Abstract Although the neutral theory of molecular evolution was proposed to explain DNA and protein sequence evolution, in principle it could also explain phenotypic evolution. Nevertheless, overall, phenotypes should be less likely than genotypes to evolve neutrally. I propose that, when phenotypic traits are stratified according to a hierarchy of biological organization, the fraction of evolutionary changes in phenotype that are adaptive rises with the phenotypic level considered. Consistently, molecular traits are frequently found to evolve neutrally whereas a large, random set of organismal traits were recently reported to vary largely adaptively. Many more studies of unbiased samples of phenotypic traits are needed to test the general validity of this hypothesis. adaptation, beneficial, genotype, molecular, organismal, phenotype Genotypic and Phenotypic Evolution Could Obey Different Laws It was through the observation of spectacular phenotypic variations among organisms combined with deductive reasoning that Darwin reached the conclusion of evolution by natural selection (Darwin 1859). According to this theory, a new phenotype will spread as long as it is present in a population, is heritable, and confers a higher organismal fitness than the existing phenotype. The theory is so simple yet so powerful that no biological phenomenon seems unexplainable by natural selection as long as one is willing to make some assumptions, most of which are usually hard to prove or disprove. Although various constraints and historical contingencies are now commonly recognized as additional factors shaping phenotypic evolution (Gould and Lewontin 1979), in the minds of the vast majority of biologists, natural selection is the predominant force driving phenotypic evolution. Contrasting this paradigm of phenotypic evolution is the neutral theory of molecular evolution (Kimura 1983), which asserts that the great majority of intra and interspecific variations in DNA and protein sequences are selectively neutral. The neutral theory originated half a century ago from the then-new observations in molecular biology, such as the surprisingly uniform yet extraordinarily rapid pace of sequence evolution of individual proteins and the agreement between the observed amino acid compositions and the corresponding random expectations from nucleotide compositions (Kimura 1968; King and Jukes 1969). Studies of population genetic behaviors of neutral alleles helped build a solid theoretical foundation of the neutral theory (Kimura 1983). That a substantial proportion of DNA sequence polymorphisms and divergences are neutral is now generally agreed upon by molecular evolutionists, although different opinions exist about the exact proportion (Nei 1987; Li 1997; Eyre-Walker 2006; Nei 2013; Graur et al. 2016). Furthermore, to a large extent, the neutral theory has successfully explained the evolution of genome architecture such as the abundances of duplicate genes, introns, and transposable elements across various lineages (Lynch 2007). Even though the neutral theory can in principle also explain phenotypic evolution, founders of the neutral theory limited it to genotypic evolution, as reflected by the formal name of the theory. In fact, Kimura, an originator and the chief proponent of the neutral theory, did not believe that it works for phenotypic evolution. In his landmark book, Kimura (1983) wrote that “The neutral theory is not antagonistic to the cherished view that evolution of form and function is guided by Darwinian selection, but it brings out another facet of the evolutionary process by emphasizing the much greater role of mutation pressure and random drift at the molecular level” (p. ix) and “laws governing molecular evolution are clearly different from those governing phenotypic evolution” (p. 326). Phenotypic evolution can occur via plastic changes, genetic changes, or a combination of the two (Ho and Zhang 2018). If we ignore the uncommon situation of phenotypic evolution that is due entirely to plasticity, phenotypic evolution must have underlying genotypic basis. Is it then theoretically inconsistent for phenotypic evolution to be largely adaptive while genotypic evolution mostly neutral? The answer is no and here is why. Evolutionary changes in DNA sequence, or substitutions, may be broadly classified into three categories: deleterious, beneficial, and effectively neutral, the latter referring to those whose absolute value of selection coefficient is smaller than the reciprocal of the effective population size (Kimura 1983). While deleterious substitutions may be rare, they can nevertheless occur by either genetic drift or hitchhiking. The fraction of DNA sequence evolution that is adaptive equals the proportion of all substitutions that are beneficial (AFG, where AF