No genetic differentiation in the rose-infesting fruit flies Rhagoletis alternata and Carpomya schineri (Diptera: Tephritidae) across central Europe

After the last glacial retreat in Europe, multiple recolonizations led to intraspecific differentiation in many of the recolonizing taxa. Here we investigate the genetic diversification across central Europe in two recolonizing taxa, the tephritid fruit flies Rhagoletis alternata (Fallén, 1814) and Carpomya schineri (Loew, 1856), which attack rose hips. Analysis of amplified and sequenced fragments of the mitochondrial genes encoding cytochrome oxidase I (800 bp), cytochrome oxidase II (470 bp) and cytochrome b (450 bp), indicate that all the individuals of R. alternata (n = 21) collected from across Europe share the same haplotype. Two individuals of C. schineri form Berlin, which is further north of the range than previously reported in the literature, differ from the other individuals (n = 13) in one nucleotide position on the cytochrome oxidase II gene fragment. This level of genetic variation in sequences with a summed length of 1720 bp is unexpectedly lower than in other insect taxa (n = 63). This might have been caused by a selective sweep by a cytoplasmic symbiont such as Wolbachia, or a recent range expansion associated with a host shift or a single recolonization event.


INTRODUCTION
Many European species survived the last ice age in one or several refugia.Owing to the isolation of these regions during this period, the taxa in the refugia differentiated (Hewitt, 1996).With the warming of the climate, species ranges expanded again and populations from different refugia came into secondary contact (Hewitt, 1996), which resulted in many taxa in Europe having complex genetic suture zones (Taberlet et al., 1998).
Here, we analyse the genetic diversification in two tephritid fruit flies that attack rose hips: Rhagoletis alternata (Fallén) and Carpomya schineri (Loew).The former, R. alternata is distributed throughout the Palaearctic region (Kandybina, 1977;Smith & Bush, 2000).In contrast, C. schineri has a more restricted distribution in southern Europe, southern Asia, and northern Africa (White & Elson-Harris, 1992).In southern Europe, these two species occur sympatrically.The larvae of both fruit flies develop in rose hips, where they feed not on the seeds but on the hypanthium.Adult females of R. alternata lay their eggs under the surface of hips, which they mark with a pheromone that deters other females.Around October, the larvae leave the hips and pupate in the soil (Bauer, 1986(Bauer, , 1998)).The larvae of C. schineri leave the hips around August and the adult flies emerge around June the following year (Hendel, 1927).
Both flies are host specific at the generic level and their distribution depends on that of their host plants, mainly the dog roses, Rosa section Caninae.Members of this rose section are the most common rose species in Europe and western Asia.They are evolutionarily young, originated by hybridization (Wissemann, 2002;Ritz et al., 2005) and expanded their range into central and northern Europe after the last ice age (Dingler, 1907).The distribution of the roses provided a platform for the flies to recolonize Europe from certain refugia.If the flies came from Mediterranean refugia there should be genetic suture zones in central Europe.We therefore screened the genetic variation in samples of flies collected from populations across this area.

Collection of larvae
Infested rose hips were collected in September 2004 from the canton Valais in Switzerland and in 2006 across Europe (Fig. 1; Supplementary material: Table S1), sent to our laboratory in perforated plastic bags and then stored at 15°C.The larvae were allowed to leave the hips and pupate within the bags.Pupae were stored in 95%-alcohol at 5°C for DNA analysis.Pupae parasitized by Utetes magnus (Fischer, 1958) are brown (T.Hoffmeister, pers.commun.),which allowed us to select nonparasitized individuals for DNA extraction.As a control, some adult R. alternata were allowed to emerge from each sample.
The amplification reactions (20 µl) contained 20-100 ng of template DNA, 10 mM Tris-HCl pH 8.3, 50 mM KCl, 1.5 mM MgCl2, 80 µM dNTP, 10 µM of each of the two primers and 1 unit of Taq DNA polymerase (Sigma, Taufkirchen, Germany).The products were purified using a Qiagen MinElute PCR purification kit.The purified PCR products were sequenced directly in both directions by Sequencing Laboratories Göttingen GmbH, Germany.

Sequence analysis
We compared the variability in mtDNA with that recorded for other insect populations.We searched for population genetic studies of insect taxa in the ISI Web of Knowledge.The aim of our ISI web search was not to construct a complete data set, but to retrieve sufficient information for statistical analysis.We used the keywords "Insect", "Diptera", "Lepidoptera", "Coleoptera" or "Hymenoptera", in combination with "genetic structure" and/or "genetic differentiation".We selected these keywords after some testing.These preliminary searches showed that the selected keyword combinations retrieved sufficient but a manageable number of papers.Furthermore, to obtain the most recent publications only the first 100 were selected.Finally, we excluded studies on parthenogenetic and social insects as they are known to show low genetic variability.From the published reports on these insect taxa, we extracted the number of haplotypes found and the number and length of the sequences (Supplementary material: Table S2).All variables were log10-transformed before analysis using a general linear model in STATISTICA (Version 6.1).Both fly species were included in the analyses.The number of expected haplotypes was calculated for both fly species using the coefficients of the model (Table 1).

