Dynamics of an Ongoing Wolbachia Spread in the European Cherry Fruit Fly, Rhagoletis cerasi (Diptera: Tephritidae)

Numerous terrestrial arthropods are infected with the alphaproteobacterium Wolbachia. This endosymbiont is usually transmitted vertically from infected females to their offspring and can alter the reproduction of hosts through various manipulations, like cytoplasmic incompatibility (CI), enhancing its spread in new host populations. Studies on the spatial and temporal dynamics of Wolbachia under natural conditions are scarce. Here, we analyzed Wolbachia infection frequencies in populations of the European cherry fruit fly, Rhagoletis cerasi (L.), in central Germany—an area of an ongoing spread of the CI-inducing strain wCer2. In total, 295 individuals from 19 populations were PCR-screened for the presence of wCer2 and their mitochondrial haplotype. Results were compared with historic data to understand the infection dynamics of the ongoing wCer2 invasion. An overall wCer2 infection frequency of about 30% was found, ranging from 0% to 100% per population. In contrast to an expected smooth transition from wCer2-infected to completely wCer2-uninfected populations, a relatively scattered infection pattern across geography was observed. Moreover, a strong Wolbachia-haplotype association was detected, with only a few rare misassociations. Our results show a complex dynamic of an ongoing Wolbachia spread in natural field populations of R. cerasi.


Introduction
Heritable bacterial endosymbionts are present in a broad range of arthropods with intriguing effects on the ecology and evolution of numerous species [1,2]. Endosymbionts are usually transmitted vertically from infected females to their offspring. By manipulating the reproduction through the induction of cytoplasmic incompatibility (CI), parthenogenesis, feminization, and male-killing, bacterial endosymbionts increase the fitness of infected females and enhance their frequency in host populations [3,4]. Certain endosymbionts are also able to provide fitness benefits protecting their host against RNA viruses or natural enemies [5,6]. Most endosymbionts are able to invade and adapt to new species by interspecific horizontal transmission [7][8][9][10].
One of the most common bacterial endosymbionts that can manipulate arthropod reproduction is Wolbachia [11]. This bacterium has been estimated to infect more than 50% of terrestrial arthropod species [1] and is present in numerous hexapods, crustaceans, chelicerates, and nematodes [11]. The most common reproductive manipulation is CI that is expressed when a Wolbachia-infected male

Sampling and DNA Extraction
Samples were collected from infested fruits of Prunus avium and Lonicera xylosteum in July 2016. In total, specimens from 19 German locations were sampled: One site from the northern part of Baden-Württemberg, i.e., 16/1, and 18 populations from central-south Hesse, i.e., 16/2-16/19 (Table 1, Figure 1, Table S1). Individuals were collected from a single plant per location, either as larvae from infested fruits in the field or as pupae after emerging from cherries in the laboratory. To avoid the analysis of siblings, fruits from different parts of one tree/shrub were taken. Samples were stored in absolute ethanol at −20 • C. DNA was extracted from 16 individuals per location, except for the populations 16/5 (n = 15), 16/6 (n = 10), and 16/19 (n = 14; Table S1), using the GenElute Mammalian Genomic DNA miniprep kit (Sigma-Aldrich, St. Louis, MO, USA) following the manufacturer's instructions.

Wolbachia Screening
Samples were PCR-screened for the presence of wCer1 and wCer2 using strain-specific primers targeting a part of the Wolbachia surface protein (wsp) [29,34,35]. PCR reactions were performed in a total volume of 10 µL, containing 1 mg/mL BSA, 2 mM Y-buffer (PeqLab/VWR, Erlangen, Germany), 800 µM dNTPs, 0.2 µM forward and reverse primer each, 0.5 U Taq polymerase (PeqLab/VWR), and 1 µL of template DNA. PCR conditions were 2 min at 94 • C, followed by 35 cycles of 94 • C for 30 s, 55 • C for 45 s, 72 • C for 1 min, followed by a final extension at 72 • C for 10 min.
PCR-amplified fragments were electrophoretically separated on a 2% agarose gel stained with GelRed Nucleic Acid Dye (Biotum, Hayward, CA, USA). As wCer1 is fixed in European R. cerasi [16,29,30,35], screenings for this strain were performed to control for sufficient DNA quality. In order to avoid false-negatives, all results were confirmed by two independent PCR runs.

