Genome Wide Analyses Reveal Little Evidence for Adaptive Evolution in Many Plant Species

Toni I. Gossmann 2 Bao-Hua Song 1 Aaron J. Windsor 1 Thomas Mitchell-Olds 1 Christopher J. Dixon 0 Maxim V. Kapralov 0 Dmitry A. Filatov 0 Adam Eyre-Walker 2 0 Department of Plant Sciences, University of Oxford , Oxford , United Kingdom 1 Institute for Genome Sciences and Policy, Department of Biology, Duke University 2 Centre for the Study of Evolution, School of Life Sciences, University of Sussex , Brighton , United Kingdom The relative contribution of advantageous and neutral mutations to the evolutionary process is a central problem in evolutionary biology. Current estimates suggest that whereas Drosophila, mice, and bacteria have undergone extensive adaptive evolution, hominids show little or no evidence of adaptive evolution in protein-coding sequences. This may be a consequence of differences in effective population size. To study the matter further, we have investigated whether plants show evidence of adaptive evolution using an extension of the McDonald-Kreitman test that explicitly models slightly deleterious mutations by estimating the distribution of fitness effects of new mutations. We apply this method to data from nine pairs of species. Altogether more than 2,400 loci with an average length of 280 nucleotides were analyzed. We observe very similar results in all species; we find little evidence of adaptive amino acid substitution in any comparison except sunflowers. This may be because many plant species have modest effective population sizes. - Introduction The contribution of adaptive evolution relative to genetic drift is a fundamental problem in molecular evolution (Kimura 1983; Gillespie 1991). Several methods to estimate the fraction of adaptive substitutions, , have been developed based on the McDonaldKreitman (MK) test (McDonald and Kreitman 1991) that contrast polymorphism and divergence between selectively and neutrally evolving sites (Charlesworth 1994; Fay et al. 2001; Smith and Eyre-Walker 2002; Bierne and Eyre-Walker 2004; Welch 2006; Boyko et al. 2008; Eyre-Walker and Keightley 2009). These methods have been applied to a variety of species. Estimates in Drosophila (Smith and Eyre-Walker 2002; Bierne and Eyre-Walker 2004; Welch 2006; Bachtrog 2008) and rodents (Halligan et al. 2010) suggest that 50% of all amino acid substitutions have been fixed as a consequence of adaptive evolution and for microorganisms estimates may be even higher (Charlesworth and Eyre-Walker 2006; Liti et al. 2009). However, although analyses of DNA sequence diversity show signs of some adaptive evolution (Fay et al. 2001; Zhang and Li 2005), overall the level of adaptive evolution in hominids appears to be very low (Chimpanzee Sequencing and Analysis Consortium 2005; Zhang and Li 2005; Boyko et al. 2008; Eyre-Walker and Keightley 2009). The contrast between hominids and other animals has led to the suggestion that effective population size may be an important determinant of the rate of adaptive evolution because hominids typically have low effective population sizes in contrast to rodents, insects, and bacteria (Eyre-Walker et al. 2002; Fraser et al. 2007; Halligan et al. 2010). However, some caution should be exercised because the level of adaptive evolution is typically measured as the proportion of substitutions that are adaptive and this depends both on the numbers of substitutions that are effectively neutral and the number that are advantageous. It has been shown that the number of effectively neutral mutations is negatively correlated to the effective population size in many species (Woolfit and Bromham 2003, 2005; Popadin et al. 2007; Moran et al. 2008; Piganeau and Eyre-Walker 2009). Hence, the correlation between proportion of substitutions that are adaptive and Ne may be a consequence of the correlation between the proportion of effectively neutral mutations and Ne and may not reflect any change in the absolute rate of adaptive evolution. The rate of adaptive evolution has also been studied in plants. On a genome-wide scale, previous studies in Arabidopsis thaliana have shown little evidence for adaptive evolution (Bustamante et al. 2002; Barrier et al. 2003; Schmid et al. 2005). This was attributed to the high frequency of inbreeding in A. thaliana and the reduction in effective population size that this caused. However, the outcrossing species A. lyrata (Barnaud et al. 2008; Foxe et al. 2008), the partially outcrossing cultivated tropical grass Sorghum bicolor (Hamblin et al. 2006), and the mainly outcrossing Zea species (Bijlsma et al. 1986; Ross-Ibarra et al. 2009) also show little evidence of positive selection. Instead, all these species show evidence of slightly deleterious The Author 2010. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: mutations (SDMs) segregating. In contrast, to the pattern in other plant species, Strasburg et al. (2009), Ingvarsson (2010), and Slotte et al. (2010) have recently estimated that 75%, 30%, and 40% of fixed amino acid differences were driven by adaptive substitutions in sunflowers, aspen trees, and some Brassicaceae species, respectively. There is evidence that SDMs contribute to variation in many populations (Akashi 1999; Cargill et al. 1999; Fay et al. 2002; Hughes 2005; Charlesworth and Eyre-Walker 2006). They are subject to weak negative selection (Nes 1), segregate at lower frequencies than neutral mutations, and contribute proportionally more to polymorphism than to divergence when compared with neutral mutations. The presence of SDMs in the analyses of Arabidopsis sp., S. bicolor , and Zea sp. may explain why there is so little apparent positive selection because SDMs are expected to bias the estimate of adaptive evolution downward if population sizes are either stationary or contractingthey can lead to an overestimate if population sizes have expanded (McDonald and Kreitman 1991; Eyre-Walker 2002). Here, we apply a method to estimate the proportion of adaptive substitutions that controls for the effects of SDMs by estimating the distribution of fitness effects (DFE) of new mutations (Eyre-Walker and Keightley 2009). It estimates the DFE from the polymorphic data and predicts the expected number of substitutions originating from neutral and SDMs. If the observed number of substitutions is greater than the expectation inferred from the DFE, it can be attributed to advantageous substitutions, yielding an estimate of . We present a new parameterization of this method that allows us to estimate the rate of adaptive substitution relative to the rate of synonymous substitution, thereby allowing us to explore whether the rate of adaptive evolution depends on Ne independent of the effects of Ne on the number of effectively neutral substitutions. We also apply the method of Fay et al. (2001), who suggested controlling for SDMs by removing low-frequency mutations from the analysis. We apply these met (...truncated)