Evolutionary and Ecological Genomics of Arabidopsis

Integration of three biological fields is required for comprehensive understanding of organismal biology. Molecular genetic studies are necessary to reveal which gene and the precise genetic variation responsible for developmental or physiological diversity. Ecological studies demonstrate how an organism interacts with conspecific individuals, other species, and abiotic environments to survive and reproduce. Evolutionary forces provide feedback mechanisms, including natural selection for adaptive traits, genetic drift, and demographic processes, which determines the genetic constitution of subsequent generations. Evolutionary genomics studies these forces by analyzing the genetics underlying phenotypic evolution in the context of genome-wide molecular information. At the heart of these studies are attempts to understand traits in their broad ecological and evolutionary contexts. Ecology, which examines interactions among individuals, species, and their abiotic environment, requires knowledge of the response of plants to their natural environments. Evolutionary studies examine how selection, genetic drift, and evolutionary history have shaped patterns of variation at both the molecular and phenotypic levels. In this article, we will illustrate the power of Arabidopsis as a model plant for the study of ecology and evolutionary genomics, and the approaches that have led to key insights not only for these fields but also for plant biology in general (Mitchell-Olds, 2001; Shimizu, 2002). Arabidopsis (L. Heynh.; family Brassicaceae) is a weedy annual plant, occupying disturbed habitats such as the margins of agricultural fields. It is a predominantly selfing species, with a reported out-crossing rate of approximately 1% (compiled by Hoffmann et al., 2003). Most natural populations adopt the winter annual life history strategy, which is characterized by plants germinating in the fall, overwintering as vegetative rosettes, and flowering in the spring. The spring annual strategy, with germination and flowering in spring, is also commonly observed (Pigliucci, 2002). Floral morphological diversity in the genus Arabidopsis. Left, Arabidopsis Col-0 accession. Right, Arabidopsis halleri subsp. gemmifera, collected in Osaka. This self-incompatible species has large petals to attract pollinator insects, and the anthers are separated from the stigma to avoid autopollination. Scale bar = 1 mm. Arabidopsis displays a wide range of ecological relationships, including within- and between-species interactions and adaptations to abiotic environments. It responds physiologically and developmentally to a large variety of environmental cues, including light, daylength, vernalization, and nutrient and water levels (for review, see Pigliucci, 2002; Koornneef et al., 2004). Arabidopsis can be affected by bacterial and fungal pathogens and by insect herbivory. Despite being predominantly self-pollinating, flower visits by solitary bees, dipterans, and thrips have been observed in the field (Mitchell-Olds, 2001; Hoffmann et al., 2003). Furthermore, interactions with other individuals of Arabidopsis are ecologically important as mates and as competitors (Bates and Lynch, 2001; Shimizu et al., 2004; Köhler et al., 2005). The strength of Arabidopsis as a model system for ecological and evolutionary genetics is that it allows researchers to identify the genetic basis of a wide array of evolutionary and ecological phenomena. Several approaches to gene identification have been utilized to reveal the molecular genetic bases of various putative adaptive traits in this species, with the goal of determining the specific genetic polymorphism(s) responsible for ecologically and evolutionarily relevant phenotypic diversity. Many traits of interest to evolutionary biologists and ecologists, such as flowering time, water use efficiency, and trichome density, are quantitative in nature. It is no surprise that quantitative trait locus (QTL) mapping studies have been key components of Arabidopsis research, and this has been reviewed elsewhere (Koornneef et al., 2004). QTL mapping studies have been applied to examine the genetic basis of various traits, such as flowering time, inflorescence architecture, seed size, insect resistance, and light response (Koornneef et al., 2004; see below). Although only two accessions can be studied under standard QTL mapping techniques, the use of multiple populations can sample a wider range of natural variants, a point exemplified by a recent study of trichome density (Symonds et al., 2005). New genomic technologies, such as extreme array mapping using chip-based genomic arrays, are also beginning to provide more rapid mapping methods to identify QTL (Wolyn et al., 2004). A new genomics approach to trait locus mapping is linkage disequilibrium (LD) or association mapping, which may provide a new tool in the identification of genes underlying natural phenotypic (and perhaps adaptive) variation. In LD mapping, researchers exploit recombination and allele correlations that have occurred over evolutionary time to detect associations between particular genomic markers and specific phenotypes of interest. The use of LD mapping allows researchers to screen for alleles in a more diverse set of genotypes than is possible under standard QTL mapping studies. This procedure may, however, be complicated by nonindependence of individuals from each other in mapping populations due to population structure, but theoretical advances provide methods to take this into account. LD mapping in Arabidopsis has been used to detect correlations of flowering-time variation in CRY2 (Olsen et al., 2004) and FRI (Hagenblad et al., 2004). Estimates of LD in the Arabidopsis genome indicate that allele correlations can extend 50 to 250 