RESULTS AND DISCUSSION
We sequenced the chosen mtDNA fragments of 21 individuals of R. alternata from 12 localities across central and southern Europe (Fig. 1).None of the 1720 nucleotide positions from COI, COII, and Cyt b (accession numbers FJ571363, FJ571366, and FJ571369, respectively) were variable, i.e. there is only one haplotype throughout the area sampled.The nucleotide sequences of the COII gene fragment were identical to the COII sequence of R. alternata available in GenBank (acc.no.U53260).We also found no differences in the sequences of the amplified COI (acc.nos.FJ571364 and FJ571365) and Cyt b gene fragments (acc.nos.FJ571370 and FJ571371) from 13 individuals of C. schineri from 4 localities, and the nucleotide sequences of the COI gene fragment were identical to the published sequence for this species (acc.no.U53267).In contrast to the situation in R. alternata, the COII gene fragment sequences from 2 individuals of C. schineri (acc.no.FJ571367) differed at one nucleotide position from those in the other 11 individuals (acc.no.FJ571368), which shared one haplotype.The 2 individuals with the second haplotype were collected in Berlin, far north of the range of C. schineri reported in the literature (Kandybina, 1977;White & Elson-Harris, 1992;Smith & Bush, 2000).
The mtDNA genes chosen for our analyses, those encoding cytochrome oxidase I and II subunits and cytochrome b, are protein-encoding genes with considerable variability even between closely related species and populations of the same species (Rokas et al., 2002;Simon et al., 1994).Furthermore, the chosen fragments of these genes include the most variable positions of the genes and are therefore often used for population genetic studies of animals, especially insects (e.g.Lunt et al., 1998;Rokas et al., 2003;Arias et al., 2005;Pramual et al., 2005;Sezonlin et al., 2006).These criteria indicate that the selected mtDNA gene fragments should be suitable for elucidating the biogeographic history of these fly species.
We compared our findings with mtDNA sequence data for other insect populations (Fig. 2).For 21 sequences with a summed length of 1720 bp (Fig. 2) 9-10 expected haplotypes were calculated and 6-7 haplotypes for 13 sequences of the same summed length.The standardized residual of the model for R. alternata was -2.93 (P = 0.0016) and for C. schineri -1.62 (P = 0.053).
Compared to the results for other insect populations, the genetic variation in R. alternata and C. schineri was significantly lower than expected.This could have at least two explanations.One possibility is that symbionts, such as Wolbachia, shape mtDNA evolution (Hurst & Jiggins, 2005), which would constrain the suitability of mtDNA sequences for molecular biogeographic studies of insects.During the initial phase of symbiont invasion, selective sweeps may reduce mtDNA diversity, thereby producing a genetic signal similar to that produced by a population bottleneck with subsequent expansion (Hurst & Jiggins, 2005).Wolbachia is known to infect members of the genus Rhagoletis (Riegler & Stauffer, 2002) but not members of the genus Carpomya (Kittayapong et al., 2000).The second possibility is that the flies recently and rapidly expanded their range from one source population.Such an expansion could be induced by colonization or a host-shift; in both cases, lower levels of genetic variation would be expected due to founder effects (Harrison, 1991).Both of the fly species studied are specialists and therefore dependent on the distribution of their host, members of the genus Rosa section Caninae.These dog roses originated by hybridization during the last ice age (Wissemann, 2002;Ritz et al., 2005) prior to the recolonization of Europe (Dingler, 1907).Founder individuals of the two fly species may have shifted to this new host, which would provide an explanation for the low genetic variability.S2).Twenty-one sequences from Rhagoletis alternata (black square) and 13 from Carpomya schineri (white square) are included in this analysis.Grey squares, Diptera; grey circles, Coleoptera; grey triangles, Hymenoptera; grey diamonds, Lepidoptera; grey inverted triangles, other insect orders.
Today, the distribution and density of roses is influenced by humans.Rosa rugosa, for example, was introduced from East Asia about 100 years ago (Hegi, 1975) and is now cultivated in parks, gardens and along roads all over Central Europe.Also its rose hips are attacked by R. alternata, and this increase in available hosts may have triggered the spread of these flies across Europe.R. alternata disperses well, and even the Alps do not seem to be an effective geographical barrier (Vaupel et al., 2007).This increase in distribution can also be partly explained by the behaviour of the females.After oviposition, the females mark the rose hips with a pheromone (Bauer, 1986(Bauer, , 1998)).Often, a high proportion of the hips, up to 100%, are infested and marked.Females leave such locations and search for rose shrubs with a lower proportion of infested hips.These observations and our results indicate that a recent range expansion of the flies from an unknown source area may account for their low genetic variability.Nuclear markers of R. alternata, e.g.allozyme genes, also show little variability, which suggests a high level of gene flow between European populations (Leclaire & Brandl, 1994;Vaupel et al., 2007).Note also that our finding C. schineri near Berlin extends the known range of this species northwards.The range expansion of this fly species could also be influenced by humans, who may have transported the larvae in plant material.
In conclusion, we found a surprisingly low level of genetic variability in tephritid fruit fly populations across central Europe.The reasons remain uncertain, but a recent and single colonization from an unknown source or the host shift to Rosa section Caninae are plausible explanations.The reporting of such findings is likely to facilitate a pluralistic understanding of the biogeography of plants and animals living in Europe.

Fig. 2 .
Fig. 2. Number of haplotypes in relation to (A) the number of sequences analysed (all variables log10 transformed) and (B) sequence length for 63 insect taxa (Supplementary material obtained from the literature cited in TableS2).Twenty-one sequences from Rhagoletis alternata (black square) and 13 from Carpomya schineri (white square) are included in this analysis.Grey squares, Diptera; grey circles, Coleoptera; grey triangles, Hymenoptera; grey diamonds, Lepidoptera; grey inverted triangles, other insect orders.

TABLE 1 .
Estimated coefficients and analyses of variance of the general linear model.

TABLE S1 .
Localities at which Rhagoletis alternata and Carpomya schineri were sampled in Europe.

TABLE S2 .
Studies on the number of haplotypes in insects used in the comparison with those found in this study on Rhagoletis alternata and Carpomya schineri.