The ecomolecular synthesis
Nucleotide changes in the genome must, from first principles, be the basis of adaptive evolution, but if studied in isolation they cannot give a useful description of the adaptive process. Adaptation occurs through the interaction of genotype with ecosystems and their components, including abiotic and biotic factors. Plant development mediates the interaction between genotype and environment. Unlike animal development, plant development involves a continual feedback between developmental processes and the environment, as transduced through the unfolding phenotype, giving rise to a multiplicity of end results, which together are termed phenotypic plasticity. Where the organismal development involves the iteration of many units over a long period, as in a tree like poplar, a great variety of end results are possible within the boundaries of developmental canalization.
We now have the means to combine molecular, developmental and ecological studies into a new understanding of the adaptive process: an ecomolecular synthesis. In some ways the ecomolecular synthesis may be seen as an extension of a historical process that began with Darwin. Darwin's theory of natural selection was organism-centred, necessarily so because of the prevailing inadequate understanding of genetics. It fell to the architects of the modern synthesis (Fisher, Sewell-Wright, Haldane and others) to combine Darwinian selection with Mendelian genetics, so moving our concept of selection down to the level of notional ‘allele’ frequencies.
However, there is now the prospect of shifting the study of natural selection down to the level of the individual nucleotide (e.g. Purugganan & Suddith, 1998), by understanding the effect of nucleotide variation on developmental processes in an ecological context (eco-devo). Understanding how genomes and ecosystems interact will lead to an ecomolecular synthesis (Cronk, 2001, 2002, 2004a). It is to be hoped that the combination of ecology and genomics may shed light on several crucial problems that have remained intractable under the pregenomic ‘modern synthesis’.
Within populations, gene sequences unfold under environmental feedback to give functional phenotypes that reproduce differentially, depending on their interaction with the environment. This leads to gene frequency change in natural populations, and so returns to a different starting set of DNA sequences: a cyclical ‘adaptive recursion’.
In most cases the adaptive recursion is only studied by looking at single parts, such as molecular evolution or organism–environment interactions. One study that has been able to examine a substantial chunk of the adaptive recursion is that of Jamaican click beetles (Pyrophorus plagiophthalamus) and their evolving luminescence (Stolz et al., 2003). This is an ideal system for studying the adaptive recursion for several reasons.
First, the beetles are polymorphic for an easily quantified phenotypic character: the colour of their emitted light, which varies from green to yellow-green to orange. Second, the link from gene to phenotype is both simple and well understood: specific amino acid substitutions in the enzyme luciferase change the wavelength of the emitted light. Third, the ecological function of the light emission is well known: to attract mates.
In a highly illuminating piece of work, Stolz and colleagues showed that there is a long-term adaptive trend towards orange in Jamaica, driven by natural selection. The click beetle system is an unusually elegant one and it is likely that many other adaptive systems will be much harder to study. This may be particularly true of plants. As plants are prone to greater phenotypic plasticity than animals, it is difficult to quantify the phenotype except in controlled growth chambers, and study of the genotype is complication by clonality. We therefore expect the link between phenotype and genotype to be generally less tractable.
Completely sequenced genome of poplar
The poplar genome consists of some 35 000 genes spread over 19 chromosomes (2n = 38) and c. 500 million base pairs (Mb). The complete genome sequence has recently been sequenced through a collaborative effort led by the US Department of Energy Joint Genome Institute (JGI) and the Oak Ridge National Laboratory, with postgenomic activities guided by the International Poplar Genome Consortium (IPGC). It marks a milestone in tree biology. It also marks a milestone in eco-devo, as this is the first ecologically important (‘keystone’) plant species (Brunner et al., 2004) to be sequenced and one that has numerous congeners around the northern hemisphere, many with significantly different morphological and ecological characteristics.
All the congeners have the same chromosome number and apparently similar genomes (Smith, 1943). They are also often interfertile and even wide crosses (between sections) are sometimes possible (Stettler, 1980). The existence of the complete genome sequence allows eco-devo studies to go beyond a few candidate genes to whole genome scans using high throughput approaches, thereby moving from ‘fishing’ to ‘draining the lake’.
