Symbionts do not affect the mating incompatibility between the Brazilian-1 and Peruvian morphotypes of the Anastrepha fraterculus cryptic species complex

The South American fruit fly, Anastrepha fraterculus, is clearly undergoing a speciation process. Among others, two of their morphotypes, the Brazilian-1 and Peruvian, have accumulated differences in pre- and post-zygotic mechanisms resulting in a degree of reproductive isolation. Both harbor a different strain of Wolbachia, which is a widespread endosymbiotic bacterium among many invertebrates producing a range of reproductive effects. In this paper, we studied the role of this bacterium as one of the factors involved in such isolation process. Infected and cured laboratory colonies were used to test pre- and post-zygotic effects, with special emphasis in uni- and bi-directional cytoplasmic incompatibility (CI). We showed that Wolbachia is the only known reproductive symbiont present in these morphotypes. Wolbachia reduced the ability for embryonic development in crosses involving cured females and infected males within each morphotype (uni-directional CI). This inhibition showed to be more effective in the Peruvian morphotype. Bi-directional CI was not evidenced, suggesting the presence of compatible Wolbachia strains. We conclude that Wolbachia is not directly involved in the speciation process of these morphotypes. Other mechanisms rather than CI should be explored in order to explain the reduced mating compatibility between the Brazilian-1 and Peruvian morphotypes.

. Molecular characterization of Wolbachia by HVRs and wsp analyses. HVR alleles were assigned considering translated nucleotide sequences and wsp alleles were established by comparing the obtained sequences against a Wolbachia nucleotide database.
The latency to mate was statistically different among treatments [F (3,111) = 7.20, p = 0.0002, Table 2]. The crossing involving AfP− females and AfP− males mated faster than the other combinations. Also, mating duration was affected by the infection status [F (3,117) = 19.29, p < 0.0001, Table 2]. Multiple contrasts showed that matings involving AfP− females ended significantly later (p < 0.001) than those involving AfP+ females, irrespectively of the male infection status.