kb, suggesting that whole-genome scanning by LD mapping in this species may be feasible (Nordborg et al., 2002). Whether such whole-genome LD mapping scans are possible remains to be seen; a recent study of LD mapping of known flowering-time genes shows that it may be difficult to detect genes underlying a trait if the genetic architecture is complex (e.g. involving epistasis or multiple alleles; Hagenblad et al., 2004). Alternatively, researchers can exploit the high level of available molecular genetic information and use a candidate gene approach. With this method, genes that are known to affect a trait may be examined for further evidence that they are causally associated with a trait of interest in natural environments. Candidate gene approaches may also be used in conjunction with QTL and LD mapping strategies; this is most vividly illustrated by studies of the molecular genetic basis of ecological variation in flowering time (El-Assal et al., 2001). Knowledge of genetic networks has also been useful, as in a recent study in which researchers utilized information on the flowering-time pathway to identify a key epistatic regulatory interaction between the FRI and FLC genes that underlies a latitudinal cline in flowering time (Caicedo et al., 2004). It should be noted that all these association studies remain statistical associations until verified by genetic complementation (including transgenic complementation) techniques (see below). Transgenic methods provide important tools to prove that isolated genes (including candidate loci) actually underlie natural variation in the trait of interest. This is especially valuable in studying differences between species where genetic segregation analysis is impossible. For example, the transformation of SRK and SCR genes of Arabidopsis lyrata into Arabidopsis restored the self-incompatible response in the latter species, proving that mutations in these genes were responsible for the evolution of selfing (Nasrallah et al., 2002). Quantitative characters, however, may prove difficult to study by transgenic complementation, particularly if the alleles are of small to moderate effect. This is because independent transformants with the same transgenic genes very often display heterogeneous phenotypes, resulting from variability of transgene insertion locations and copy numbers. In certain cases, however, other methods may allow the analyses of quantitative characters by transgenic means (Tian et al., 2003). Traditional genetic complementation tests are also routinely used to check whether a new laboratory-induced mutation is an allele of a known or novel gene; in principle, they could also be employed to identify genes underlying ecological traits or evolutionary changes. Maloof et al. (2001) used a complementation test to determine that a naturally occurring allele in the Lm-2 accession, which has reduced far-red sensitivity, is an allele of the PHYA gene. Finally, recent theoretical and experimental work suggests the utility of reverse genomic approaches collectively referred to either as adaptive trait locus or as hitchhiking mapping. These techniques rely on specific predictions of molecular evolutionary and population genetic theory on the levels and patterns of genetic variation expected for genes experiencing positive, or directional selection (selection that fixes an allele harboring an advantageous mutation that increases individual fitness; positive selection is often referred to as Darwinian selection) or balancing selection (selection that maintains variant alleles in a population, which may arise from heterozygote advantage, selection in variable environments, or fitness values that depend on allelic frequency; Luikart et al., 2003). For example, recent positive selection on a gene leads to reduced levels of nucleotide variation, while balancing selection is associated with increased levels of molecular diversity surrounding the selected mutation. Moreover, recurrent selection on protein sequence results in elevated levels of nonsynonymous (KA) nucleotide changes compared to synonymous (KS) substitutions. The latter approach has been used to identify genes associated with the divergence between Arabidopsis and its closely related congener A. lyrata (Barrier et al., 2003), and to examine the evolution of pollen genes between these two species (Schein et al., 2004). Reverse genetic approaches can then be used to determine the functions of identified adaptive trait genes and the phenotypic consequences associated with differential selection. Determining the genetic basis of adaptation is a central focus of evolutionary and ecological research. There have been concerted efforts in recent years to assess the genetics underlying putatively adaptive traits that vary within and between species. Using the approaches described above, we can identify the genes (and the specific polymorphisms within these genes) and determine the functional mechanisms underlying these adaptive traits and the evolutionary histories that gave rise to them (Mitchell-Olds, 2001; Shimizu, 2002; Luikart et al., 2003). Successful studies along these lines have employed evolutionary genomic analyses to draw inferences on the evolutionary forces (including selection, drift, and population structure) that have shaped the history of these adaptive loci. A key concept of evolutionary genomics is that selection is a deterministic force that affects single genes, while population-level processes, such as population expansion and migration, are stochastic forces that affect all genes in the genome. Recent theoretical advances and the availability of genome-wide polymorphism data now permit researchers to discriminate between selective forces and population-level processes and thus identify genes underlying adaptive evolution (Luikart