Phylogenetic context of Populus studies
Poplar (like Arabidopsis) is a member of the clade ‘eurosid I’ in the recently updated Angiosperm Phylogeny Group (APG) system (APG, 2003). It is a member of the family Salicaceae in the order Malpighiales (which also includes families Violaceae and Malpighiaceae). Arabidopsis by contrast is a member of the ‘eurosid II clade’, being placed in family Brassicaceae in the order Brassicales (along with the Capparaceae and Tropaeolaceae).
The family Salicaceae (now including the Flacourtiaceae: Leskinen & Alstrom-Rapaport, 1999; Chase et al., 2002) consists of numerous woody, mainly tropical genera. The closest relatives of the genus Populus are the eight other ‘salicoid’ genera: Salix (including Chosenia), Olmediella, Bennettiodendron, Idesia (Fig. 1), Carrierea, Poliothyrsis, Itoa and Macrohasseltia (Table 1). The sister group of poplar is the clade comprising Salix and Chosenia.
|Genus||No. of species||Distribution|
|Chosenia[ = Salix]||2||Asia|
The salicoid genera form the phylogenetic context against which Populus morphology and evolution has to be viewed. All the genera are dioecious or monoecious except for Macrohasseltia, which appears to be hermaphrodite (M.H. Alford, pers. comm., 2004). Salix is cosmopolitan and Populus from the northern hemisphere, but the other genera are Asian with the exception of the Central American Olmediella and Macrohasseltia. All genera are predominantly insect-pollinated except the wind pollinated Populus. All genera except Salix and Chosenia have a perianth (although it is very small in poplar: Boes & Strauss, 1994; Kaul, 1995).
The evolutionary history and diversity of poplar
The first unequivocal records of poplar fossils are from the Eocene (Manchester et al., 1986). These fossils consist of leafy shoots with female catkins attached. Leaf records from the Cretaceous and Palaeocene are highly dubious and require careful checking, although there are somewhat plausible records from the late Palaeocene (Collinson, 1992). There are, however, numerous Eocene records, and by this period the genus seems to have become well established. There is also leaf-fossil evidence that by the Miocene plants referable to all the modern sections of poplar were present (Chaney & Axelrod, 1959; Mai, 1995). Eckenwalder suggested that the extant Populus mexicana forms its own section on the basis of some putatively primitive characters (Eckenwalder, 1977). Manchester et al. (1986) report fossil from the Eocene with similar leaf morphology to P. mexicana, which supports this view. Populus has a fossil history in Mexico known to date back to the Oligocene (Ramirez & Cevallos-Ferriz, 2000).
There are numerous problems associated with interpreting fossil material of Populus. The characters of salicoid (gland-tipped) teeth, camptodromous secondary venation, and elliptic, lanceolate or deltoid shape are not unique to Salix and Populus. Poor material has often wrongly been identified as poplar in the early literature (particularly by L. Lesquereux, 1806–1889). It is worth noting that material originally named as fossil Populus has subsequently been re-assigned to a diverse array of other genera (Cercidiphyllum, Celtis, Tilia, Cocculus, Viburnum, Ficus, Salix, Magnolia, Vitis, Betula, Quercus, Platanus, Celastrinites and Rhamnacidium) (Lamotte, 1952). A presumed female catkin originally identified as Populus sp. by E. W. Berry (1875–1945) was later thought to be a Taxodium cone (Lamotte, 1952).
Poplar leaves are notoriously heteromorphic and this has probably caused an unnecessary proliferation of fossil leaf species (Eckenwalder, 1980). Other problems are caused by the presence of similar leaves (including salicoid teeth) in closely related species of Salicaceae within genera such as Salix, Idesia, Poliothyrsis and Carrierea. Salix-like leaves from the Eocene of North America were found to be associated with cymose inflorescences terminal on shoots (Boucher et al., 2003), and thus classified in a new genus Pseudosalix (these characters are primitive in Salicaceae but are found in a much reduced form in Populus ilicifolia from Kenya).