stands for adaptive fraction and the subscript G stands for genotype). Due to prevalent epistasis (Zhang 2017) and genotype-by-environment interactions (Wei and Zhang 2017), the fitness effect of a substitution depends on both the genetic background and environment. Hence, any fitness effect discussed here refers to that in the genetic background and environment when the substitution takes place. Substitutions that do not have phenotypic effects must be neutral, because otherwise they would impact at least one phenotypic trait—fitness (fig. 1A). Substitutions that have phenotypic effects may be neutral, beneficial, or deleterious (fig. 1A). Hence, AFG is the ratio between the length of the blue bar and the total length of all bars in figure 1A. The fraction of phenotype-altering substitutions that are adaptive (AFP, where the subscript P stands for phenotype) is the ratio between the length of the blue bar and the total length of the three bars of the top row in figure 1A. Clearly, AFP exceeds AFG. In other words, phenotype-altering substitutions have a higher probability to be adaptive than average substitutions. Under the reasonable assumption that an average beneficial substitution impacts at least as many phenotypic traits as does an average phenotype-altering neutral or deleterious substitution, we can further deduce that the adaptive fraction of phenotypic evolution exceeds the adaptive fraction of nucleotide substitutions. In short, genotypes and phenotypes can in theory obey different laws of evolution. Fig. 1. View largeDownload slide Schematics explaining the expected fraction of adaptive phenotypic evolution. (A) Numbers of deleterious, beneficial, and effectively neutral substitutions stratified by the presence/absence of any phenotypic effect. The length of a horizontal bar represents the hypothetical number of substitutions in the category. Fraction of phenotype-altering substitutions that are adaptive (AFP) equals the length of the blue bar divided by the total length of the three bars of the top row, whereas the fraction of substitutions that are adaptive (AFG) equals the length of the blue bar divided by the total length of all bars. Therefore, AFP > AFG. This conclusion can be extended to adaptive fractions of phenotypic and genotypic changes under a reasonable assumption (see main text). (B) Numbers of deleterious, beneficial, and neutral substitutions stratified by their effects on traits of different phenotypic levels. At each phenotypic level, substitutions impacting traits of that level but not any higher level are considered. According to our assumption, substitutions impacting traits at a phenotypic level always impact at least one trait at each lower phenotypic level. The length of a horizontal bar represents the hypothetical number of substitutions in the category. (C) Fraction of phenotype-impacting substitutions that are beneficial (AF) rises with the phenotypic level considered. At each level, AF is computed by the length of the blue bar divided by the total length of all bars at this phenotypic level and above in panel B. Under a simple assumption (see main text), AF also describes the adaptive fraction of evolutionary changes at each phenotypic level. Fig. 1. View largeDownload slide Schematics explaining the expected fraction of adaptive phenotypic evolution. (A) Numbers of deleterious, beneficial, and effectively neutral substitutions stratified by the presence/absence of any phenotypic effect. The length of a horizontal bar represents the hypothetical number of substitutions in the category. Fraction of phenotype-altering substitutions that are adaptive (AFP) equals the length of the blue bar divided by the total length of the three bars of the top row, whereas the fraction of substitutions that are adaptive (AFG) equals the length of the blue bar divided by the total length of all bars. Therefore, AFP > AFG. This conclusion can be extended to adaptive fractions of phenotypic and genotypic changes under a reasonable assumption (see main text). (B) Numbers of deleterious, beneficial, and neutral substitutions stratified by their effects on traits of different phenotypic levels. At each phenotypic level, substitutions impacting traits of that level but not any higher level are considered. According to our assumption, substitutions impacting traits at a phenotypic level always impact at least one trait at each lower phenotypic level. The length of a horizontal bar represents the hypothetical number of substitutions in the category. (C) Fraction of phenotype-impacting substitutions that are beneficial (AF) rises with the phenotypic level considered. At each level, AF is computed by the length of the blue bar divided by the total length of all bars at this phenotypic level and above in panel B. Under a simple assumption (see main text), AF also describes the adaptive fraction of evolutionary changes at each phenotypic level. Relative Prevalence of Adaptive Changes Should Rise with Phenotypic Level Because the phenotype of an organism includes all measureable phenotypic traits, to study phenotypic evolution systematically as well as to simplify the matter, it seems natural to classify all traits into a hierarchy based on different levels of biological organization. For instance, for multicellular organisms, a potential sequence from low to high levels of biological organization can be traits at molecular, cellular, tissue, organ, and organismal levels. We may add on top of organismal traits the trait of Darwinian fitness, which is the summary of all traits. In this hierarchy, molecular traits would include concentrations and molecular functions of all (non-DNA) molecules such as RNAs, proteins, and metabolites. Cellular traits would include, for example, cell size, weight, and shape. Tissue, organ, and organismal traits similarly include characteristics of tissues, organs, and organisms, respectively. Because the phenotypic levels outlined above are based on a hierarchy of biological organization, it is reasonable to assume that a substitution impacting a trait at one level must also influence at least one trait at each phenotypic level below the focal level. That is, all substitutions affect the genotype, but only some substitutions affect molecular traits, among which only a subset affect cellular traits, and so on. Substitutions impacting fitness, the top-level trait, are by definition nonneutral; most of them should be beneficial although a minority may be deleterious (fig. 1B). Substitutions impacting organismal traits or traits at any lower phenotypic level but not fitness are by definition neutral (fig. 1B). The fraction of fitness-altering substitutions that are adaptive (AFfitness; fig. 1C) is the length of the blue bar divided by the total length of the blue bar and red bar in figure 1B. Now let us consider the fraction of organismal trait-altering substitutions that are adaptive (AForganism). Because every fitness-impacting substitution must also influence some organismal traits, as assumed above, AForganism equals the number of beneficial substitutions divided by the sum of the number of beneficial substitutions, number of deleterious substitutions, and number of neutral substitutions impacting organismal traits (i.e., the length of the blue bar divided by the total length of all bars of the top two rows in fig. 1B). Clearly AForganism < AFfitness (fig. 1C). Similarly, the fraction of organ trait-altering substitutions that are adaptive (AForgan) equals the number of beneficial substitutions divided by the sum of the numbers of beneficial substitutions, deleterious substitutions, organismal trait-altering neutral substitutions, and organ trait-altering neutral substitutions (i.e., the length of the blue bar divided by the total length of all bars of the top three rows in fig. 1B). Hence, AForgan < AForganism (fig. 1C). It can be shown that AFG < AFmolecule < AFcell < AFtissue < AForgan < AForganism < AFfitness (fig. 1C). That is, the fraction of phenotype-altering substitutions that are adaptive increases with the phenotypic level considered and is higher than the overall fraction of nucleotide substitutions that are adaptive. The sequence of the phenotypic levels considered above is but one of many possibilities. For instance, one could add more phenotypic levels or collapse some levels. The cellular, tissue, organ, and organismal levels all collapse to one level in unicellular organisms. The essential criterion in the classification is that phenotypic levels follow a hierarchy of biological organization such that a phenotypic change at one level must involve at least one phenotypic change in each level below. In the foregoing analysis, we considered fractions of phenotype-altering substitutions that are adaptive. Because of pleiotropy, a substitution can potentially impact multiple traits of the same phenotypic level in addition to influencing traits of different levels. At a phenotypic level, let k be the ratio between the expected number of traits affected by a beneficial substitution and that affected by a neutral or deleterious substitution impacting the level. Under the reasonable assumption that k at any phenotypic level is no smaller than that at any level below it, the above result on AF can be extended to the proportion of phenotypic evolutionary changes that are adaptive. In other words, the adaptive fraction of phenotypic evolution is also expected to exceed AFG and to rise with the phenotypic level. Evolution of Molecular Traits Empirically verifying the above theoretical prediction about the relative abundances of neutral and adaptive