Introduction

Amongst the most important, and perhaps most controversial, questions in biology is that of speciation (e.g. Barton 2001). Biological species evolve from lineages between which a barrier to genetic exchange has been established (Coyne 2007). The process of allopatric speciation is widely accepted, in which this barrier is between different geographical areas. In this case, geographical structure leads to a degree of genetic population structure, which, if large enough, may ultimately lead to genetically different lineages that no longer successfully interbreed (e.g. Turelli et al. 2001). There are, however, other ways that may lead to genetic structure within populations, and thus, potentially, speciation. A much more controversial mechanism is sympatric speciation (Via 2001; Berlocher and Feder 2002; Bolnick and Fitzpatrick 2007; Coyne 2007). This type of speciation does not require geographic separation between lineages. It may theoretically result from the establishment of an evolutionary novelty, such as a mutation influencing habitat use or mate choice (e.g. Jiggins 2008).

One example where convincing evidence has been obtained for genetic differentiation between sympatric lineages is the host plant use of phytophagous insects (Berlocher and Feder 2002; Drès and Mallet 2002; Dambroski et al. 2005; Egan et al. 2008; Xie et al. 2008). Here, in several cases, part of an insect population has acquired the ability to exploit a novel host plant species. Once such an evolutionary novelty has arisen, barriers to genetic exchange between insects using the ‘old’ host plant(s) and those able to use the new one as food may be generated through, for example, phenological constraints (Feder and Filchak 1999; Filchak et al. 2000; Dambroski and Feder 2007), pleiotropic effects (Mackenzie 1996), host preference (e.g. Forbes and Feder 2006) or assortative mating (Malausa et al. 2005). Thus, host plant races may evolve in the phytophagous insect that may, given enough time, lead to sympatric speciation.

Whereas, in the view of sympatric speciation geographical disjunction is not required, this does not imply the total absence of a geographical component in the precise shaping of the evolutionary pathway. It has been shown in many organisms that populations are genetically structured, e.g. through geographical discontinuities (see e.g. Mopper and Strauss 1998). Although genetic differentiation, for example in host plant races of phytophagous insects, may locally arise sympatrically, further, more global, spread of adaptations involved in host race formation may still be influenced by processes of migration, selection and drift. The eventual geographic distribution of adaptations involved in host plant use may thus be influenced by local differences in selection, as well as insect mobility. Thus, to understand the variation in host plant use by a phytophagous insect, it is crucial to detect the different factors that may limit the genetic exchange between subpopulations. This is the subject of the present paper, in which we report factors associated with population structure in the flea beetle Phyllotreta nemorum L. (Coleoptera: Chrysomelidae) and propose an explanation for the present-day distribution of its adaptation to an atypical host plant, Barbarea vulgaris ssp. arcuata R. Br. (Brassicales: Brassicaceae).

P. nemorum is an oligophagous flea beetle, feeding on a relatively narrow range of cruciferous host plants, most notably Sinapis arvensis L. (Brassicales: Brassicaceae) (Nielsen 1997 a,b). In East Denmark individuals have been found that used the atypical (though locally abundant) host plant B. vulgaris. This plant species has two types, called P- and G-type (with pubescent and glabrous leaves, respectively) of which the latter is chemically defended against at least some phytophagous insects (Nielsen 1997a). Thus, the majority of P. nemorum cannot feed on this plant. However, some populations of P. nemorum were found feeding on the G-type of B. vulgaris in East Denmark (Nielsen 1997a,b; de Jong and Nielsen 1999). These turned out to be genetically resistant to the plant defences. Whereas the resistant beetles can use the G-type of B. vulgaris as well as other host plant species that were studied, beetles that did not contain the resistance allele(s) could use the same range of host plants, except the G-type of B. vulgaris. This could represent a unilateral barrier to gene flow, namely from beetles using other plants than B. vulgaris (G-type) to those using this toxic plant.