Extant poplars are divided into three main groups: aspens, including white poplars, poplars in the narrow sense, and turangas (Table 2). The poplars are further divided into balsam poplars, aigeiros poplars and large-leaf poplars. In addition, it has recently been suggested that P. mexicana, which is otherwise classified in section Aigeiros, should be placed in its own section, Abaso (Eckenwalder, 1977). Recent molecular studies (Shi et al., 2001; Hamzeh & Dayanandan, 2004) suggest that a major division is between the aspens (section Populus) and the other sections.
|Turangas||Balsamiflua (Griff.) Browicz||Turanga Bunge||P. euphratica Oliv., P. pruinosa Schrenk|
|Tsavo river poplar||Balsamiflua (Griff.) Browicz||Tsavo (Jarn.) Browicz||P. ilicifolia (Engl.) Rouleau|
|Aspens (incl. White poplars)||Populus||Populus||P. tremula L., P. tremuloides Michx., P. alba L.|
|Balsam poplars||Populus||Tacamahaca Spach||P. trichocarpa Torr. & A. Gray, P. balsamifera L., P. maximowiczii Henry|
|Aigeiros poplars||Populus||Aigeiros Duby||P. deltoides Marsh., P. nigra L., P. fremontii S. Wats.|
|Large-leaf poplars||Populus||Leucoides Spach||P. heterophylla L, P. glauca Haines., P. wilsoni C.K. Schneid.|
Taking a broad species concept, there are estimated to be three extant species of turangas, 10 species of aspens, four species of aigeiros poplars and some 20 species of balsam poplars and five large-leaf poplars. The most species-rich group, the balsam poplars, diversified primarily in Asia. In interpreting these species numbers it should be noted that Populus is widely regarded as a taxonomically difficult genus, due to hybridization, species variation, leaf heteroblasty and clonal reproduction. Narrower species concepts result in greatly increased species numbers; for instance, the Flora of China (Wang & Fang, 1984) account includes 71 species and hybrids. The most extant diversity of poplar is in Asia, with only 10 species in North America, four species in Mexico, three species native to Europe and two species native to Africa.
The majority of poplars have wide, sometimes very wide, species distributions. These distributions often encompass markedly different climatic regimes, making poplar an ideal system for the study of climatic adaptation (Farmer, 1996). In some cases these distributions parallel each other. For instance P. deltoides occurs from Texas to Manitoba and P. trichocarpa from California to Alaska. Both therefore have a considerable latitudinal range and must cope with very different photoperiod conditions, thus forming ‘replicates’ for the study of environmental adaptation.
Poplars have often been used for the study of hybridization (e.g. Martinsen et al., 2001; Schweitzer et al., 2002). Hybrids are frequent within sections but rare between sections. An exception to this rule occurs between the aigeiros poplars and the balsam poplars (Eckenwalder, 1984; Floate, 2004), which are frequently interfertile. These intersectional hybrids often show great hybrid vigour as is evident in P. x generosa (P. deltoides×P. trichocarpa).
Potentially adaptive traits broadly characteristic of the genus Populus
Dioecy Like many of the other salicoid genera of the Salicaceae (Gunter et al., 2003; Semerikov et al., 2003), poplar is dioecious (although occasional monoecious individuals occur). However, unlike other salicoid genera, poplar is wind-pollinated. Populus is thus the first genome to be sequenced of a plant that is both dioecious and has relatively recently acquired characteristics of wind-pollination. Like many wind-pollinated plants its inflorescences are reduced to catkins. However, in this case the catkin does not appear to be a primary adaptation to wind pollination as its sister genus Salix, which is mainly insect-pollinated, also has inflorescences reduced to catkins.
Seed hairs The possession of long seed hairs is biologically significant both for the importance of this trait in dispersal and because of their economic importance in other species. Seed hairs in cotton (Gossypium spp.) and kapok (Ceiba spp.) form the basis of important industries. Thus, although Populus does not have economically important seed fibres, it may serve as a useful general model for seed-fibre developmental genetics, and an interesting comparison for functional studies when the complete genome of Gossypium becomes available.