phenotypic changes in evolution is challenging. A number of methods have been developed to test adaptive versus neutral evolution hypotheses for phenotypic traits (Lande 1976, 1977; Chakraborty and Nei 1982; Lynch and Hill 1986; Turelli et al. 1988; Lynch 1990; Spitze 1993; Orr 1998), and numerous authors have studied the evolution of individual traits in this context (Endler 1986; Kingsolver et al. 2001). Although such studies are valuable for identifying cases of adaptive or neutral phenotypic evolution, they cannot tell, even in aggregation, whether most phenotypic traits evolve neutrally or adaptively, because of the strong preference in choosing potentially adaptive traits for investigation and the bias in publishing studies that detect adaptive signals. To obtain a general picture on the neutrality or adaptation in phenotypic evolution requires studying large sets of unbiasedly sampled traits. This is easier said than done, because (1) an organism has an almost infinite number of phenotypic traits even at one phenotypic level, making it impossible to study all traits, (2) phenotypic traits are often correlated (e.g., the lengths of the two arms in humans), making it unclear how to sample them fairly, and (3) biologists are attracted to certain phenotypic traits (e.g., those exhibiting clear trends of geographic variation or conspicuous to human sensory systems) more than other traits, potentially biasing the sampling. This situation contrasts sharply that of genotypic studies, where it is straightforward to circumvent sampling bias by analyzing either an entire genome or random segments of a genome. Although certain types of genotypic changes (e.g., single nucleotide substitutions) are studied more often than other types (e.g., insertions and deletions), the results are at least unbiased for the type of genotypic changes considered. Notwithstanding, it has become possible to analyze either one complete set of traits or a random sample in a set of traits, most notably for molecular traits. For example, gene expression level, or the cellular concentration of the mRNA product of a gene, is a phenotypic trait at the molecular level. Measuring such traits simultaneously for all genes in a genome is now routinely performed thanks to advancements in genome technology in the last decade. Although the mRNA concentrations of different genes are not independent from one another, the sampling can be unbiased because typically either all genes or a random subset of genes are studied. When genome-wide gene expression data are analyzed, stabilizing or purifying selection instead of positive selection is usually found for the vast majority of genes. For instance, gene expression variation among wild strains of Caenorhabditis elegans worms is smaller than that among mutation accumulation (MA) lines that diverged from one another under virtually no selection, despite that the divergence times are much longer among the wild strains than among MA lines (Denver et al. 2005). Such results are similar to the general trend in genotypic evolution where the dominant force is purifying selection rather than positive selection (Zhang 2010). Furthermore, patterns of gene expression similarities among nine yeast strains of three species were found to be almost exclusively determined by the phylogenetic relationships instead of environmental similarities of these strains, further suggesting that the intra and interspecific expression variations of most genes are not the result of environmental adaptations (Yang et al. 2017), although a minority of yeast genes exhibit adaptive expression differences (Bullard et al. 2010; Fraser et al. 2010; Qian et al. 2012). In addition to mRNA concentrations, the relative prevalence of neutral and adaptive evolution has been assessed for several other large sets of molecular traits, most of which are posttranscriptional or posttranslational modifications. For instance, A-to-I RNA editing enzymatically converts adenosines to inosines in RNA molecules and is the most common editing of transcripts of animal nuclear genes (Nishikura 2016). Whether an A site is edited to I and the fraction of RNA molecules in which the A site is edited (i.e., editing level) are phenotypic traits. Because I is recognized as guanine (G) by ribosomes, coding A-to-I editing could result in nonsynonymous changes, which generate new protein variants and thus could be adaptive. If coding RNA editing is largely beneficial for protein diversity, the proportion of A sites that are edited and the editing level should both be higher for nonsynonymous editing than synonymous editing, which is expected to be more or less neutral. The observation from transcriptome-wide analysis in humans, however, showed that nonsynonymous editing has lower frequencies and lower editing levels