Cross
Latency Duration

Discussion
In the present study we evaluated the reproductive compatibility within and between two morphotypes belonging to the A. fraterculus cryptic species complex infected by single and different Wolbachia strains, considering the potential role of the endosymbiont in the previously observed phenotypic and behavioural differences 30,37 . Within each colony, cured females that mated with infected males had a significantly reduced egg hatch when compared to the other type of crosses. These results on this key variable suggest uni-directional CI. The absence of bi-directional CI was evidenced in crosses involving both infected morphotypes, as no egg hatch reduction was observed. Despite large similarities in the studied parameters observed among heterotypic crosses, our results suggest that the presence of Wolbachia might affect them in a different way. This could be a consequence of the differences between Wolbachia strains 29,30,37 , differences in the genetic background of the hosts 30,39 or a distinctive interaction between each morphotype and its reproductive symbiont. Differences in the density of the symbiont within cells, which was not assessed, may have also accounted for the observed differences as previously shown in Drosophila species 40,41 and the wasp Nasonia vitripennis (Hymenoptera: Pteromalidae) 42 . Yet, independently of the mechanism, the reduced number of matings between morphotypes was confirmed.
Based on the sequence of the wsp gene, different Wolbachia strains and 100% prevalence were detected in the A. fraterculus colonies used in our experiments. In the case of AfP, our results showed the presence of wPer strain, as was previously reported by Cáceres and his colleagues 30 . Sequence analyses showed identical MLST allelic profiles between wPer and the wAfraCast1_A strain infecting A. fraterculus (Brazilian-1 morphotype) from Argentina 37 . In addition, we confirmed the allelic patterns of Wolbachia infecting AfC (wAfraCast2_A) as was previously characterized by Conte et al. 37 .
Differences between morphotypes were observed from pre-zygotic tests. For AfP, the lack of an effect on the mating percentage suggests that the infection does not affect flies' mating propensity for either sex. However, Figure 3. Percentage of matings between Wolbachia-cured (−) or infected (+) males and females belonging to the Brazilian-1 morphotype of Anastrepha fraterculus (AfC). Cross notation: Female × male. an infected status may represent possible fitness costs, as the latency to mate decreased and the mating duration increased when flies were cured from Wolbachia. These characteristics have been proposed to be advantageous 43,44 . In the case of AfC, the absence of Wolbachia in both sexes correlates with a lower tendency to copulate, again suggesting a negative effect associated with the curing protocol. Cured AfC males needed more time to mate and all matings involving a cured fly lasted longer, which may indicate that the curing process or the absence of some bacterial taxa (even though not checked, probably removed by the antibiotic treatment) affected the mating parameters in each morphotype in a slightly different way, as suggested by Ikeya and colleagues 45 for Drosophila. Alternatively, and as pointed out for Drosophila simulans 46 , a short mating duration may be explained by a lower amount of transferred sperm when infected with Wolbachia.
Pre-mating sexual isolation between Peruvian and Brazilian-1 morphotypes has been previously demonstrated 30,35,36 . Because these morphotypes are infected by distinct Wolbachia strains, it was hypothesized that Wolbachia could, at least indirectly, be involved in this mechanism and therefore its removal would, partially or totally, revert the observed reproductive isolation. Here, the percentage of mated AfP females and AfC males was not affected by the infection status of the flies. However, when comparing the reciprocal crosses, a significant reduction in the percentage of mating was found when Wolbachia was removed from both sexes, which may have resulted from a detrimental effect of the antibiotic treatment on the mating competitiveness 47 . Thus, our results do not support the idea that Wolbachia infection could be one of the evolutionary forces behind the pre-zygotic isolation between the Peruvian and Brazilian-1 morphotypes. Regarding the time variables, even when statistical differences were found in some cases, they seem to be related to the morphotype per se more than to the infection status. These results are in concordance with previous results on mating latency and duration 30,35,36 .
As previously commented, the effects of antibiotic treatment on host physiology should be taken into account as a possible explanation for the effects on some biological parameters in crosses involving Wolbachia-cured individuals 48,49 . Despite the fact that the cured flies were reared for at least two generations without antibiotics before performing the experiments (allowing the reposition of the main gut microbiota), it is possible that some important symbionts have not been restored. If they played a role in mating propensity as suggested by Juárez et al. 47 , the curing process (alone or interacting with other factors) might have inadvertently and negatively affected the fitness of cured flies. Another factor that may have caused the observed differences among the experiments is the different genetic background of the two morphotypes. There are examples in which the same strain of Wolbachia has different effects in closely related hosts 39 . Transinfection experiments (i.e., wPer in cured AfC flies and wAfraCast2_A in cured AfP flies) may help to test this hypothesis.
Most known examples of reproductive phenotypes caused by Wolbachia are related to phenomena taking place after fertilization (i.e., post-zygotic effects). In our work, the infection status did not affect the egg-laying ability in any of the two morphotypes tested independently, and only a detrimental effect in the heterotypic cross between uninfected AfP females and AfC males was found. This reduction cannot be explained either by the origin of the flies or by their infection status and may be related to a specific biochemical stress associated to that particular combination of factors including the removal of undetected microorganisms potentially important to the normal development of the offspring. After egg-laying, uni-directional CI was confirmed for both morphotypes given the observed reduction in egg hatch, with an even higher intensity in the Peruvian morphotype. For AfC, there seems to be no effect after the larvae had hatched, which is evidenced in the similarity in the percentages of pupation and F1 emergence among crosses. Yet, AfP showed a significant decrease of pupation percentage only when both sexes were cured (and slightly decreased when only the female was cured) and this could be explained by some missing factors that co-evolved between host and bacteria (i.e., biochemical imbalance).
Cáceres et al. 30 observed a reduction in egg hatch and a sex ratio distortion as a degree of post-zygotic isolation between flies from the Brazilian-1 and Peruvian morphotypes of A. fraterculus. These authors provided evidence suggesting major genetic differences based on the chromosomal asynapsis recorded in their hybrids. They also suggested the presence of a nuclear-cytoplasmic interaction, probably involved in the differences observed in reproductive parameters. Based on molecular, behavioural, and morphometric traits, the two morphotypes have accumulated important differences 30,34 , to the point in which they are considered undergoing speciation. In our study, no bi-directional CI was evidenced even though the two morphotypes are infected with two distinct Wolbachia strains. However, it is not known if these two strains, wPer and wAfraCast2_A, carry the same or closely related bacteriophage WO genes, cifA and cifB, that have recently been shown to contribute to the induction of -and in the case of cifA also in the rescue of-cytoplasmic incompatibility [50][51][52][53] . The fact that the cross of infected AfP female and infected AfC male has shown an even higher egg hatch percentage when compared with the respective control (i.e., the cured cross) might indicate that some other factors are potentially involved in these crosses that were not evaluated in our experiments. Further studies on mating compatibility between morphotypes considering their microbiota composition, interactions, and other physiological changes induced in the host species, will complement the information generated here to understand the host-symbionts dynamics and its association with phenotypic traits elicited.
In sum, we have evaluated the effects of the most widespread reproductive symbiont of insects infecting two morphotypes of the A. fraterculus cryptic species complex that show an intermediate degree of sexual isolation compared to other combinations of entities. Uni-directional CI was confirmed for both, with a different degree of penetrance. Bi-directional CI was not evidenced. Differences in the observed general pattern suggest biochemical imbalances produced directly or indirectly by the antibiotic treatment or by the absence of other relevant bacteria.  www.nature.com/scientificreports www.nature.com/scientificreports/ et al. 55 . A purified colony containing only wAfraCast2_A was obtained at IGEAF and sent to IPCL, which was treated with antibiotics (see below) and used as the AfC cured colony in the experiments. The AfC infected colony was previously established at IPCL from the same origin but contained both Wolbachia variants. It was considered that AfC cured and infected strains possess the same genetic background.