Fast growth rate and heterosis
Many poplar species are adapted to conditions of abundant nutrients and unlimited water availability, and under such favourable circumstances achieve very high relative growth rates and consequently rapid biomass accumulation. However, even greater growth rates are achieved by F1 hybrids, which characteristically exhibit high degrees of hybrid vigour (or heterosis) (Li & Wu, 1997; Pearce et al., 2004). Rapid growth is a valuable agronomic trait in economically important plants and the phenomenon of heterosis has been used to enhance this trait. Poplar is therefore of great potential importance as a model for aspects of growth and vigour.
The flutter syndrome Many poplars, notably the aspens and aigeiros poplars, have leaves that hang loosely and flutter in a slight breeze. This phenomenon aids photosynthetic efficiency and contributes to fast growth rates (Roden & Pearcy, 1993). The flutter is associated with a syndrome of morphological characteristics (Fig. 2): long, laterally compressed petioles and leaves that are isobilateral (i.e. have similar, photosynthetically effective, upper and lower leaf surfaces: Russin & Evert, 1984). These isobilateral leaves are sometimes called ‘unifacial’ in the literature, but this is a misnomer as that term has a different, and very specific, meaning in plant morphology.
Phytochemistry Poplars defend themselves by the production of complex phenolic-based resins (Curtis & Lersten, 1974; Greenaway et al., 1992). The balsam poplars, as their name implies, are particularly important resin producers. This interesting aspect of plant chemistry now has the benefit of a completely sequenced genome to aid understanding of the enzymatic machinery involved in producing this phytochemistry.
Vegetative reproduction Extensive root suckering is of great ecological importance in the aspens as single clones can attain enormous proportions. A clone in Utah, named Pando by researchers, covers 43 ha (Mitton & Grant, 1996). Furthermore vegetative reproduction is possible at high altitudes and high latitudes, even when sexual reproduction is aborted because of low temperatures. By contrast the balsam poplars and the aigeiros poplars sucker less but their shoots (unlike those of aspens) have great rootability. This feature (shared with willows) allows clonal fragments washed downstream to root and grow. These highly divergent vegetative properties within a genus make poplar a possible model for the study of ecologically important reproductive and physiological traits.
Heteroblasty Poplars characteristically display a range of leaf form that varies according to developmental stage, associated with both position of the shoot on the tree and the sequence of leaves on a shoot: Curtis & Lersten, 1978; Eckenwalder, 1980; Thiebaut, 2000; Yuceer et al., 2003; Cronk, 2004b). This heteroblasty appears to be under hormonal control and provides an intriguing model system for studying the interaction between growth hormones and leaf developmental genes.
Potentially adaptive variation within and between Populus species
Leaf shape, indumentum and serrature Poplars are enormously variable in leaf characters. Much of this variation is heteroblastic varying according to growth stage. However, clones within species differ and it is clear that genetic variation exists (Wu et al., 1997). Leaves in different genets of P. trichocarpa sampled from across its range and grown at the University of British Columbia differ from broadly cordate to narrowly linear, from glabrous to hairy, and from subentire to deeply toothed. It is possible that this variation has ecological significance in relation to temperature and water relations, but there does not appear to be an obvious correlation with geographical distributions. One aspect of leaf ecophenotypy that may be a promising way to connect genotype and phenotype is fluctuating asymmetry (FA). The idea behind FA is that deleterious mutations may act though perturbing developmental pathways slightly. This perturbation, it is argued, is reflected in asymmetry of otherwise symmetrical organs. If the proponents of FA are right asymmetry can thus be used as a proxy for deleterious genetic variation. Recent studies have reported lower rates of herbivory on symmetrical vs asymmetrical leaves of oak (Quercus spp.) (Cornelissen & Stiling, 2005). The genomics tools now available for poplar make new approaches to FA possible.
