Diversity and Phylogenetic Analyses of Bacterial Symbionts in Three Whitefly Species from Southeast Europe

Bemisia tabaci (Gennadius), Trialeurodes vaporariorum (Westwood), and Siphoninus phillyreae (Haliday) are whitefly species that harm agricultural crops in many regions of the world. These insects live in close association with bacterial symbionts that affect host fitness and adaptation to the environment. In the current study, we surveyed the infection of whitefly populations in Southeast Europe by various bacterial symbionts and performed phylogenetic analyses on the different symbionts detected. Arsenophonus and Hamiltonella were the most prevalent symbionts in all three whitefly species. Rickettsia was found to infect mainly B. tabaci, while Wolbachia mainly infected both B. tabaci and S. phillyreae. Furthermore, Cardinium was rarely found in the investigated whitefly populations, while Fritschea was never found in any of the whitefly species tested. Phylogenetic analyses revealed a diversity of several symbionts (e.g., Hamiltonella, Arsenophonus, Rickettsia), which appeared in several clades. Reproductively isolated B. tabaci and T. vaporariorum shared the same (or highly similar) Hamiltonella and Arsenophonus, while these symbionts were distinctive in S. phillyreae. Interestingly, Arsenophonus from S. phillyreae did not cluster with any of the reported sequences, which could indicate the presence of Arsenophonus, not previously associated with whiteflies. In this study, symbionts (Wolbachia, Rickettsia, and Cardinium) known to infect a wide range of insects each clustered in the same clades independently of the whitefly species. These results indicate horizontal transmission of bacterial symbionts between reproductively isolated whitefly species, a mechanism that can establish new infections that did not previously exist in whiteflies.


Introduction
Insects share a long-term and intimate association with bacterial symbionts, which has helped their hosts to become well adapted to numerous challenges present in their environment [1][2][3][4]. Primary bacterial symbionts are essential for insects to live on nutritionally poor diets, while secondary

Whitefly Populations
Populations of the three whitefly species were collected in the summers from 2009 to 2013 in regions with intense agricultural activity in Croatia, Montenegro, Macedonia, and Serbia ( Figure 1, Table S1). Whitefly specimens were collected from open fields or greenhouses when high population densities were available. Sampling locations, whitefly species, and host plants are shown in Table S1 and Table 3. Whitefly populations from Macedonia and Serbia were newly examined; two laboratory populations of B. tabaci from Israel were used for comparison, whereas other whitefly populations from Croatia and Montenegro were previously screened in Skaljac et al. (2010Skaljac et al. ( , 2013 [39,41] and used in this study as well. Adults of the three whitefly species were collected using a Pasteur pipette attached to a hand-held aspirator and then moved into absolute ethanol and kept at −20 • C until processing.

Whitefly Populations
Populations of the three whitefly species were collected in the summers from 2009 to 2013 in regions with intense agricultural activity in Croatia, Montenegro, Macedonia, and Serbia ( Figure 1, Table S1). Whitefly specimens were collected from open fields or greenhouses when high population densities were available. Sampling locations, whitefly species, and host plants are shown in Tables S1 and 3. Whitefly populations from Macedonia and Serbia were newly examined; two laboratory populations of B. tabaci from Israel were used for comparison, whereas other whitefly populations from Croatia and Montenegro were previously screened in Skaljac et al. (2010Skaljac et al. ( , 2013 [39,41] and used in this study as well. Adults of the three whitefly species were collected using a Pasteur pipette attached to a hand-held aspirator and then moved into absolute ethanol and kept at −20 °C until processing.  Table S1) with overview of overall infection frequencies of secondary bacterial symbionts (according to Table 2) in the three whitefly species in Southeast Europe. Each chart is marked with a code (B, T, and S) in order to distinguish between the whitefly species. B, Bemisia tabaci; T, Trialeurodes vaporariorum; S, Siphoninus phillyreae.