Molecular detection and characterization of Wolbachia.
Total DNA was isolated from 20 randomly selected adult flies (whole body) from each colony using the CTAB method 55 . Wolbachia detection was based on PCR and direct visualization of the present/absence of amplicons. For this purpose, a fragment (438 bp) of the gene encoding the 16S ribosomal RNA (16S rRNA) was amplified using the Wolbachia-specific primers wSpecF and wSpecR 56 . The results of PCR amplifications were visualized by electrophoresis in 1.5% w/v agarose gels and stained with ethidium bromide 55 . Images were captured with an UVP reveller (Fotodyne Inc. Hartland, WI, USA).
The identification of the Wolbachia strain infecting AfC and AfP colonies was performed by PCR amplification and sequence analysis of a portion (590 to 632 bp) of the Wolbachia surface protein (wsp) gene, obtained by PCR using 81F/691R primers 57 . Amplicons were purified using a Wizard SV Gel and PCR Clean-Up System (Promega) and sequenced. Forward and reverse sequences were obtained using an Abi 3130XL Genetic Analyzer (Applied Biosystem, Thermo Fisher Scientific Inc.). Sequences were manually edited and aligned using Bioedit 58 and Staden Package 59 . The characterization of the sequences includes allele assignment based on comparisons of the nucleotide sequence obtained against the Wolbachia database and the analysis of hypervariable regions (HVRs), which were in turn determined by comparisons among available translated nucleotide sequences 60 . HVR allele definition is based on the analysis of the amino acid motifs of the wsp gene sequence (61-573 bp) in respect to wMel. Also, the sequences of five genes (gatB, coxA, hcpA, fbpA and ftsZ) were analysed by means of an MLST-based approach as a further genetic characterization 61 . Wolbachia MLST public database was used for analyses and assignment of wsp alleles and MLST sequence type (https://pubmlst.org/wolbachia/).