Plant architecture Branch angle and branch outgrowth (apical dominance) are both unusually variable in poplar (Ceulemans et al., 1990; Bradshaw & Stettler, 1995). Extreme variants are known in the form of horticultural selections such as the erect ‘Lombardy poplar’ (P. nigra var. italica). Loss of apical dominance in aspen (P. tremuloides) caused by cutting stimulates a massive growth of root suckers, whereas this is not a marked phenomenon in other poplars.
Stomatal number and physiology A north–south cline of stomata density, apparently under genetic control, is evident in P. trichocarpa. (Gornall & Guy, 2002). This is potentially driven by adaptation to water relations (R. Guy, pers. comm., 2004). Clones also differ in their responsiveness to water stress. In particular, clones from wet areas sometimes have leaf conductances (stomatal apertures) that are relatively unresponsive to xylem pressure potential changes (Bassman & Zwier, 1991).
Tolerance of environmental stress: salt, cold, drought Turanga poplars are well known for their tolerance of extremely high salinities and it is likely that intraspecific variation for this trait will occur in other species. The wide geographical distribution of poplar species is an indication that such variation will occur. For instance, P. trichocarpa occurs from California to Alaska, experiencing wide differences in environmental conditions, especially temperature extremes, over that latitudinal gradient.
Dormancy, growth and photoperiodic response For the same reasons that we expect variation in stress tolerance, we would also expect considerable variation in phenological adaptation (stress avoidance). When transplanted to the equable maritime climate of the UK, the high-altitude species from the Himalayan region, P. glauca, does not leaf until the end of June, later than all other temperate trees at this site (Fig. 3). The closely related P. wilsonii from China leafs at the normal time. Other provenance patterns revealed by common garden experiments are well known. The genetics of bud-flush and bud set timing have been studied in a cross between a clone of P. trichocarpa from Washington State and P. deltoides from Texas (Frewen et al., 2000), which demonstrate very different phenologies when grown in common garden experiments. Phytochrome genes are possible candidates for involvement in adaptation to photoperiod and the diversity of phytochrome genes in poplar have been studied by Howe et al. (1998). Curiously Populus trichocarpa has no genes of the PHYE or PHYC/F subfamilies, although these are present in most other angiosperms that have been examined.
Trophic interactions Genetic variation exists for susceptibility to insect pests (Havill & Raffa, 1999) and fungal pathogens (Villar et al., 1996). In addition, hybrid poplars have been shown to act as sinks for herbivorous insects (Floate et al., 1997; Kalischuk et al., 1997). Trophic interactions and competitive interactions thus act as an ‘extended phenotype’ through which genomic constitution affects ecosystem structure and composition. The idea of ecosystems as the extended phenotypes of their constituent species has recently been championed by Whitham in a series of influential papers (e.g. Whitham et al., 2002).
Genetic variation in Populus
All poplar species are obligately outbreeding and have what appears to be exceptional genetic variation. The high individuality of poplar clones is due mainly to small changes in sequence, called single feature polymophisms (SFPs). These are either single nucleotide polymorphisms (SNPs), deletion insertion polymorphisms (DIPs), which when in coding regions usually consist of three base-pair differences, so preserving the reading frame of the protein, or short tandem repeat polymorphisms (STRPs). The wild individual of P. trichocarpa that has been sequenced (Nisqually-1) is typical in being highly heterozygous. Preliminary results from the sequencing project indicate that there is a heterozygous position (SFP) on average approx. every 100 base pairs in the genome (S. DiFazio, pers. comm., 2004).
Taking into account genotypic differences between clones, particularly in a geographically widespread species such as P. trichocarpa, we expect that the total SNP rate is higher still, perhaps one polymorphic position (PP) every 50 base pairs. If between-species variation is taken into account, to give total variation in the entire genus, then PP rates of 1 per 20 base pairs (a total of 25 million PP in the all species genome) are plausible. Only a tiny percentage of this vast reservoir of variation will be significant in determining phenotypic differences between individuals of a species and between species within the genus Populus. One of the greatest challenges of eco-devo is to sort significant from insignificant SFPs.