Earlier work has shown that, whereas flea beetles sampled on the G-type of B. vulgaris were all resistant, the proportion of resistant beetles found on other host plants was always much lower, with a maximum of approximately 70% (de Jong and Nielsen 1999; Nielsen and de Jong 2005). Although an allozyme study indicated that there is some limitation to gene flow in P. nemorum due to the geographical distance between the samples, it was concluded that this barrier was unlikely to explain the geographical distribution of resistance in this beetle (de Jong et al. 2001). However, the existence of host races, partly isolated by unidirectional migration, and potentially also by other factors limiting gene flow of beetles from B. vulgaris to other host plants (e.g. phenological differences across host plants, preference, assortative mating, or selection against resistant migrants towards other host plants) could explain the geographical distribution of resistance. This could also represent an early step in sympatric speciation. In the present study we investigated whether there was any evidence for genetic diversification between resistance-phenotypes of P. nemorum, when corrections were made for effects of geographical distance between the samples.

Materials and Methods

Results

Distribution of different resistance-phenotypes

In all samples collected on B. v8.4x10-3ulgaris, the proportion of resistant P. nemorum ranged from 98–100%. Samples collected on other plants, even those less than 1 km. from B. vulgaris, usually contained many fewer resistant beetles (15.0–71.4%; see Nielsen and de Jong 2005). One exception involved two samples that were collected in Ejby: one on B. vulgaris, and the other on S. arvensis growing next to it. Here, the percentage of resistant beetles on B. vulgaris was 99.0%, and on S. arvensis 96.0%.

Figure 1.

Locations in Denmark where samples were collected. The numbers refer to the sampling locations. In some cases, a number of samples were collected close together at one location; these are then indicated by one dot on the map. 1–6: Kværkeby; 7: Maglebrænde; 8: Vigersted; 9: Taastrup; 10–12: Ejby; 13: Viby; 14: Lynæs; 15: Svebølle; 16–17: Try Enge; 18: Suserup; 19: No. Vedby. Locations 13, 16–17, and 19 were not used for the analysis.

General description of samples

The average size of the 22 samples was 31.1. The average proportion of polymorphic loci (99% criterion) across the samples was 0.82. The average number of alleles per locus was 2.98 (per polymorphic locus 3.44). The mean proportion of heterozygotes across all samples was 0.23 (expected: 0.23).

Table 1.

The 22 samples collected at the 14 different localities in East Denmark, including the host plant on which collections were made, resistance phenotype of each sample (+: resistant to defences of B. vulgaris), and the number of beetles that were used for the analysis.

Discussion

These results reveal two independent sources of genetic differentiation in the flea beetle P. nemorum: geographical distance (which was earlier found by de Jong et al. 2001) as well as resistance phenotype (which had not been studied before). Despite the limited allelic variability at the allozyme loci used in this study, and therefore the low resolution of population differentiation, pairwise theta values between samples containing different resistance phenotypes were statistically significantly larger than those comparing the same phenotype, after correction for the effect of geographical distance. This result implies that the genetic exchange between different phenotypes is limited, independently of isolation by distance.

Figure 2.

Pairwise genetic distance (theta) against pairwise geographical distance. Closed symbols: comparison between different resistance phenotypes; open symbols: comparison between identical resistance phenotypes.

An earlier study (de Jong and Nielsen 2001) concluded that no genetic differentiation was detectable between beetles collected on different host plants with overlapping local distributions. Those results involved comparisons between samples consisting of mixtures of resistance phenotypes (note that on other host plants than B. vulgaris, both resistant and non-resistant flea beetles occur). The present analysis, therefore, is the first rigorous test of the existence of genetic differentiation between beetles that can, and cannot, use B. vulgaris (G-type) as food. This may indicate the beginning of host race formation.

Various factors may be responsible for the genetic differentiation between different resistance-phenotypes of the flea beetles. First, non-resistant flea beetles cannot use the G-type of B. vulgaris as food. They are therefore more likely to mate with flea beetles living on other host plants than B. vulgaris that would have a mixture of non-resistant and resistant beetles. In this case, gene flow from nonresistant to resistant flea beetles would be partially impeded. Other factors that may further enhance the isolation between the resistance phenotypes include host preference, different host phenologies leading to asynchrony in host plant use by the different resistance phenotypes, or selection limiting gene flow from resistant to non-resistant flea beetles.