than synonymous editing, indicating that if anything, nonsynonymous editing is generally harmful and thus has been reduced by purifying selection (Xu and Zhang 2014). These and other lines of evidence suggest that most observed coding sequence editing events probably result from promiscuous activities of editing enzymes and are selectively permitted rather than selectively favored (Xu and Zhang 2014). Given this finding, one would expect rapid turnovers of edited sites in evolution, which is indeed confirmed by the observation of little overlap between human and mouse edited sites (Pinto et al. 2014). Similar results were recently reported for C-to-U RNA editing, the second most common RNA editing of animal nuclear genes (Liu and Zhang 2018a). Furthermore, transcriptome-wide analysis of methylation of A at the nitrogen-6 position (m6A) shows that this common posttranscriptional modification is subject to no detectable purifying selection within species and is only weakly conserved between species, suggesting that most m6A modifications are likely nonfunctional off-target modification errors (Liu and Zhang 2018b). For such phenotypes, neutral evolution is expected to be prevalent because the proportion of phenotype-altering substitutions that are neutral is high. In addition, studies of alternative splicing (Saudemont et al. 2017), protein phosphorylation (Landry et al. 2009), and protein glycosylation (Park and Zhang 2011) all suggest that neutral changes are common and are likely more abundant than adaptive changes in the evolution of these molecular traits. Therefore, we may conclude that neutral changes dominate the evolution of molecular traits so far studied. This said, it is notable that, in at least two groups of species, one including octopus, squid, and cuttlefish (Liscovitch-Brauer et al. 2017) and the other including the red bread mold Neurospora crassa and related fungi (Liu et al. 2017), A-to-I RNA editing frequency and/or level are higher for nonsynonymous editing than synonymous editing, indicative of the action of positive selection for amino acid-altering editing. But in neither case is the specific advantage of the pervasive nonsynonymous editing known. Other molecular traits of interest in the assessment of relative abundances of neutral and adaptive evolution include protein and metabolite concentrations, strengths of various molecular interactions such as protein–protein, protein–DNA, and protein–RNA interactions, RNA and protein structures, and chromatin structures, in addition to many other posttranscriptional and posttranslational modifications. It is expected that further technological improvements in genomics, proteomics, and structural biology will make these analyses possible in the near future. Evolution of Traits at Higher Phenotypic Levels When one moves upward from molecular traits to cellular, tissue, organ, and organismal traits, it becomes increasingly difficult to study all traits in an objectively defined set at the phenotypic level considered. Hence, unbiased sampling becomes more important. For example, Ohya et al. (2005) measured hundreds of morphological traits of yeast cells using fluorescent microscopic images of cells stained with three different chemicals that, respectively, dye the cell wall, actin cytoskeleton, and nuclear DNA. These traits were measured not because of their potential relevance to adaptation but because they can be quantified. For this reason, they represent a random subset of all morphological traits of yeast cells regarding adaptive versus neutral evolution. Using 210 of these morphological traits, Ho et al. (2017) showed that, the more important a morphological trait is to fitness, the more the trait varies within and between yeast species, even after the control for mutation size variation among traits. Under the neutral hypothesis, compared with traits that are relatively unimportant to fitness, relatively important traits should be subject to stronger purifying selection and evolve more slowly given the same speed of mutational input, analogous to the neutral paradigm that functional genes evolve more slowly than functionless pseudogenes (Li et al. 1981). Hence, Ho et al.’s (2017) observation is unexplainable by neutral evolution and requires invoking prevalent adaptations, under which, more important traits are subject to stronger positive selection and could evolve faster than less important traits. By contrast, a similar analysis of between-strain differences in expression levels of thousands of yeast genes showed that, given the same mutational input, the more important a gene is to fitness, the less its expression varies, consistent with the neutral nature of most gene expression variations aforementioned. Therefore, evolution of molecular traits and that of cellular traits