et al., 2003). Here, we discuss several examples that illustrate the utility of combining various approaches to understanding the diversification of adaptive traits, and an increasing number of such investigations using Arabidopsis are being employed to address evolutionary and ecological issues (Fig. 1). The evolutionary transition from out-crossing to selfing is one of the most prevalent trends in flowering plants. Charles Darwin (1876) proposed the reproductive assurance model to explain the prevalence of self-pollination in plants, suggesting that selfing can be evolutionarily advantageous when pollinators or mates are scarce in spite of inbreeding depression. Darwin's model also underlies Baker's rule, which states that colonizing species that disperse over long distances are generally self-compatible (Charlesworth, 2003; Shimizu et al., 2004, and refs. therein). Self-incompatibility is a major mechanism to prevent selfing in plants. A. lyrata and many Brassicaceae species have a self-incompatible recognition system controlled by the Sterility (S)-locus, which harbors at least two functional genes, the female receptor gene SRK/Aly13 and the male ligand gene SCR/SP11. A number of S-haplotypes with divergent sequences are maintained by balancing selection in these species. Arabidopsis, however, has pseudogenes of SRK and SCR. Transgenic experiments showed that the loss of functional alleles at these genes is responsible for the emergence of selfing (Nasrallah et al., 2002). The pseudoSCR1 gene in 21 Arabidopsis accessions has low levels of nucleotide diversity compared with neighboring genes in the pseudo S-locus and with genomic average. This low value is consistent with the hypothesis that the pseudogene allele of SCR1 was advantageous and recently spread to fixation in the species. Computer simulation based on coalescent theory (a mathematical theory to analyze the genealogy of DNA sequences, often used to derive inferences about demographic, population-level forces, and natural selection) demonstrates that this selection event most probably occurred very recently. The 95% confidence interval of the time estimate spans 0 to 320,000 years ago, when the planet experienced 100,000-year cycles of glacial-interglacial climate changes. Within this interval, the likelihood of the time estimate for the selective sweep was highest at T equals approximately 0 years, a time frame consistent with the expansion of the species range approximately 17,000 years ago after the last glacial retreats. If indeed selfing evolved during postglacial species expansion, it provides support for Darwin's reproductive assurance model, since rapid expansion would be accompanied by scarcities of mates and pollinators and thus selfing plants would have a selective advantage during long-distance dispersals (Shimizu et al., 2004). Also, the evolution of self-fertilization must have been followed by rapid evolution in floral morphological traits to facilitate selfing (Fig. 2). Those traits, including small flower size and autopollination, must have evolved after becoming self-compatible because they would have been deleterious if plants remained self-incompatible. These examples support the hypothesis that rapid adaptive evolution is a major response to climate change (for review, see Davis and Shaw, 2001). Disease resistance genes are fascinating targets of selection, with their evolutionary dynamics driven by coevolution between the plant and the attacking pathogen (Bergelson et al., 2001). The Arabidopsis Col-0 accession is resistant to the bacterial pathogen Pseudomonas syringae avrRpm1, whereas the Nd-0 accession displays susceptibility. Mapping studies between Col-0 and Nd-0 showed resistance to this pathogen is conferred largely by one gene, resistance to Pseudomonas syringae pv maculicola (RPM1), which was isolated by map-based cloning (Grant et al., 1995). Sequencing of the RPM1 gene in 26 Arabidopsis accessions revealed that resistant accessions had a functional allele while nonresistant accessions contained a large deletion spanning the gene. These two alleles showed high divergence in their flanking sequences, suggesting their long-term maintenance within the species, and the high divergence of resistance genes has been as evidence for an evolutionary This model that the dynamics of resistance genes selection in the plant to of polymorphism at however, indicate that balancing selection has maintained the resistant and nonresistant alleles for long evolutionary with a high of alleles as by the model et al., proposed a model in which advances and of resistant allele variation for resistance as a polymorphism et al., has the allele been ecological hypothesis is that is a associated with the resistance test this et al. a transgenic to the fitness associated with plant by RPM1 in plants with and the functional RPM1 gene in field and in the of a in seed number in plants with a functional RPM1 a high fitness associated with the maintenance of studies have to the possible of balancing selection on other resistance recent study the gene, which resistance to the fungal pathogen associated with the of the levels of polymorphisms of both the plant gene and the fungal gene an interaction by balancing selection et al., 2004). This study the of the molecular dynamics associated with interactions between species. major in plant resistance to insect QTL mapping studies in Arabidopsis have a major locus that underlies both the diversity of and resistance to and insect high levels provide resistance to but not to et al., consistent with the ecological hypothesis that by is but is by analysis for resistance genes showed that one of the major QTL for this trait harbors a gene family which determine the of the in the Molecular population genetic analysis of this family in accessions