Mitochondrial Genotyping of R. cerasi
To study the association of R. cerasi mitochondrial haplotypes with the Wolbachia infection status, genotyping of the flies was performed by applying restriction fragment length polymorphism (RFLP) as used in [16]. In brief, a part of the mitochondrial COI gene was PCR-amplified using the primers Pat and Dick [36]. Following PCR (conditions were the same as described above), 10 µL of the PCR product was incubated with 2 U of HaeIII (Thermo Fisher Scientific, Waltham, MA, USA) at 37 • C for 3 h. Fragments were separated on a 2% agarose gel-HT2 is cut into a 342 bp and a 204 bp fragment, while HT1 remains undigested [16].
A similar pattern was found in the northern locations where a wCer2-uninfected population (Lich, 16/12) was surrounded by two transitional populations 5 and 12 km apart, and just 14 km distant from the completely wCer2-infected population in Gießen/Prunus (16/14) (Figure 1c).

Mitochondrial Genotyping of R. cerasi and Haplotype-Wolbachia Associations
As only two mitochondrial R. cerasi haplotypes across Europe have been described previously, we used an RFLP approach to determine the haplotype affiliation of each individual. Genotyping of flies showed that the Wolbachia infection status was strongly associated with the two mitochondrial haplotypes of the host (Figure 2b). Almost all wCer2-uninfected individuals were associated with HT1 (205 out of 206). Just one individual from Alsfeld (16/16) was wCer2-uninfected but associated with HT2. In contrast, 88.8% of wCer2-infected R. cerasi were associated with HT2, whereas 10 wCer2-infected individuals from three populations were associated with HT1 ( Figure 2b). Haplotype-Wolbachia misassociations were found in populations in geographically close proximity, i.e., Gießen/Lonicera (16/13), Gießen/Prunus (16/14), and Lahnau (16/15). In Gießen/Lonicera (16/13), two of the 12 wCer2-infected samples were associated with HT1, whereas in Lahnau all three wCer2-infected flies were associated with HT1. The most significant deviation was found in Gießen/Prunus (16/14), where all 16 individuals were wCer2-infected, but only 11 (68.8%) were associated with HT2.

Wolbachia Dynamics and Haplotype Associations in Time and Space
Out of the 19 R. cerasi populations studied here, eight had already been screened for their Wolbachia infection status and mitochondrial haplotypes between 1999 and 2014 [16]. We compared our results with these historic data to infer the dynamics of Wolbachia infection and the associated mitochondrial haplotype of the fly. Rhagoletis cerasi collected from Dossenheim (16/1)-where all individuals screened were infected by wCer2-was already completely invaded by this strain in 1999 and 2008 (Table 2). In Stockstadt ( Comparing Wolbachia-haplotype associations between different years revealed that wCer2-infected flies associated with HT1 occurred in the same populations before (Figure 2a). Overall, the number of wCer2-infected R. cerasi associated with HT1 increased in all three populations. In contrast, the population in Alsfeld (16/16), where one wCer2-uninfected individual was associated with HT2, did not show this pattern in 2001. However, two out of five wCer2-infected flies were associated with HT1 in 2001, a pattern that was not confirmed in 2016 (Table S2).

Discussion
Here, we studied the frequency of wCer2 in R. cerasi in the central part of Germany, a region where this strain is currently spreading. Compared to previous work in this area [16], our fine-scale sampling and screening of 19 populations allowed us to characterize the spatial distribution of wCer2 in this transition zone. Furthermore, comparing our results with historic data from the last two decades [16] allowed us to assess the temporal dynamics of this Wolbachia strain in certain populations. In contrast to our expectation of a transitional zone with smooth gradients from completely infected to entirely uninfected flies, as described in other parts of the species' range [30], we found a surprisingly scattered pattern of wCer2 infection frequencies across different populations. Comparison of the Wolbachia infection status and the mitochondrial haplotype of individuals showed a strong relationship, however, occasional misassociations suggest events of intraspecific horizontal transfer. Taken together, our data provide new insights into the ongoing Wolbachia spread in European R. cerasi.