Antibiotic treatment.
In order to obtain Wolbachia-cured individuals from the AfP colony, an antibiotic treatment was applied to the larval diet using 0.1% tetracycline (Sigma) at IPCL. Similarly, cured AfC individuals were obtained at IGEAF by application of 0.01% rifampicin (Laboratorios Richet, Buenos Aires, Argentina) in the larval diet. After 6 generations under antibiotic treatment, each colony was reared without antibiotic for at least two generations before the initiation of any experimental work. The cured status was confirmed through the 16S rRNA gene PCR approach as described above. In addition, the presence of other commonly present symbionts was tested using primers and conditions described previously (Spiroplasma sp. 62 , Cardinium sp. 63 , Rickettsia sp. 64 , and Arsenophonus sp. 65 ).
The following colonies were considered for three experiments: Peruvian infected A. fraterculus (AfP+), Peruvian cured A. fraterculus (AfP−), Brazilian-1 infected A. fraterculus (AfC+), and Brazilian-1 cured A. fraterculus (AfC−). Each colony was reared in different cages following standard procedures 66 and kept under environmentally controlled conditions (25 °C, 60% HR, 12 L:12D), with adult diet (sugar and hydrolysed yeast at a 3:1 ratio) and water. Incompatibility tests. AfP and AfC were first assessed independently in uni-directional cytoplasmic incompatibility (CI) tests, crossing individuals in four different ways (treatments) according to the sex and the infection status (Experiments 1 and 2 in 'Results' section). Also, infected or cured heterotypic crosses were performed (Experiment 3 in 'Results' section) in order to understand to what extent, the post-zygotic incompatibility reported between the two morphotypes 30 is related to the infection by different Wolbachia strains (i.e. bi-directional CI).
Pre-zygotic incompatibility tests. In order to determine if the presence of Wolbachia affects mating compatibility, no-choice experiments were conducted with one virgin male and one virgin female inside a mating arena. To this end, adults were sorted by sex after emergence and kept in separate containers with adult food and water until sexual maturation (10 days after emergence for males 67  , respectively, were set up. As bi-directional CI can be revealed only when different and incompatible Wolbachia strains are present, no crosses involving both cured and infected flies were tested. Notation throughout the paper is female × male. The couples were checked continuously for mating occurrence. For each couple, the time of mating start and end were recorded. The experiment ended after a 30-minutes span in which no matings were recorded. Mated couples were then provided food and water and kept for further post-zygotic incompatibility tests. Post-zygotic incompatibility tests. On the day after the mating test, mated females were offered an oviposition substrate. This device consisted of a Petri dish whose base (6 cm in diameter) was replaced by a black silicon-coated mesh. The dish was filled with water and placed on top of the acrylic box to allow constant oviposition through the mesh. Every two days, eggs from each female were transferred to a black filter paper and placed on larval diet within a small, plastic Petri dish (5 cm in diameter). Eggs were counted under a stereo-microscope (60 ×, Olympus, Japan). Samples with less than 10 eggs were discarded. Egg collection stopped after reaching at least 50 eggs per couple in more than one collection. Five to seven days after egg collection, hatched eggs were counted, the filter paper was removed, and the developing larvae were transferred into a closed, ventilated box with sawdust to allow pupation. In the case there were no hatched eggs, female spermathecae were dissected to check for sperm transfer following the protocol described by Segura et al. 69 . The females that showed empty spermathecae were not considered in the post-zygotic analyses. After all larvae had pupated, the Petri dish with the remaining larval diet was removed from the box to avoid fungal contamination. After emergence, the adults were counted, and their sex recorded. Emerged and non-emerged pupae were recorded as well. Parental couples were preserved in 96% ethanol to confirm the corresponding Wolbachia infection status (see procedures in section 'Molecular detection and characterization of Wolbachia').
After mating experiments, a total of 200 flies, representing all the combinations of morphotypes, sexes and infection status were screened for Wolbachia by the PCR-based protocol described above (16S rRNA gene amplification). Whenever one or both insects of the pair did not show the initially assigned status, the mating pair was eliminated from the data set and no longer considered in the statistical analyses of pre-and post-zygotic analyses.
Statistical analyses. Pre-zygotic incompatibility tests. The percentage of matings was compared among crosses within each experiment using the number of mated and unmated couples in a Chi-square test of homogeneity. The latency to mate was calculated as the difference between the mating start time and the time at which females were released in the boxes. To standardize this value among treatments within each experiment, the lowest latency value obtained from each treatment was subtracted from the respective replicates. Latency was compared among crosses within each experiment by means of a one-way ANOVA. When assumptions were not met (Experiment 2), latency was transformed using log (latency + 1). The duration of the mating was calculated as the time elapsed between the starting and ending times of mating. This variable was analysed by means of a one-way ANOVA, using log(e) as transformation (Experiment 1 and 3) in order to meet homoscedasticity assumption.
Post-zygotic incompatibility tests. The percentage of females that laid at least 50 eggs (in more than one oviposition event) was compared among treatments by means of Chi-square test of homogeneity using the number of egg-laying females and non-egg-laying females. The percentage of egg hatch was calculated out of the total number of eggs collected in several opportunities for every mated female. This variable was analysed by means of a one-way ANOVA, using square root as transformation in those cases were the assumptions were not met (Experiment 1 and 2). The percentage of pupation was calculated as the percent of the hatched eggs that reached the pupal stage. A one-way ANOVA was used for testing differences among treatments. The percentage of adult emergence was compared among crosses by means of one-way ANOVA using rank transformation. Finally, sex ratio was calculated as the proportion of emerged females from the total number of emerged adults and it was compared among crosses by a one-way ANOVA. For Experiment 1, sex ratio was transformed to 1/(sex ratio +1) whereas data from Experiment 2 were log-transformed. For statistical purposes, cases with less than 10 eggs, pupae, or adults were not considered in the data analysis.
Whenever statistical differences in the Chi-square test of homogeneity were found, crosses were compared in pairs in order to detect the treatment responsible of the statistical significance. Significant p-values after ANOVAs were followed by Tukey tests. All statistical analyses were performed with STATISTICA version 7 70 .

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary Information file). Wsp gene sequences generated in this study from wAfraCast2_A have been deposited in GenBank, National Center for Biotechnology Information (NCBI) databases under accession number KC589027.1. Allelic profile of MLST scheme of five genes (gatB, coxA, hcpA, fbpA and ftsZ) and HVRs allelic profile of wAfraCast2_A are available on Wolbachia MLST database. Nucleotide sequences of the groEL, gltA, dnaA, sucB, aspC, atpD and pdhB genes from Wolbachia infecting Brazilian-1 A. fraterculus were submitted to GenBank under accession numbers MG977022-28 respectively. Nucleotide sequences of MLST from Wolbachia infecting Peruvian A. fraterculus corresponding to gatB, coxA, hcpA, fbpA, ftsZ and hcpA genes were deposited in GenBank under AN MN145458-MN145462".