complex molecular variants, including evidence for gene loss and gene These data also the possible of balancing selection in this variation at et al., 2003), and may indicate that an ecological between the of these alleles and the resistance patterns to and insect is responsible for genetic diversity. The of flowering is a major life history transition in flowering plants and is to various including and (Koornneef et al., 2004). These vary with and evolutionary adaptation to these ecological would be expected to to latitudinal in flowering Arabidopsis is a over a wide latitudinal range in and variation in flowering time be expected to be associated with molecular polymorphisms in genes that flowering time in response to environmental A number of genes have been identified that flowering time in Arabidopsis. in several of including the pathway gene CRY2 (El-Assal et al., 2001; et al., 2004) and the gene FRI (Hagenblad et al., are known to underlie natural variation in flowering time in this species. variation in flowering time with has been in field experiments et al., 2004). This cline to be by an epistatic interaction between FRI and the gene which a (Caicedo et al., 2004). is a major ecological for plants. It affects diverse ecological including and flowering Maloof et al. (2001) of Arabidopsis in accessions and that this trait with the of of the from generally had The that Arabidopsis has to light at by being to A latitudinal cline in has been using but in and far-red light than light et al., 2002). ecological response to light is the which allows plants to to the of and is in by (Pigliucci, 2002). and the response in other accessions but no evidence of the that this response is a evolutionary (Pigliucci, 2002). The and in interaction and subsequent have been major in reproductive theory that and have in the of from the to its and pollen from multiple can in in one the interest is to for to the of by pollen from other plants. In the interest is to the among of the gene the molecular of these results in being only in the female which the interest by the of due to the of a gene et al., 2005). In interaction between female is observed among female has a of since of their on is after analysis using showed that the female was responsible for the of and of the female by more male pollen The of female to of bees, also results in genetic (Shimizu and and female interactions also underlie since species are and maintained by reproductive that evolution is often the for genes responsible for reproductive traits and in and 2004). It is thus that the gene family pollen also displays the phenotypic consequences of this rapid diversification have to be et al., 2004; et al., 2004). Most phenotypic of Arabidopsis traits, whether by natural variation among or QTL mapping are generally in controlled Although of these are to natural environments, they are A study by Hoffmann for example, showed that high and were not in in the natural range of this species. are necessary to plant phenotypes, it should be that controlled may be a of the environments observed in nature. It is that phenotypic traits and their underlying genetic bases may between natural and et al. for example, showed that the genetic for flowering-time variation in lines in controlled environments field in and Several of the QTL identified in environments were not in field suggesting that genes may have on flowering-time variation under these two field several novel were identified only in field and not in controlled environments. The of field is further by recent work that shows that a latitudinal cline in flowering time, associated with the epistatic interaction between FRI and can only be observed if the phenotypes are in field experiments (Caicedo et al., 2004; et al., 2004). Although to analyze to natural environments may have experimental and statistical methods to plant to complex environments. and for example, examined the of water on Arabidopsis and life not only the but also the of the water water are more of natural the of patterns in the water that were from under water including differences in which specific morphological or traits display to water also interactions between for example, that a water in of a and that was highest with a rate of low water this allow molecular to novel experiments such as and analysis in that the natural that plants Although has been in the last a between ecological molecular and evolutionary field to a and interaction among these fields remains There is a to these fields for two further understanding of evolutionary and ecological processes requires to the molecular underlying organismal phenotypes and It is only by relevant genes, the of variation at these and determining the functional consequences of such variation that we can to into the and history of ecological and evolutionary Molecular analysis can study of adaptive including that have occurred in the evolutionary Molecular has to for example, the ecology of resistance or the adaptive evolution of selfing during climate Molecular tools are also being in other species of Arabidopsis, which display a variety of ecological phenotypes and allow to extend studies to other ecological and evolutionary (Mitchell-Olds, 2001). understanding the ecological and evolutionary in which genes allows to a of their of genes are generally described in or physiological can also be however, in light of the ecological context in which they are or in their associations with evolutionary the power of ecological and evolutionary may to new genes underlying traits, such as flowering time, which may under standard in provides a for a more comprehensive of genome It is in both these that Arabidopsis to the power of genetics and genomics to further understanding of ecology and on one and of and on the and the for of the