To assay variation across the geographic range of P. trichocarpa a standard population of 140 clones has been established at the University of British Columbia (UBC) and used for SNP discovery by ecotilling (Till et al., 2003; Comai et al., 2004). Tilling is an enzymatic SNP detection method that uses the enzyme CEL1 to cut DNA strands at mismatched base pairs. So far, surprisingly high levels of polymorphism have been detected in most (but not all) genes (E. Gilchrist, pers. comm.), which is consistent with results from P. tremula (Ingvarsson, 2004).
Eco-devo seeks to bring together molecular developmental genetics and ecology to determine how ecologically significant (and perhaps adaptive) traits arise in the genome and how they are selected for and maintained in populations or species.
In poplar it is easy to assay polymorphisms in any part of the genome using standard clone collections (SCCs, such as the UBC P. trichocarpa whole range geographical SCC). However candidate genes are needed to check whether there is a correspondence between gene and phenotype, either by functional experiment or association genetics. Ideally every gene in the genome should be considered a candidate gene for a given phenotype and studied accordingly. However, in the absence of high throughput techniques to study gene-phenotype correspondence in all genes in the genome simultaneously, some means must be found to narrow the field. Candidate genes can be obtained from orthology, reverse genetics, forward genetics, QTL mapping, microarray genotyping or expression studies.
Orthology, as determined by gene phylogenies, is a powerful way of connecting new research organisms like poplar to functional studies that have already been carried out in established model organisms. If a particular gene is shown to affect a trait in, for example, Arabidopsis, the orthologue of that gene is a good candidate for the same trait in poplar. On the other hand, the poplar orthologue may have acquired a different function during the long evolutionary divergence from other model plants. The Arabidopsis LEAFY gene and its poplar orthologue PTLF provide an example of such apparently divergent function (Rottmann et al., 2000). The prediction of function from orthology (e.g. Citerne et al., 2003) is increasingly termed ‘phylogenomics’.
Reverse genetics involves the production of mutagenised populations (such as transposon insertion mutagenesis) hoping that a knock-out of a gene will occur that will affect a particular trait. This has proved a very powerful tool in annuals such as Arabidopsis and Antirrhinum but it lacks power in poplar because of the further breeding required to make the recessive knock-out mutation homozygous. The breeding problem also limits the power of using naturally occurring variation to determine QTLs and genetically mapping these to candidate genes on the chromosomes, although major QTL studies have already been carried out (e.g. Frewen et al., 2000).
As an alternative, forward genetics have been used to activate genes (activation tagging), so creating dominant mutations that are immediately visible (Busov et al., 2003). This technique is enormously powerful and particularly well suited to poplar, in which selfing to create homozygous recessives is impossible.
Microarrays provide a further means of generating candidate genes. Short oligomer arrays such as those manufactured by Affymetrix have the ability to detect single base pair differences between the array and the probe. This has allowed the use of such arrays to scan whole genomes for SFPs and for extreme array mapping (XAM) in Arabidopsis (Borevitz et al., 2003; Wolyn et al., 2004).
Microarrays are more commonly used for expression studies. These may take the form of either ‘comparative’ or ‘response’ approaches. In comparative approaches RNA pools representing individuals having genetically different phenotypes are used to generate probes (e.g. small-leaved vs large-leaved plants). In response approaches, genes responsive to environmental factors are detected. For instance, genes up-regulated in response to drought are candidate genes for adaptation to drought.
Studies of this sort will generate huge numbers of candidate genes (although, it is to be hoped, far fewer than the full complement of c. 35 000). High-throughput techniques are needed for performing association tests on natural variation, or for performing functional studies to rule in or out candidates. Prerequisite tools are: first excellent, well-phenotyped SCCs, carefully chosen for specific hypotheses and established in common gardens and (preferably) multiple garden trials; and second high-throughput SFP detection pipelines for whole genome genotyping using microarrays or for the genotyping of large sets of candidate genes in the SCCs. At present the biggest bottleneck appears to be the first of these, and a great deal of fieldwork needs to be done. For eco-devo using poplar it seems that laboratory work and field work will be inextricably linked for the foreseeable future, which is to be welcomed.
Such dual faceted studies have the potential to usher in a new era of the study of the genotype in relation to environment.