appear to obey different rules in yeast. Because yeast is a unicellular organism, cellular traits are also organismal traits. That variations of organismal traits are largely adaptive while those of molecular traits are largely neutral is consistent with the theoretical prediction in figure 1C. This is certainly just the beginning of a systematic and unbiased empirical assessment of the relative abundances of neutral and adaptive phenotypic changes in evolution. Many more studies, especially in multicellular organisms, are needed in this area. If confirmed in multiple species, the theoretical prediction in figure 1C provides an explanation of the relative prevalence of adaptation at different phenotypic levels and can guide those who are interested in identifying adaptive or neutral phenotypic variations. Outlook While the neutral theory was proposed specifically to explain DNA and protein sequence evolution, the impact of the neutral theory is beyond the field of molecular evolution. For evolutionary biologists, neutrality is commonly considered the null hypothesis against which adaptation is tested even in studies of phenotypic evolution. Recent investigations of large numbers of molecular traits, however, showed that neutrality is more than a null hypothesis that awaits rejection but an appropriate description of the evolution of most molecular traits examined. This finding, along with the theoretic prediction outlined in figure 1C regarding the relative prevalence of neutral and adaptive evolution at different phenotypic levels, will hopefully stimulate systematic studies of evolutionary forces driving phenotype evolution. This research program will ultimately help understand the functions as well as evolution of phenotypic variations. While Kimura (1983) did not believe in neutral phenotype evolution, let us be reminded that Darwin wrote in Origin of Species that “Variations neither useful nor injurious would not be affected by natural selection, and would be left a fluctuating element, as perhaps we see in the species called polymorphic” (p. 81) (Darwin 1859). Acknowledgments I thank Wei-Chin Ho, Xinzhu Wei, and two anonymous reviewers for valuable comments. This study was supported in part by National Institutes of Health research grant R01GM120093. References Bullard JH, Mostovoy Y, Dudoit S, Brem RB. 2010. Polygenic and directional regulatory evolution across pathways in Saccharomyces. Proc Natl Acad Sci U S A. 107( 11): 5058– 5063. Google Scholar CrossRef Search ADS PubMed Chakraborty R, Nei M. 1982. Genetic differentiation of quantitative characters between populations or species: I. Mutation and random genetic drift. Genet Res. 39( 03): 303– 314. Google Scholar CrossRef Search ADS Darwin C. 1859. On the Origin of Species by Means of Natural Selection . London: J. Murray. Denver DR, Morris K, Streelman JT, Kim SK, Lynch M, Thomas WK. 2005. The transcriptional consequences of mutation and natural selection in Caenorhabditis elegans. Nat Genet. 37( 5): 544– 548. Google Scholar CrossRef Search ADS PubMed Endler JA. 1986. Natural Selection in the Wild . Princeton, NJ: Princeton University Press. Eyre-Walker A. 2006. The genomic rate of adaptive evolution. Trends Ecol Evol. 21( 10): 569– 575. Google Scholar CrossRef Search ADS PubMed Fraser HB, Moses AM, Schadt EE. 2010. Evidence for widespread adaptive evolution of gene expression in budding yeast. Proc Natl Acad Sci U S A. 107( 7): 2977– 2982. Google Scholar CrossRef Search ADS PubMed Gould SJ, Lewontin RC. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc R Soc Lond B Biol Sci. 205( 1161): 581– 598. Google Scholar CrossRef Search ADS PubMed Graur D, Sater AK, Cooper TF. 2016. Molecular and Genome Evolution . Sunderland, MA: Sinauer. Ho WC, Ohya Y, Zhang J. 2017. Testing the neutral hypothesis of phenotypic evolution. Proc Natl Acad Sci U S A. 114( 46): 12219– 12224. Google Scholar CrossRef Search ADS PubMed Ho W-C, Zhang J. 2018. Evolutionary adaptations to new environments generally reverse plastic phenotypic changes. Nat. Commun . 9( 1): 350. Google Scholar CrossRef Search ADS PubMed Kimura M. 1968. Evolutionary rate at the molecular level. Nature 217( 5129): 624– 626. Google Scholar CrossRef Search ADS PubMed Kimura M. 1983. The Neutral Theory of Molecular Evolution . Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS King JL, Jukes TH. 1969. Non-Darwinian evolution. Science 164( 3881): 788– 798. Google Scholar CrossRef Search ADS PubMed Kingsolver JG, Hoekstra HE, Hoekstra JM, Berrigan D, Vignieri SN, Hill CE, Hoang A, Gibert P, Beerli P. 2001. The strength of phenotypic selection in natural populations. Am Nat. 157( 3): 245– 261. Google Scholar CrossRef Search ADS PubMed Lande R. 1976. Natural selection and random genetic drift in phenotypic evolution. Evolution 30( 2): 314– 334. Google Scholar CrossRef Search ADS PubMed Lande R. 1977. Statistical tests for natural selection on quantitative characters. Evolution 31( 2): 442– 444. Google Scholar CrossRef Search ADS PubMed Landry CR, Levy ED, Michnick SW. 2009. Weak functional constraints on phosphoproteomes. Trends Genet. 