, 1915).  Moreover, the offspring of females who prefer brighter feathers would inherit the father?s genes for better fitness and the mother?s genes for preferring bright feathers.  Across generations, the increasing co-occurrence within individuals of the preference genes and the fitness genes would lead to a powerful positive-feedback process that could fuel the rapid evolution of brighter feathers?a process termed ?runaway sexual selection? (Fisher, 1930).       Why would healthier birds have brighter feathers or bigger tails?  One possibility is a mechanism called ?the handicap principle? (Zahavi, 1975).   A peacock?s tail takes considerable energy to grow, maintain, and display.  This cost could make it a reliable indicator of fitness, because only the fittest peacocks can afford the energy necessary to grow large and colorful tails.  As a result, peahens would evolve a preference for the extravagant extreme.  The handicap principle and several related mechanisms produce extravagant traits in theoretical models (Andersson, 1994; Hasson, 1989; Michod & Hasson, 1990)?even in monogamous species (Hooper & Miller, submitted).  Moreover, empirical work has shown that some sexually selected traits bear the three hallmarks of fitness indicators (Andersson, 1994): 1) they vary greatly in size, loudness, complexity, or other qualities acro! ss individuals; 2) that variance correlates with underlying fitness and condition; and 3) potential mates prefer the high-fitness extreme.       But this leads to another question?why don?t all peacocks have big, beautiful tails?  Tails vary greatly in size and complexity, and that variation is somewhat heritable.  However, in a group of peafowl, the one or two peacocks with the most elaborate tails sire virtually all the offspring (Petrie, Halliday, & Sanders, 1991).  Why don?t the genes for big tails proliferate and why don?t the genes for less elaborate tails disappear?  This question is called ?the paradox of the lek? (Kirkpatrick & Ryan, 1991)?a lek being the clearing in which male birds display their ornaments as females inspect and choose?not unlike a singles bar.  Recently, several investigators have suggested a common potential resolution of the ?lek paradox? (Houle & Kondrashov, 2002; Kotiaho, Simmons, & Tomkins, 2001; Michod & Hasson, 1990; Pomiankowski & Moller, 1995! ; Rowe & Houle, 1996).  This resolution, discussed subsequently, is at the heart of the explanatory and predictive power of our hypothesis regarding schizophrenia. The Lek Paradox Resolved       The resolution requires a distinction between ?good? and ?bad? genes.  ?Good genes? are those versions of genes (?alleles?) best suited to an animal?s current ecological niche, and to the rest of its species-typical genome.  Individuals with ?good genes? grow better bodies and brains, find more food, resist more parasites, avoid more predators, survive longer, and, thereby, leave more offspring.  However, to reproduce, they must make sperm or ova.  In that process, they must copy DNA, and DNA cannot be copied perfectly.  Copying errors produce new versions of genes that are (almost always) less well-suited to the niche.  These altered genes are called fitness-reducing mutations or ?bad genes.?  They reduce the chances that offspring will survive and reproduce.  In every generation, copying errors s! upply new ?bad genes.?  For example, the average human child has two to four new harmful mutations that neither parent had (Eyre-Walker & Keightley, 1999).  Selection immediately removes fatal mutations, and quickly removes very harmful mutations.  Mildly harmful mutations, however, can persist for many generations.  A mutation causing a 1% reduction in fitness will persist in the population for 100 generations, on average (Falconer, 1996).  The balance between mutation and selection leads to an equilibrium frequency of ?bad genes? in a population (Keller & Miller, in press). For example, the average human carries 500 to 2,000 old mutations inherited from his or her ancestors?mutations that have not yet been eliminated by selection (Fay, Wyckoff, & Wu, 2001; Sunyaev et al., 2001).  The number and type of ?bad genes? (referred to, in composite,  as ?mutation load?) varies across individuals and is responsible for most of the heritable variation in fitness (Houle & Kondrashov, 2002; Michod & H! asson, 1990; Rowe & Houle, 1996).       Mutation load reduces fitness and is the key to resolving the lek paradox.  In panel ?a? of Figure 1, we?ve modeled variation in fitness as a normal distribution with a mean of 50 and standard deviation of 10.  Panel ?b? shows a hypothetical relationship between fitness and the ultimate attractiveness of a sexually selected trait.  Panel ?c? shows the result of applying the function in ?b? to the distribution in ?a.?  For now, apply the figure to the attractiveness of peacock tails.  Ignore the dashed lines in panels ?b? and ?c? as well as the reference to schizophrenia in panel ?c.?   {Insert Figure 1 about here}       Imagine that we could pick out the peacock embryos lucky enough to have been conceived with very few ?bad genes? (i.e., a low mutation load) and therefore high fitness, say ?75? on the fitness scale in panels ?a? and ?b.?  These embryos have ?good genes? for precise cell migration, efficient feeding, parasite resistance, predator evasion, and any other process that can ultimately affect tail size.  Thus, embryos with ?good genes? for general fitness tend to develop into adult peacocks with very large and elaborate tails at about ?7? on the attractiveness scale.       However, most peacock embryos contain some ?bad genes? and end up with somewhat smaller, less elaborate tails.  A few peacock embryos at the low-fitness extreme of the distribution contain more than their share of bad genes.  Imagine we could pick out embryos with a fitness score of ?35.?  The ?bad genes? in these embryos are so numerous or so severe that they interfere with several of the hundreds of developmental processes that can affect tail size.  By impairing anything from embryonic cell migration to adult feather preening, they disrupt tail development or maintenance enough that these peacocks tend to grow small, dull tails, at less than ?1? on the attractiveness scale.       Thus, the tail?s sensitivity to fitness converts otherwise subtle variation in fitness into obvious variation in attractiveness.  But, if only those peacocks with the most attractive tails get to mate, why are there any offspring with unattractive tails in the next generation?  This is the lek paradox.       The answer may lie in the new ?bad genes? that arise during the formation of ova and (especially) sperm.  The risk of a copying error in any one gene is very low.  But so many genes influence tail size that there is high risk that at least one is copied incorrectly in each gamete.  This is especially a problem in males, since sperm production involves many more cell-copying events than egg production does in females. For example, mature human females carry eggs that have gone through only about 20 DNA replications, whereas age-30 males carry sperm that have gone through about 380 DNA replications, and age-50 males carry sperm that have gone through about 840 DNA-replications (Crow, 2000).  Thus, mutation load rises rapidly with paternal age, but not maternal age.  This onslaught of new mutations in every generation?especially from older males?restores the distribution of heritable fi! tness in panel ?a? and ensures a wide range of tail sizes, including small, dull ones, in every generation.  This is a potential resolution of the ?lek paradox.?(Houle & Kondrashov, 2002; Kotiaho et al., 2001; Michod & Hasson, 1990; Pomiankowski & Moller, 1995; Rowe & Houle, 1996). <snip .....