No evidence has been published so far supporting host preference in the flea beetles. If resistant flea beetles preferred to feed on the G-type of B. vulgaris, it would limit the gene flow from resistant to non-resistant flea beetles, leading to a larger degree of isolation between the phenotypes than that only caused by the inability of non-resistant individuals to feed on B. vulgaris, G-type.

There is some support for phenological effects that may affect genetic exchange between the phenotypes. B. vulgaris overwinters as a rosette, and is available immediately when the flea beetles emerge from overwintering in early spring. The other major host plant, S. arvensis, is an annual, and this plant becomes available later in the season after germination of the seeds (Nielsen 1997a). Thus, in early spring B. vulgaris (G-type) may be colonized by resistant flea beetles that would then mate assortatively, leading to a degree of temporal isolation between the phenotypes. Such phenological factors influencing host race formation have also been documented for the applemaggot fly (e.g. Feder and Filchak 1999).

Evidence has also been obtained for the third possible factor enhancing isolation between flea beetle resistance phenotypes, namely selection against resistant flea beetles. De Jong and Nielsen (2000) found a negative pleiotropic effect of resistance: homozygosity for resistance reduced the chance of survival in a crossing experiment with strains that had been reared in the laboratory for five generations. In the field, therefore, resistant migrants competing with non-resistant conspecifics on other host plants than B. vulgaris may be selected against (de Jong and Nielsen 2002). This would mean that genetic exchange between the resistance phenotypes is reduced by selection in two different directions: 1) non-resistant beetles are selected against when they attempt to feed on B. vulgaris (G-type), and 2) resistant beetles are selected against when they compete with non-resistant ones on other plants than B. vulgaris. To what extent the restriction in gene-flow due to selection at the resistance-locus is also reflected in neutral allozyme variation, depends on the timing of effects of selection relative to reproduction; gene-flow at neutral loci may be very high despite strong selection at an adaptive locus if neutral alleles can be established in the population before the effects of selection are manifested.

The possible causes of genetic differentiation between resistance-phenotypes in P. nemorum described above are not mutually exclusive. It is, however, likely, given the low estimates of genetic differentiation that were found in the present study and previous work (de Jong and Nielsen 2001), that a substantial amount of gene flow exists across resistance phenotypes and geographical distance. On the continuum of host plant races to (sympatric) species, therefore, the flea beetles are likely to be positioned at the very beginning of host plant associated differentiation.

The restriction of genetic exchange between the resistance phenotypes of P. nemorum as found in the present study helps understand the geographical distribution of resistant versus non-resistant beetles. Field populations of P. nemorum living on B. vulgaris G-type have all been found to be resistant to the plant defences. This is not surprising, since the individuals were always sampled in the larval stage, and these larvae are leaf miners with a restricted mobility (Alford 1999). They can only feed on the defended type of B. vulgaris if they are resistant to its defences. The interesting question with respect to the spatial distribution of resistance phenotypes of P. nemorum is: why are resistant larvae relatively rare on other plants than B. vulgaris (G-type)? In the absence of limitations to genetic exchange between beetles that are, and are not, able to use B. vulgaris (G-type), one would expect a rapid spread of the resistance trait in the flea beetles. The fact that a large proportion of flea beetles on other host plants than B. vulgaris are non-resistant to the defences of this plant can be understood by the barrier to gene flow that apparently exists across the resistance phenotypes. This barrier can be formed by any of the mechanisms described above, or combinations between them. The exact proportion of resistant to non-resistant flea beetles may vary locally, depending on the topographical and ecological circumstances influencing the amount of gene flow, e.g. proximity of patches of B. vulgaris G-type that are used by resistant flea beetles, relative abundance of different host plants, local differences in phenology of different host plants, and so forth. The global relationship between P. nemorum and its host plants is likely to be best described by a mosaic of local interactions between the beetles, their host plants and their enemies, leading to locally adapted subpopulations that are genetically isolated to some extent (cf. Thompson 2005). In a number of these localities, host race formation may be actually occurring. One of the first steps that should be followed now is to identify the variable mechanisms leading to the observed genetic differentiation between the resistance phenotypes of these flea beetles.