Wolbachia Infection Frequencies in Time and Space
At least five Wolbachia strains have been described from R. cerasi, and one individual fly can harbor various strains [29,35]. The spread of the strain wCer2 in Central Europe from south to north represents a rare event of an ongoing Wolbachia invasion in natural field populations [16,29,30]. By screening 295 individuals from 19 populations, we found an overall wCer2 infection frequency of about 30%, ranging from 0% to 100% per population. Instead of gradual transitional zones between completely wCer2-infected and entirely wCer2-uninfected R. cerasi populations [30], a relatively scattered infection pattern was found. Populations completely infected with wCer2 were in close proximity to locations with low wCer2 infection frequencies. Our results, together with previous data [16,29,30], provide a comprehensive picture of an ongoing Wolbachia spread in natural field populations of R. cerasi and suggest that the spatial pattern of this endosymbiont infection is influenced by various factors, such as long distance migration of flies and/or passive movement with infested cherries.
Although Wolbachia is one of the best-studied bacterial endosymbionts, only a limited number of studies have given insight into its spatial dynamics under natural field conditions [16,30,31,37,38]. In addition to empirical observations, theoretical modelling can help to provide a basic understanding on the mode of Wolbachia spread [14,39,40]. One of the best studied examples of an ongoing Wolbachia spread in the field is the mosquito Aedes aegypti, artificially infected with the strain wMel, that naturally occurs in Drosophila melanogaster [41,42]. wMel was found to spread relatively slowly in release areas with a rate of 100-200 m per year [41]. Major factors influencing this spread are deceased fitness of transinfected mosquitoes and a low dispersal rate of the host [41]. This is in contrast to the estimated spread of wCer2 in central Europe of 1-1.9 km per year [30], a spread that might benefit from the long adaptation of wCer2 to its host-with expected low fitness costs-and the higher dispersal capacity of the fly. Both systems showed that human-mediated dispersal of insects can influence their long-range migration and can result in unexpected spatial patterns of host organisms and their associated symbionts [16,43].
Generally, Wolbachia that causes fitness costs in its host needs to reach a sufficiently high equilibrium frequency to get established in a host population [41]. Thus, infections in low frequencies do not result in an establishment of the bacterium [40]. In contrast, Wolbachia can be established in a host population even from very low initial infection frequencies by providing positive fitness effects to their host, as it was reported from Drosophila simulans in Australia [37] and California [31]. Although the spatial pattern of wCer2 infection rates in Germany is different from other regions in the range of R. cerasi [16,29,30], our data suggest that a certain equilibrium infection frequency in a population is necessary for this strain to become established [16,30]. This was shown by the comparison of Wolbachia infection frequencies at the same sites among different years. For example, in the locations Ober-Ramstadt (16/3), Stockstadt (16/4), Idstein (16/6), Lich (16/12), and Lahnau (16/15), infection frequencies remained at low levels over a period of eight to 15 years. In contrast, in Gießen/Lonicera (16/13), wCer2 infection frequencies of R. cerasi collected from honeysuckle increased rapidly from 10% in 2001, to 68.8% in 2008 [16], and to 75% in 2016. We assume that wCer2 reached a sufficiently high infection frequency over time, and we expect that this strain will get fixed in the following years. The importance of an equilibrium infection frequency is further supported by the absence of intermediate infection frequencies over the sampling years, as it was found in almost all locations.
In Gießen/Prunus (16/14), however, R. cerasi was already completely wCer2 infected. This suggests an influence of the fly's host plant on the Wolbachia infection dynamics. Some lines of evidence propose host plant-related differentiation patterns in R. cerasi. For example, individuals from Prunus and Lonicera show slightly different eclosion times in spring, maybe a response to a differing fruiting phenology of the host [44]. Potential ecological differentiation between host ecotypes of R. cerasi infesting cherry or honeysuckle [18,20] might reduce gene flow between different populations, influencing the Wolbachia infection dynamics. A contrasting pattern was observed in the northernmost location Alsfeld (16/16) where the ratio of wCer2 infections decreased significantly from 50% in 2000 to 12.5% in 2016. Since none of the wCer2-uninfected individuals had the mitochondrial haplotype HT2, we exclude events of occasional loss of wCer2 as a reason for the decrease of this strain over time. The accidental anthropogenic introduction of wCer2-uninfected flies into this region, for example, via trade with R. cerasi-infested cherries, could explain this finding.