25( 5): 193– 197. Google Scholar CrossRef Search ADS PubMed Li WH. 1997. Molecular Evolution . Sunderland, MA: Sinauer. Li WH, Gojobori T, Nei M. 1981. Pseudogenes as a paradigm of neutral evolution. Nature 292( 5820): 237– 239. Google Scholar CrossRef Search ADS PubMed Liscovitch-Brauer N, Alon S, Porath HT, Elstein B, Unger R, Ziv T, Admon A, Levanon EY, Rosenthal JJC, Eisenberg E. 2017. Trade-off between transcriptome plasticity and genome evolution in cephalopods. Cell 169( 2): 191– 202. Google Scholar CrossRef Search ADS PubMed Liu H, Li Y, Chen D, Qi Z, Wang Q, Wang J, Jiang C, Xu JR. 2017. A-to-I RNA editing is developmentally regulated and generally adaptive for sexual reproduction in Neurospora crassa. Proc Natl Acad Sci U S A. 114( 37): E7756– E7765. Google Scholar CrossRef Search ADS PubMed Liu Z, Zhang J. 2018a. Human C-to-U coding RNA editing is largely nonadaptive. Mol Biol Evol. 35( 4): 963– 969. Google Scholar CrossRef Search ADS Liu Z, Zhang J. 2018b. Most m6A RNA modifications in protein-coding regions are evolutionarily unconserved and likely nonfunctional. Mol Biol Evol. 35( 3): 666– 675. Google Scholar CrossRef Search ADS Lynch M. 1990. The rate of morphological evolution in mammals from the standpoint of the neutral expectation. Am Nat. 136( 6): 727– 741. Google Scholar CrossRef Search ADS Lynch M. 2007. The Origins of Genome Architecture . Sunderland, MA: Sinauer. Google Scholar PubMed PubMed Lynch M, Hill WG. 1986. Phenotypic evolution by neutral mutation. Evolution 40( 5): 915– 935. Google Scholar CrossRef Search ADS PubMed Nei M. 1987. Molecular Evolutionary Genetics . New York: Columbia University Press. Nei M. 2013. Mutation-Driven Evolution . Oxford: Oxford University Press. Nishikura K. 2016. A-to-I editing of coding and non-coding RNAs by ADARs. Nat Rev Mol Cell Biol. 17( 2): 83– 96. Google Scholar CrossRef Search ADS PubMed Ohya Y, Sese J, Yukawa M, Sano F, Nakatani Y, Saito TL, Saka A, Fukuda T, Ishihara S, Oka S, et al. 2005. High-dimensional and large-scale phenotyping of yeast mutants. Proc Natl Acad Sci U S A. 102( 52): 19015– 19020. Google Scholar CrossRef Search ADS PubMed Orr HA. 1998. Testing natural selection vs. genetic drift in phenotypic evolution using quantitative trait locus data. Genetics 149( 4): 2099– 2104. Google Scholar PubMed Park C, Zhang J. 2011. Genome-wide evolutionary conservation of N-glycosylation sites. Mol Biol Evol. 28( 8): 2351– 2357. Google Scholar CrossRef Search ADS PubMed Pinto Y, Cohen HY, Levanon EY. 2014. Mammalian conserved ADAR targets comprise only a small fragment of the human editosome. Genome Biol. 15( 1): R5. Google Scholar CrossRef Search ADS PubMed Qian W, Ma D, Xiao C, Wang Z, Zhang J. 2012. The genomic landscape and evolutionary resolution of antagonistic pleiotropy in yeast. Cell Rep. 2( 5): 1399– 1410. Google Scholar CrossRef Search ADS PubMed Saudemont B, Popa A, Parmley JL, Rocher V, Blugeon C, Necsulea A, Meyer E, Duret L. 2017. The fitness cost of mis-splicing is the main determinant of alternative splicing patterns. Genome Biol. 18( 1): 208. Google Scholar CrossRef Search ADS PubMed Spitze K. 1993. Population structure in Daphnia obtusa—quantitative genetic and allozymic variation. Genetics 135( 2): 367– 374. Google Scholar PubMed Turelli M, Gillespie JH, Lande R. 1988. Rate tests for selection on quantitative characters during macroevolution and microevolution. Evolution 42( 5): 1085– 1089. Google Scholar CrossRef Search ADS PubMed Wei X, Zhang J. 2017. The genomic architecture of interactions between natural genetic polymorphisms and environments in yeast growth. Genetics 205( 2): 925– 937. Google Scholar CrossRef Search ADS PubMed Xu G, Zhang J. 2014. Human coding RNA editing is generally nonadaptive. Proc Natl Acad Sci U S A. 111( 10): 3769– 3774. Google Scholar CrossRef Search ADS PubMed Yang JR, Maclean CJ, Park C, Zhao H, Zhang J. 2017. Intra and interspecific variations of gene expression levels in yeast are largely neutral. Mol Biol Evol. 34( 9): 2125– 2139. Google Scholar CrossRef Search ADS PubMed Zhang J. 2010. Evolutionary genetics: progress and challenges. In: Bell MA, Futuyma DJ, Eanes WF, Levinton JS, editors. Evolution Since Darwin: The First 150 Years . Sunderland, MA: Sinauer. p. 87– 118. Zhang J. 2017. Epistasis analysis goes genome-wide. PLoS Genet. 13( 2): e1006558. Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Molecular Biology and Evolution – Oxford University Press
Published: Apr 5, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera