Introduction

The invasion of land by plants and animals that are geographically removed from their native ranges is a common biological phenomenon that has occurred repeatedly since such organisms first became terrestrial. In recent years, much emphasis and interest has focused on invasions resulting from human activities following the accidental or intentional introduction of plants and animals to new sites. Some of these introductions have established successfully in their new surroundings, become invasive, and spread rapidly from points of introduction. Sometimes this has had an adverse effect on native components of the flora and fauna in the area invaded (Pimentel 2002; Gurevitch and Padilla 2004). On other occasions, the main effect has been to enrich the flora and fauna of a particular region (Sax and Gaines 2003; Walker 2007). The latter is particularly true in areas that were glaciated during the Pleistocene as illustrated by the flora of Britain and Ireland. According to Preston et al. (2002), just half the species that comprise this flora are native while the remainder is introduced, mainly within the past 500 years. Much of Britain and Ireland was covered by a thick ice-sheet as recently as 18,000 years ago, at the time of the Last Glacial Maximum, and thus virtually all of the current British and Irish Flora, apart from some arctic-alpine elements, is of relatively recent origin and the product of biological invasion. The same is likely to be true of floras in many other regions of the world that were subjected to glaciation during the last ice age (Brochmann et al. 2003).

Studying a species that recently invaded a new territory allows one potentially to reconstruct events leading to its successful establishment and spread, and the factors that were important in this process. If there is a sufficiently good historical record of when a species was first introduced and how it subsequently spread, it is possible to determine, for example, whether there was a lag phase between introduction and spread (as is often the case), the speed at which the species spread following establishment, and what effects it has had on other components of the local flora and fauna. There is also the possibility of a detailed analysis of the source of introduced material to determine its genetic composition, and evolutionary changes that occurred during establishment and spread across ecological gradients (Lee 2002; Rieseberg et al. 2007). Consequently, it is possible to determine if rapid evolutionary change occurred in the invasive populations and to reveal which particular traits promoted invasiveness. It is not surprising, therefore, that invasive species are increasingly used as models for the study of rapid evolutionary change (Lee 2002).

A particularly interesting finding in recent years has been the importance of hybridization in plant invasions. From a review of the literature, Ellstrand and Schierenbeck (2000) reported that invasiveness is frequently preceded by hybridization either between different varieties, races or subspecies of a species, or between different species. Provided that the hybrids are not sterile, it is expected that hybridization between genetically divergent stocks, followed by recombination and segregation, will generate a wide array of recombinant genotypes on which selection can act. These recombinant genotypes may provide the means for adaptation to changed conditions during establishment and also at later stages as material spreads widely from its initial site(s) of establishment (Lee 2002). The importance of hybridization will extend further if, during its invasive phase, the species hybridizes with other components of the introduced or native flora, and acquires genes through introgression that provide adaptations to local conditions (Milne and Abbott 2000; Abbott et al. 2003). Harlan and de Wet (1963) were first to promote this idea and termed such species ‘compilospecies’, from the Latin compilo, to plunder or to rob (see also Fuertes Aguilar et al. 1999). Additionally, hybridization involving an invasive species might lead to the origin of new hybrid taxa at the homoploid level (Rieseberg et al. 1991; Abbott 1992; Bleeker 2003) or via allopolyploidy following chromosome doubling (Abbott 1992; Ashton and Abbott 1992). On the other hand, outbreeding depression and/or gene swamping could be experienced by a native species that hybridizes with an invading species, leading possibly to its extinction and loss of native biodiversity (Bleeker et al. 2007).

It is often considered that successful establishment of an introduced species following long distance dispersal will be enhanced by uniparental reproduction, and this has been termed Baker’s Law (Baker 1955; Stebbins 1957). However, there are many exceptions to this rule (Lane and Lawrence 1993; Radford 1997; Sun and Ritland 1998) including species with self-incompatibility (SI) systems such as Senecio squalidus (Abbott and Forbes 1993; Hiscock 2000a, b). Until recently, there had been no detailed analysis of the population genetics of SI in an invasive species. However, this deficiency has been rectified recently by an analysis of the population genetics of SI in the Oxford ragwort, Senecio squalidus (Brennan et al. 2002, 2003a, b, 2006).

Senecio squalidus is a particularly interesting invasive species, in that it originated recently in the British Isles from hybrid material (Abbott et al. 2000; James and Abbott 2005), it exhibits sporophytic SI (Hiscock 2000a, b), and it has hybridized with a native species to give rise to two new self-fertile hybrid species and one stabilized introgressant form of a native self-fertile taxon (Abbott et al. 2003; Abbott and Lowe 2004). In this paper, we review the evidence for these different phenomena and consider evolutionary changes that may have occurred in Oxford ragwort since its origin in the British Isles.

Oxford ragwort in the British Isles: introduction and spread

Oxford ragwort, S. squalidus, is a short-lived, perennial herb derived from material introduced to Britain from Mount Etna, Sicily, at the beginning of the eighteenth century (Harris 2002). A reconstruction of the history of this event suggests that plants were first grown at the Duchess of Beaufort’s Garden at Badminton, before being cultivated at the Oxford Botanic Garden, sometime before 1719, where they have been in continuous cultivation ever since. Linnaeus described the species in 1753 from plants raised by him from seed sent from Oxford (Walker 1833; Harris 2002). In 1794, the species was reported growing on walls in the Oxford area (Sibthorp 1794) and was described as a garden escape. The subsequent spread of the species from Oxford was aided initially by exchange of seeds between botanic gardens, but in the late nineteenth century, following the establishment of the railway system in the British Isles, S. squalidus began spreading rapidly from Oxford via the railway network. Druce (1927) pointed out that railways provided the species with a substrate similar to the lava soils that it grows on in Sicily. More recently, the species has spread along motorway verges in Britain, and is now common on waste-ground especially in urban areas. The spread of the species in the British Isles was chronicled by Kent (1955, 1956, 1957, 1960, 1963, 1964a, b, c, 1966). Records show that it began to spread northwards in the late nineteenth century and reached different parts of northern England between the early to middle of the twentieth century (Fig. 1). The species became established in the wild in the Central Belt of Scotland during and after the mid-1950s (Kent 1955; 1966), and has continued to spread further north in Scotland and into Northern Ireland in recent years (Fig. 1).

Fig. 1
figure 1

The spread of Senecio squalidus in the British Isles from Oxford where it was first recorded in the wild in 1794. Maps for 1820–1930 are redrawn from Crisp (1972) and are based on the presence (black) or absence (white) of the species in Vice-Counties of the British Isles. Maps for 1962, 1988 and 1999 were obtained from the Biological Records Centre, Centre of Ecology and Hydrology, with each black dot representing at least one record of the species in a 10 km square of the National Grid

Full size image

Hybrid origin of S. squalidus

Druce (1927) and Kent (1956) described S. squalidus as introduced from Sicily, Italy. Crisp (1972) later proposed that the species originated in the British Isles from hybrids between the species S. aethnensis Jan. ex DC and S. chrysanthemifolius Poiret which grow at high and low altitudes, respectively, on Mount Etna in Sicily. Crisp (1972) advanced that following almost a century of cultivation in the Oxford Botanic Garden, stabilized derivatives of this introduced hybrid material escaped and spread rapidly to many parts of the British Isles. He concluded that “British S. squalidus is of hybrid origin…” and “…can be treated as a separate species… because it is both geographically isolated from the parental species, and it has evolved over…. many generations to a state where it is morphologically distinct from either of them, although still polymorphic and in general intermediate between them.” The hypothesis that S. squalidus is a diploid hybrid derivative has since been tested by surveys of morphological and isozyme variation (Abbott et al. 2000, 2002), and more recently by a detailed survey of Randomly Amplified Polymorphic DNA (RAPD) and inter-simple sequence repeat (ISSR) variation (James and Abbott 2005) in S. squalidus, its two putative parents, and hybrid plants from Mount Etna.

In the wild, Senecio aethnensis produces large capitula and entire, glaucous leaves, while S. chrysanthemifolius has smaller capitula and highly dissected, non-glaucous leaves. These differences are retained in plants raised under ‘common garden’ conditions (Abbott et al. 2000). A principal component analysis of 32 morphological traits measured on plants raised from seed in a glasshouse (Abbott et al. 2000) showed that S. squalidus closely resembles plants derived from the hybrid zone on Mount Etna, and is morphologically intermediate to its putative parents. Interestingly, this morphological intermediacy is due to plants possessing a mix of traits that distinguish the two species, rather than possessing characters with means intermediate to those of the two parents. Two samples of S. squalidus were examined in the study—one from Oxford and the other from Edinburgh. The two samples differed in mean for only two characters—total floret number and disc floret number per capitulum. In Edinburgh material, there was no evidence for the mean of any character measured being either significantly higher or lower than the means recorded for the same character in both putative parent species. Thus there was no evidence that transgressive segregation (Rieseberg et al. 1999) had generated morphological novelty in this material. In contrast, there was evidence of transgressive expression in morphology in Oxford material, but for only 3 of the 32 characters measured (one capitulum trait and two leaf characters). The fact that most morphological characters in S. squalidus do not appear to exhibit transgressive expression is interesting because transgressive segregation is reported to occur frequently in hybrid derivatives (Rieseberg et al. 1999), and has been suggested to be important in generating novel traits that could aid the adaptation of hybrids to new habitats from which parent species are excluded (Rieseberg et al. 2003; Lexer et al. 2003; Gross and Rieseberg 2005). Notwithstanding this, it is entirely possible that S. squalidus exhibits transgressive change in metabolic and physiological characters relative to its parents that have facilitated its colonization of Britain. This hypothesis is supported by studies of gene expression in S. squalidus using microarrays, which have revealed transgressive changes in the expression patterns of a number of metabolic genes relative to their expression in S. chrysanthemifolius and S. aethnensis (Hegarty et al. 2009).

Morphological analysis provides only an indication of whether a species is of hybrid origin. More substantial evidence requires molecular analysis (Rieseberg and Ellstrand 1993; Chapman and Abbott 2005). A survey of allozyme variation provided some initial molecular support for S. squalidus being of hybrid origin (Abbott et al. 2000, 2002) in that some British populations of S. squalidus were genetically similar to material collected from the hybrid zone between S. aethnensis and S. chrysanthemifolius on Mount Etna. In particular, at one allozyme locus (Acp2), S. squalidus and hybrids were both polymorphic for two alleles, one of which was fixed in S. aethnensis, while the other was fixed in S. chrysanthemifolius. Molecular confirmation that S. squalidus is of hybrid origin recently came from a survey of RAPD/ISSR variation (James and Abbott 2005). This showed that S. squalidus contained 11 of 13 RAPD/ISSR markers recorded at high frequency in S. chrysanthemifolius, but which were absent or occurred at low frequency in S. aethnensis, and 10 of 13 markers for which the reverse was true. Bayesian admixture analysis using the computer program STRUCTURE (Pritchard et al. 2000) showed that all individuals of S. squalidus surveyed (72 individuals from 6 populations) were of mixed ancestry with relatively high mean proportions of ancestry derived from both S. chrysanthemifolius and S. aethnensis (0.644 and 0.356, respectively). Populations varied significantly in mean S. chrysanthemifolius ancestry ranging from approximately 50–75% (Fig. 2). Although of hybrid origin, further analysis by principal coordinate analysis showed that each S. squalidus individual possessed a multilocus RAPD/ISSR phenotype clearly different from those possessed by nearly all individuals screened from the hybrid zone on Mount Etna and that individuals clustered as a separate group. Moreover, all individuals of S. squalidus only possess the cpDNA haplotype that characterizes S. chrysanthemifolius, and never the one also found in S. aethnensis, whereas hybrids on Mount Etna may possess either one of these haplotypes. This is evidence, therefore, that S. squalidus has a unique genetic make-up, and should not simply be considered as an indistinguishable part of the hybrid swarm between S. aethnensis and S. chrysanthemifolius on Mount Etna that has been displaced to the British Isles. Crosses between S. squalidus and its parents usually produce vigorous and fertile F1 and F2 offspring under glasshouse conditions (R. J. Abbott, persernal observation). Consequently, there is little evidence of intrinsic postzygotic isolation between S. squalidus and its parent taxa under these conditions. However, it is hypothesized that reproductive isolation between the hybrid species and its parents is maintained by habitat isolation aided by spatial separation. Transplant studies are currently being conducted to test this hypothesis.

Fig. 2
figure 2

Mean proportion of S. chrysanthemifolius ancestry in six populations of S. squalidus in the British Isles (from James and Abbott 2005), and estimates of genetic diversity in these populations, based on a survey of RAPD/ISSR variation

Full size image

Population genetic structure

Allozyme variation

An examination of the population genetic structure of an invasive species may help to establish how the distribution of variation within and among populations has been affected by such factors as: (a) founder effects during the initial colonization phase; (b) range expansion; (c) genetic drift and gene flow; (d) mating system; and (e) selection on non-neutral variation.

A survey of allozyme variation for six enzymes representing eight loci within and among nine populations of S. squalidus in the British Isles showed that genetic diversity (measured in terms of mean number of alleles per locus, percentage of polymorphic loci, and gene diversity within populations (H E) was lower in S. squalidus than in either of its parent species, S. aethnensis and S. chrysanthemifolius, or in hybrid populations from Mount Etna (Table 1; Abbott et al. 2000). This indicates that during the introduction of hybrid material to the British Isles and/or during the subsequent origin of S. squalidus, only a proportion of the genetic diversity present in Mount Etna material became incorporated into that now present in Britain. Seven out of the nine populations were at Hardy–Weinberg (H–W) genotype frequency equilibrium. Thus in most populations random mating can be assumed. Typical random mating within populations as measured by H–W equilibrium has been confirmed by an additional allozyme study investigating a different set of eleven populations for twelve loci (Brennan 2003a, b; Brennan et al. 2005). For both studies, significant genetic differentiation between populations was detected for allozyme variation based on F ST estimates (F ST = 0.09–0.385; significantly >0 at a 95% confidence level, Table 1). However, there is little obvious geographical pattern to population differentiation. The allozyme survey based on nine populations did not detect significant isolation-by-distance (Mantel’s test; R 2 = 0.11, P = 0.28), while that based on eleven populations resolved a pattern of isolation-by distance but the correlation only explained a small proportion of the variation (Mantel’s test; R 2 = 0.19, P < 0.01). It seems likely that population differences in allozyme variation are largely the product of founder effects and genetic drift occurring during and following the colonization of new sites, and that the recent range expansion of S. squalidus in the British Isles has prevented the accumulation of much isolation-by-distance between populations.

Table 1 Estimates of genetic diversity in Senecio squalidus based on surveys of variation over (a) 21 RAPD/ISSR loci, (b) eight allozyme loci (Abbott et al. 2000), and (c) 12 allozyme loci (Brennan 2003a, b)
Full size table

RAPD/ISSR variation

The results of a survey of RAPD/ISSR variation conducted on 12 individuals from each of 6 populations of S. squalidus sampled from different parts of England, Wales and Scotland (Fig. 2) were used to demonstrate that S. squalidus is of hybrid origin (see above) and to examine its population structure (James and Abbott 2005). However, here we restrict population genetic analysis to 21 of the 26 loci previously examined. At these 21 loci a band was present in 1 or more S. squalidus individuals examined, whereas at the other 5 loci no band was recorded in any individual tested. Analysis shows that each of the 72 individuals examined possessed a unique RAPD/ISSR phenotype and that each population contained a relatively high level of genetic diversity (Fig. 2). Mean percentage of polymorphic loci and mean gene diversity within populations (H E) were considerably greater than the same estimates of diversity based on allozyme variation (Table 1). An analysis of molecular variance (AMOVA) revealed that most RAPD variation across populations was due to differences between individuals within populations (79%), although variation between populations (21%) was significant. A Mantel test (Mantel 1967) showed that the correlation between pairwise genetic distances between populations (in terms of PhiPT) and geographical distances among populations, although higher than that recorded for allozyme variation (see above), was not significant (R 2 = 0.320, P = 0.092; Fig. 3). These results also suggest that genetic differences between populations result from founder effects and genetic drift and that the lack of an association between genetic and geographical distance between populations is most likely due to the recent rapid range expansion of S. squalidus in the British Isles (Fig. 4).

Fig. 3
figure 3

The number, identity and distribution of self-incompatibility (S) alleles in S. squalidus (from Brennan et al. 2006)

Full size image
Fig. 4
figure 4

A plot of pairwise genetic distance (based on PhiPT values) against pairwise geographic distance (km) between six populations of S. squalidus surveyed for RAPD/ISSR variation and described in Fig. 2. The slope of the regression line and R 2 value were calculated using a Mantel test with the software GenAlEx6 (Peakall and Smouse 2006)

Full size image

Mating system

According to Baker’s Law (Baker 1955), successful colonizing and invasive plants are usually self-fertile (self-compatible (SC)) (Stebbins 1957; Baker 1967). Studies of the mating system of S. squalidus, however, have shown that individuals exhibit strong SI across its entire British range and, as in other species of Asteraceae, this SI is regulated sporophytically by a single polymorphic S locus (Abbott and Forbes 1993; Hiscock 2000a, b; Brennan et al. 2002, 2005, 2006). Baker’s Law is clearly not ‘set in stone’ and there are other examples of SI species becoming successful colonists, (e.g., Papaver rhoeas (Lane and Lawrence 1993), Senecio madagascariensis (Radford 1997) and Centaurea solstitialis (Sun and Ritland 1998)), but the finding of strong SI in S. squalidus is intriguing, particularly in the light of the extreme population bottleneck that its ancestors must have experienced during its introduction and early colonization. Following a population bottleneck, allelic diversity at the S locus will be lowered and opportunities for mating (between individuals carrying different S alleles) correspondingly reduced (Hiscock 2000b; Brennan et al. 2002). Under such conditions, selection for reproductive assurance frequently leads to an erosion of the SI system and the evolution of self-compatibility (SC) (Reinartz and Les 1994; Cheptou et al. 2002; Nielsen et al. 2003). This has not happened in S. squalidus indicating that either the numbers of S alleles in British populations are sufficient to support effective mating and thereby maintain strong SI, or that other mating system characteristics (e.g., long flowering season, and generalist pollinators) can compensate for reduced allelic diversity at the S locus.

An extensive survey of the number and frequency of S alleles in populations of S. squalidus across the British Isles revealed that there are approximately seven S alleles in the entire British population (Brennan et al. 2006). Local populations maintain subsets of these 6–7 S alleles (mean 5.3), with the Oxford population maintaining a majority (6), in line with it being the center of introduction (Brennan et al. 2006; Fig. 3). These numbers of S alleles are very low when compared to population level estimates for other SI species which typically maintain ~10–43 S alleles (Lawrence 2000; Brennan et al. 2003b).

In species with SSI, such as S. squalidus, the incompatibility phenotype of the haploid pollen is controlled by the diploid genome of its parent plant, so pollen grains carry products of two S alleles as opposed to one (c.f. gametophytic SI). When S alleles are codominantly expressed, this increases the efficiency of kin recognition up to 50%, although this may also have the effect of further limiting the number of compatible mates in populations with few S alleles (Hiscock and McInnis 2003). However, with SSI, but not GSI, dominance relationships among S alleles are possible that mask recessive S allele phenotypes and introduce a degree of control over compatibility levels within populations independent of S allele number (Hiscock and Tabah 2003; Hiscock and McInnis 2003). Mate availability increases with greater levels of S allele dominance interactions with most improvement when there are few S alleles (Byers and Meagher 1992; Vekemans et al. 1998; Hiscock and Tabah 2003). Interestingly, populations of S. squalidus support these predictions, with few S alleles and high levels of dominance interactions. Comparisons between the limited number of population level studies of S allele dominance indicate that there is a negative correlation between S allele number and frequency of dominance (Brennan et al. 2003b). This suggests that the observed high frequency of dominance interactions among S. squalidus S alleles serves to optimize mate availability in these S allele-depleted populations and facilitates reproductive assurance during colonization (Brennan et al. 2006). Whether these high levels of dominance interactions represent a chance ‘pre-adaptation’ already present in the founding ancestors of S. squalidus, or whether dominance evolved among S alleles after the introduction event is being determined through studies of SI in populations of S. squalidus’ relatives on Mount Etna (A. C. Brennan, S. J. Harris and S. J. Hiscock, in preparation).

Populations of S. aethnensis, S. chrysanthemifolius and their hybrids all maintain relatively large numbers of a diverse pool of S alleles, with estimates ranging from between 9 and 26 S alleles per population in line with ‘typical’ population S allele numbers for species with SSI (Lawrence 2000). The frequency of individuals exhibiting S allele dominance interactions in the Mount Etna populations appears relatively high, but lower than the frequency observed in British S. squalidus (A. C. Brennan, S. J. Harris and S. J. Hiscock, in preparation) supporting the hypothesis that the high frequencies of dominance interactions among S alleles in S. squalidus may be a relatively recent phenomenon that evolved as a consequence of establishment and invasiveness in Britain. The genetic basis of dominance among S alleles in S. squalidus is not known, but work on SSI in Brassica indicates that dominance is regulated epigenetically (Shiba et al. 2002) through reversible DNA methylation (Shiba et al. 2006). Epigenetic control of S allele dominance interactions clearly offers the potential for dominance levels to be highly responsive to selection. For now, we can only predict that dominance among S alleles in Senecio may have an epigenetic basis, because the molecular basis of SSI in S. squalidus is different to that found in Brassica, and is as yet unresolved (Hiscock et al. 2003; Hiscock and Tabah 2003; Tabah et al. 2004).

Despite strongly expressed SI in populations of S. squalidus across Britain, it is possible to identify a few individuals with ‘leaky’ SI that set small numbers of seeds after selfing within most populations (Brennan et al. 2005). This phenomenon has been observed in many SI species, especially in wild populations, and is usually referred to as ‘pseudo-self-compatibility’ (PSC), since while individuals are able to produce a few seeds after self-pollination, the number of seeds produced is far lower than that produced following a compatible cross-pollination (Levin 1996; Stephenson et al. 2000). PSC is a quantitative trait controlled by variable numbers of ‘modifier loci’ that modulate the strength of SI and which are unlinked to the S locus (Vogler et al. 1998; Stephenson et al. 2000; Vogler and Kalisz 2001).

In S. squalidus, 3.1% of all individuals sampled across the UK were classified as PSC (Brennan et al. 2005). These individuals produced 1–10 seeds (counted as the number of fruits or achenes) per capitulum after self-pollination, compared to 40–80 seeds per capitulum following a cross-pollination. It is likely that PSC has contributed to the success of S. squalidus as a colonist, by providing a basis for limited uniparental reproduction, because the frequency of PSC varies between populations (ranging from 0 to 8% of individuals), suggesting that PSC is sensitive to selection. When populations are small and mating opportunities are few (e.g., due to reduced S allele diversity), modifiers that weaken SI may be at a selective advantage because they deliver reproductive assurance, whereas in larger populations, modifiers which impart PSC may be at a selective disadvantage because they are associated with inbreeding depression; the latter has been shown in Crepis sancta (Cheptou et al. 2002). Indeed, an investigation into rates of selfing and associated inbreeding depression in S. squalidus revealed high levels of inbreeding depression associated with ‘forced’ selfing (induced by self-pollinating stigmas treated with dilute salt solution, Brennan et al. 2005) indicating that selfing through PSC carries a fitness cost upon which selection can act. The balance between the selective forces of inbreeding depression (maintaining SI) and reproductive assurance (favouring PSC) is likely to be critical in maintaining equilibrium between SI and PSC in S. squalidus populations. One could therefore predict that the frequency of PSC might be higher in pioneer populations experiencing metapopulation dynamics compared to more stable populations, but currently the empirical data are inconclusive (Brennan et al. 2005, 2006) because all UK populations so far studied appear to function as metapopulations (A. C. Brennan and S. J. Hiscock, unpublished). More ecological studies of PSC are therefore needed to test this hypothesis rigorously.

Expression of PSC in S. squalidus appears to be determined by at least three genetic loci and its strength can be affected by specific environmental factors, such as the presence of salt at the stigma surface (Tabah 2004; Brennan et al. 2005). Interestingly, regular exposure to salt is often a feature of many habitats of S. squalidus, particularly roadsides and to a lesser extent railway lines, suggesting that environmentally-induced PSC may also have contributed to the colonizing success of S. squalidus (Hiscock 2000b; Brennan et al. 2005).

Studies of mating system variation in natural populations of a wide range of outcrossing species, including S. squalidus, are now highlighting the fact that SI is not simply a qualitative trait (the S locus) because it can also be subject to variable levels of quantitative control through the action of modifiers (Vogler et al. 1998; Stephenson et al. 2000; Vogler and Kalisz 2001; Brennan et al. 2005). Species like S. squalidus, with a high degree of quantitative control over SI will have greater reproductive flexibility during population disturbances or metapopulation dynamics and may therefore make better colonists or become invasive. The extent to which the SI system of S. squalidus has been modified by its population history in the UK or was pre-adapted for colonizing success on account of its hybrid origin on Mount Etna will become clearer when studies of the SI mating system of its parental taxa and their hybrids on Mount Etna are complete. Evidence so far suggests that ‘flexibility’ within the mating system of S. squalidus in the form of variable levels of PSC and perhaps S allele dominance has probably contributed to its success as an invasive colonist.

Evolutionary change

Invasive species are increasingly recognized as suitable models for the study of rapid evolutionary change (Lee 2002; Rieseberg et al. 2007) and a number of studies have already shown that rapid evolution can take place when such species spread across ecological gradients (e.g., Weber and Schmid 1998; Huey et al. 2000). The fact that S. squalidus occurs in a climatic zone (temperate and oceanic) very different from that of its parents (Mediterranean) would suggest that significant evolutionary change might have occurred during the origin and early establishment of S. squalidus, aiding its adaptation to conditions in the British Isles and enhancing its ecological isolation from its parents. Moreover, further evolutionary change may have occurred as the species spread northwards from the warmer and drier climate in the south of the British Isles to the colder and moister climate in the north. Currently, there is no information on the nature and extent of possible adaptive divergence in S. squalidus in the British Isles. Clearly, an excellent opportunity is presented to obtain such information and to determine which characters and genes have been most influenced by selection. To this end, genomic screens and microarray analyses are currently being conducted in our laboratories to identify those parts of the genome, which may have been subjected to selection during the origin and subsequent establishment of S. squalidus in the British Isles. Others (John Pannell, University of Oxford, personal communication) are conducting reciprocal transplant experiments to establish if populations sampled from the south and north of the British Isles show local adaptation to environmental conditions at their respective locations.

Interspecific hybridization with native taxa and subsequent evolution

During its recent spread throughout a large part of the British Isles, S. squalidus has hybridized with the native species, S. vulgaris L. (Common Groundsel) (Marshall and Abbott 1980), and this has led to the origin of two new species, S. cambrensis Rosser and S. eboracensis Abbott & Lowe, and also the origin of a stabilized introgressant form of S. vulgaris, i.e., var hibernicus Syme. These events have been reviewed in detail elsewhere (Abbott and Lowe 1996, 2004; Abbott et al. 1998, 2005) and therefore will only be summarized briefly here.

Origin of S. cambrensis

Senecio cambrensis is the allohexaploid (2n = 6x = 60) of diploid S. squalidus (2n = 2x = 20) and tetraploid S. vulgaris (2n = 4x = 40), which originated in North Wales, in the early to mid-part of the twentieth century (Rosser 1955; Ashton and Abbott 1992). Isozyme and chloroplast DNA evidence shows that there were at least two independent origins of the species in the UK: at least one in North Wales, and another in Edinburgh, Scotland (Ashton and Abbott 1992; Harris and Ingram 1992). Senecio cambrensis is SC and reproduces mainly by self-fertilization (Ingram and Noltie 1995). It exhibits a metapopulation structure in the wild with subpopulations frequently becoming extinct and new ones establishing through colonization. The Edinburgh lineage, which was discovered in 1982 (Abbott et al. 1983), is now extinct in the wild (Abbott and Forbes 2002) due to sites being built on where it occurred, and a massive reduction in number of alternative potential sites for colonization from the early 1990s onwards (Abbott and Forbes 2002). The species remains present in north Wales (Ingram and Noltie 1995; Abbott and Lowe 2004) where its populations fluctuate greatly in size between years and locations (Ingram and Noltie 1995). However, there has been a notable decline in both number of populations and individuals per population in north Wales over the past 25 years (Abbott et al. 2007). This recent decline has occurred despite the fact that S. cambrensis contains high levels of genetic variability (Abbott et al. 2007) and also exhibits large-scale alterations in its gene expression relative to that of its parents and the triploid hybrid S. X baxteri (Hegarty et al. 2005, 2006). It might be expected that both of these factors would increase the ability of the species to adapt to conditions different from those experienced by its parents, and thus favor its maintenance (and possible spread) in the wild. However, its recent decline in north Wales is probably due to a marked loss of suitable sites for colonization and also to increased use of herbicide to control weed populations (Abbott et al. 2007).

Senecio eboracensis

Senecio eboracensis was first recorded in the British Isles in 1979 (Lowe and Abbott 2003). It is thought to have originated shortly before its discovery in York, England, as a tetraploid product of hybridization between S. squalidus and S. vulgaris. The hybrid status of S. eboracensis is evident from its intermediate morphology and its possession of an additive isozyme profile (Irwin and Abbott 1992). It is prezygotically isolated from both parent taxa due largely to its predominant self-fertilization in the wild (Lowe and Abbott 2004). However, its maintenance in the wild is currently in doubt. Following its discovery in York in 1979, populations of the species were maintained at two sites in the city until the late 1990s. Both of these sites have been built on in the last few years and also sprayed at regular intervals with herbicide, and since 2,000 S. eboracensis has not been found at either site or elsewhere in York.

Senecio vulgaris var. hibernicus

The inland radiate form of S. vulgaris (var. hibernicus) in the British Isles, originated following introgressive hybridization between the discoid form of S. vulgaris (var. vulgaris) and S. squalidus (Ingram et al. 1980; Abbott et al. 1992). The radiate form has flower heads (capitula) with ray florets, which are lacking in the discoid form. Presence/absence of ray florets is controlled by a single gene with var. hibernicus homozygous for the ray allele and var. vulgaris homozygous for the alternative allele (Trow 1912). There is molecular evidence that the ray allele was introgressed into S. vulgaris from S. squalidus to give rise to var. hibernicus (Abbott et al. 1992; Kim et al. 2008). The radiate variant was first recorded in 1832 in Oxford (Crisp 1972) and has subsequently been reported in many parts of the British Isles (Abbott et al. 1992; 2003). A number of factors are believed to have favoured its establishment, including greater seed production under some conditions (Oxford and Andrews 1977; Abbott 1985) and a difference in germination behavior (Abbott et al. 1998). However, pure stands of the variant are never found and an understanding of the factors that maintain populations that are polymorphic for both variants of S. vulgaris, and their heterozygote, remains elusive (Abbott et al. 1998). Marshall and Abbott (1982, 1984a) showed that the radiate morph exhibits a higher maternal outcrossing rate than the discoid morph, due to pistillate ray florets showing greater outcrossing than hermaphroditic disc florets (Marshall and Abbott 1984b), and also because radiate plants are more attractive to pollinators (Abbott and Irwin 1988).

Conclusions

Senecio squalidus is an excellent model for studying many aspects of the ecology, genetics, and evolution of an invasive diploid plant species. The species is known to have originated in the very recent past through diploid hybrid speciation, and then to have spread rapidly through a large part of the British Isles across an ecological gradient involving changes in temperature and rainfall. From the work conducted on this species, much has been learnt about its origin, its mating system (particularly in regard to the nature and population genetics of its SI system), its overall population genetic structure, and several important evolutionary events resulting from its hybridization with the native tetraploid species, Senecio vulgaris. However, much remains to be discovered in regard to why the species has been such a highly successful invasive plant. For example, it will be of interest to determine what genetic changes may have occurred during and following the origin of S. squalidus that adapted it to conditions in the British Isles and which in turn may have enabled it to spread widely thereafter. Currently, we are using a number of approaches, including genomic screens and microarray analysis, to isolate genes that may have been important in adapting S. squalidus to conditions in the British Isles. The isolation of such genes in this and other invasive plant species will lead to a greatly improved understanding of the genetic basis of invasiveness in plants.