Wolbachia Infection and Mitochondrial Haplotype of the Host
The genetic structure of European R. cerasi on the mitochondrial level is generally low, reflected by the presence of only two haplotypes [16]. This low genetic diversity might be the result of a previous Wolbachia sweep by the strain wCer1. Schuler et al. [16] hypothesized that a few individuals of R. cerasi with HT1 got previously infected by wCer1 via horizontal transmission. This strain might have provided fitness benefits to the fly and swept through host populations, replacing all other haplotypes. Later, HT2 evolved, acquired wCer2 horizontally, and is now hitchhiking through European populations [16].
The mitochondrial haplotypes are tightly associated with Wolbachia, where wCer2-uninfected flies are associated with HT1 and wCer2-infected individuals with HT2. In our study, out of 295 samples, only 11 had a different Wolbachia-haplotype association. In 10 cases, these misassociations represented a wCer2-infection in individuals with HT1. This finding can be a result of intraspecific horizontal transfer [45]. For example, parasitoid wasps are known to be a potential source of horizontal transmission among species [16,[45][46][47][48]. Parasitoid species of R. cerasi were found to be infected with Wolbachia. Their role in horizontal transmission between and within fly species, however, has yet to be studied [45]. An additional mechanism of horizontal transfer might be via cannibalism among R. cerasi larvae [45]. Although multiple larvae within one cherry are uncommon [26], under certain environmental conditions [26] or when resources for oviposition are scarce, several larvae might develop in one fruit. In this case, a larva could acquire Wolbachia by feeding on a co-occurring larva but would remain associated with its mitochondrial haplotype.
Since all cases of a wCer2-infection in HT1 flies were found exclusively in transitional populations (where flies with both Wolbachia infection types are present), this phenomenon was interpreted as transient, where HT1 is assumed to be lost in populations completely invaded by wCer2 [16]. We found just three populations in a restricted area-in Gießen (16/13 and 16/14) and Lahnau (16/15)-with individuals belonging to HT1 harboring wCer2. The comparison with historic data from 2001, 2008, and 2014 [16], however, showed a general increase of individuals with these misassociations. In the location Gießen/Prunus (16/14) where R. cerasi is completely wCer2-infected, 31% of the individuals were associated with HT1. Subsequent studies are needed to understand if these flies can form a stable co-existence of wCer2 and HT1 or if they will be lost in future generations, as simulated by Schuler et al. [16].
Finally, one single wCer2-uninfected individual associated with HT2 was found. This suggests a potential case of unsuccessful maternal transmission of Wolbachia. The low occurrence of these misassociations reflects a strong CI-inducing effect of wCer2 with nearly perfect transmission from females to their offspring [16,19].

Conclusions
Our screening of the European cherry fruit fly with focus on the CI-inducing strain wCer2 provides new insights into this unique endosymbiont-host system. In contrast to an expected smooth transition and continuous gradient of infection frequencies from wCer2-infected to completely uninfected R. cerasi populations [30], we found a rather scattered geographic infection pattern of wCer2 infections. Our combined analysis of the Wolbachia infection status and the associated host genotype show a highly complex picture of just partially increasing wCer2 frequencies, possibly shaped by endosymbiont losses, intraspecific horizontal transmission events, and potential anthropogenic effects.
Supplementary Materials: The following are available online at http://www.mdpi.com/2075-4450/10/6/172/s1, Table S1: Detailed overview on the Wolbachia infection status and mitochondrial haplotypes of R. cerasi, Table S2: Overview on historic data on the Wolbachia infection status and the mitochondrial genotypes of R. cerasi.