Screening and Sequencing of Secondary Bacterial Symbionts and Molecular Identification of B. tabaci Species
Adults (n = 10-20 per population) from the whitefly populations (Table S1) were tested for the presence of secondary symbionts and B. tabaci species (MED or MEAM1) identification. Genomic  Table S1) with overview of overall infection frequencies of secondary bacterial symbionts (according to Table 2) in the three whitefly species in Southeast Europe. Each chart is marked with a code (B, T, and S) in order to distinguish between the whitefly species. B, Bemisia tabaci; T, Trialeurodes vaporariorum; S, Siphoninus phillyreae.

Screening and Sequencing of Secondary Bacterial Symbionts and Molecular Identification of B. tabaci Species
Adults (n = 10-20 per population) from the whitefly populations (Table S1) were tested for the presence of secondary symbionts and B. tabaci species (MED or MEAM1) identification. Genomic DNA of each individual was extracted in lysis buffer as previously described in Chiel et al. (2007) [37] and used for further screening. Bemisia tabaci MED or MEAM1 groups were identified using microsatellite markers by polymerase chain reaction (PCR) amplification using Bem 23 primers and fragment sizes. The product size obtained from MEAM1 was 200 bp, and from the MED, it was 400 bp. In addition, we amplified mtCOI gene fragment for phylogenetic analyses of B. tabaci species (Table 1). Secondary bacterial symbionts were tested by PCR using genus-specific primers amplifying 16S or 23S rDNA (Table 1). We did not detect Fritschea in our previous studies, and due to the lack of a positive control, Fritschea was not considered in this study. PCR was carried out in 25 µL volumes containing 4 µL DNA lysate, 10 pmol of each primer, 10 mM dNTP mix, 10× PCR Rxn Buffer, 2 mM MgCl 2 , and 5 units/µL of Platinum Taq DNA polymerase (Invitrogen). PCR products were visualized on a 1.5% agarose gel containing ethidium bromide. The PCR products were eluted, and DNA was ligated into a pGEM-T Easy Vector (Promega, Madison, WI, USA) following the manufacturer's instructions. After transfection into RapidTrans TAM1 competent E. coli (Active Motif, Carlsbad, CA, USA), cells were grown on LB agar (Invitrogen) at 37 • C overnight. Clones were then screened for PCR inserts using the primers listed in Table 1. Colonies were then grown overnight in 5 mL LB medium containing ampicillin (100 µg/mL) in a shaking incubator at 200 rpm and 37 • C. Plasmid DNA from independent colonies was purified using a QIAprep Spin Miniprep kit (Qiagen, Hilden, Germany) and sent for sequencing (3730xl DNA analyzer, Macrogen Europe, Amsterdam, The Netherlands). The sequences were obtained from 10-20 individual specimens per primer set and compared with those in the databases using the BLAST algorithm at the NCBI [63].

Phylogenetic Analyses
Mitochondrial COI and secondary symbiont sequences for all B. tabaci, T. vaporarium, and S. phillyreae obtained during this study, together with reference sequences from GenBank, were aligned using the program MAFFT 7 [64] followed by visual inspection and manual adjustment. The appropriate model of evolution was estimated using the Bayesian Information Criterion (BIC) with Jmodeltest 2.1.6 on CIPRES Science Gateway [65], and selected models were the same for ML inferences. Analyses were initiated from random starting trees. The models selected were GTR+G for mtCOI, HKY for Hamiltonella, GTR+G for Arsenophonus and Wolbachia, and HKY+I for Cardinium and Rickettsia. Using these models, phylogenetic reconstruction was carried out using Bayesian inference (BI). For BI analysis, two independent runs of Markov Chain Monte Carlo (MCMC) were done for 10 million generations using eight chains and a sampling frequency of 1000 generations by MrBayes ver. 3.2 [66]. The first 25% of the samples were discarded by setting the burn-in fraction to 0.25. This burn-in setting was shared across the SUMT command to discard 25% of sampled trees. These phylogenetic trees were viewed by FigTree V. 1.4.2, available at http://tree.bio.ed.ac.uk/software/figtree/. Pairwise genetic distances for whitefly mtCOI sequences were analyzed using the p-distance model of evolution in MEGA 6.0 [67]. The whitefly sequences characterized in this study were assigned to different species on the basis of the rule of >3.5% pairwise sequence divergence [19].

Nucleotide Sequence Accession Numbers
The nucleotide sequences of the genes identified in this study have been deposited in GenBank under the accession numbers KY623456 to KY623475 for Arsenophonus, from KY621483 to KY621519 for Hamiltonella, from KY994554 to KY994569 for Rickettsia, from KY994545 to KY994553 for Wolbachia, from KY994541 to KY994544 for Cardinium, and from KY628431 to KY628448 for mtCOI from B. tabaci.

Whitefly Infection with Secondary Bacterial Symbionts
Ten to 20 individuals from each population of the three whitefly species were tested for the presence of the secondary bacterial symbionts. Infection frequencies with multiple secondary symbionts present in 24 whitefly populations used in this and our previous studies are given in Tables 2 and 3, and Figure S1. All the tested individuals were positive for the primary symbiont of whiteflies, P. aleyrodidarum, and this was considered to control the quality of extracted DNA.
We could not find a correlation between infection with bacterial symbionts and host plants that whiteflies infested, as was suggested in our previous studies (Table 2) Table S1). All populations used in this study were identified as MED species and they belonged to groups Q1 and Q2 (Tables 2 and 3). These populations had high prevalence of mixed infections with Hamiltonella, Arsenophonus, Rickettsia, Wolbachia, and Cardinium (Table 2). Hamiltonella was highly prevalent or fixed in B. tabaci MED (Q1) populations from Macedonia (Doiran, Gevgelija) and Montenegro (Darza). Arsenophonus was prevalent in populations from Montenegro (Bar-MED Q2, Lastva Grbaljska-MED Q1), whereas Rickettsia was fixed in population collected in Bar (MED Q2). Individuals of both Q1 and Q2 groups of B. tabaci MED were mainly infected with one (5-75%) or two symbionts (15-35%), whereas infections with three and four symbionts were only common in Q2 B. tabaci MED populations from Zadar (Croatia) and Bar (Montenegro) ( Table 3). The two populations of B. tabaci MED from Croatia (Turanj-Q2, Split-Q1) had 35% of individuals that did not contain any of the tested secondary symbionts. Rickettsia infected around 70-75% of individuals of both Israeli B. tabaci MED (Q2) and MEAM1 populations. Additionally, Arsenophonus was fixed in Israeli MED (Q2), whereas Hamiltonella was fixed in MEAM1 population. Beside Rickettsia and Arsenophonus, around 20% of individuals of the Israeli MED (Q2) population were infected with Wolbachia. All individuals of both Israeli MED (Q2) and MEAM1 populations contained at least one secondary symbiont. In Israeli MED (Q2), 25% of individuals were infected with one, 55% with two symbionts, and 20% of individuals with three symbionts. Israeli MEAM1 population contained individuals infected with one (30%) and two (70%) secondary symbionts.

Trialeurodes vaporariorum Infection with Secondary Bacterial Symbionts
T. vaporariorum populations were collected in Croatia, Montenegro, Macedonia, and Serbia ( Figure 1 and Table S1). All populations used in this study were predominantly infected with Hamiltonella and Arsenophonus ( Table 2). The only exception was T. vaporariorum population collected in Montenegro (Podgorica), which was additionally infected with Rickettsia, Wolbachia, and Cardinium. Arsenophonus was fixed or close to fixation in the populations collected in Montenegro, Macedonia, and Serbia, while the infection rate for Hamiltonella varied between 5% and 90% independently of the location ( Table 2). Individuals of T. vaporariorum were mainly infected with one (10-100%) or two symbionts (5-90%), whereas there were no infections detected with three or four symbionts. The populations of T. vaporariorum from Croatia (Split) and Serbia (Opovo, Zorka Subotica) had 5-25% of individuals that did not contain any of the tested secondary symbionts.

Siphoninus phillyreae Infection with Secondary Bacterial Symbionts
S. phillyreae populations were collected in Croatia and Montenegro ( Figure 1 and Table S1). All populations used in this study had high prevalence of mixed infections with Hamiltonella, Arsenophonus, Wolbachia, and Cardinium (Table 2). Interestingly, Rickettsia was not found in any of the tested populations. Both Arsenophonus and Hamiltonella were highly prevalent in all the tested S. phillyreae populations (25-100% and 65-100%, respectively). All the tested S. phillyreae populations were infected with Wolbachia (15-55%), while Cardinium was detected only in populations from Croatia (Supetar, Pucisca, Ljuta). Individuals of S. phillyreae were commonly infected with one (5-60%), two (20-50%), and three symbionts (5-55%) ( Table 3). Infections with four symbionts were rare and appeared only in individuals from Croatian S. phillyreae populations from Opuzen and Ljuta in 5% and 10%, respectively. Overall, all S. phillyreae individuals contained at least one secondary symbiont.

Phylogenetic Relationships of B. tabaci Sequences
A total of 18 B. tabaci sequences were characterized from Croatia, Montenegro, and Macedonia. In this study, B. tabaci sequences from Israeli populations served as a reference. A phylogenetic tree was constructed from the mtCOI sequences of B. tabaci and divergence analyses were compared with 40 putative species previously reported [10,15,[19][20][21][22][23][24][25][26][27] (Figure 2). Our results showed that B. tabaci populations from Croatia, Montenegro, Macedonia, and Israel were identified as MED (Q1 and Q2) species. Populations of B. tabaci MED from Macedonia (Doiran, Gevgelija), Montenegro (Darza, Lastva Grbaljska), and Croatia (Split) belonged to the Q1 group, whereas populations from Croatia (Turanj) and Montenegro (Bar) belonged to the Q2 group ( Figure 2). The exception was a B. tabaci MED population from Croatia (Zadar), which contained individuals from both Q1 and Q2 groups ( Figure 2). In addition, the B. tabaci MEAM1 species was identified from the Israeli population only.

Phylogenetic Analysis of Secondary Symbionts in Whiteflies
All secondary symbionts sequenced in this study were compared with bacterial reference sequences from GenBank (https://www.ncbi.nlm.nih.gov/genbank/). Sequence analysis revealed that, in addition to species diversity, several bacterial subgroups/strains may exist within the different bacterial genera tested. For Hamiltonella, the sequences from B. tabaci, T. vaporarium, and S. phillyreae formed two groups (H1 and H2) (Figure 3). Clade H1 contains Hamiltonella from B. tabaci and T. vaporarium from Croatia, Montenegro, Serbia, and Macedonia. Group H2 contains Hamiltonella from S. phillyreae and T. vaporarium from Croatia and Montenegro, respectively. 23rDNA sequences of Arsenophonus cluster into two major groups A1 and A2 (Figure 4). The A1 group contains strains from B. tabaci and T. vaporarium. Group A2 contains a distinct Arsenophonus strain from S. phillyreae from Croatia and T. vaporarium from Montenegro, which differs from the A1 group. Rickettsia was grouped into two major clusters, R1 and R2 ( Figure 5). The top clade (R1) consists of Rickettsia from both B. tabaci and T. vaporarium from Croatia and Montenegro. Clade R2 consists of Rickettsia from only T. vaporarium from Montenegro. Interestingly, none of the S. phillyreae individuals tested positive for Rickettsia. In the case of Wolbachia, to date, 16 supergroups successively named A-Q infecting a wide range of arthropods and filarial nematodes have been reported [68]. In our analysis, the B

Phylogenetic Analysis of Secondary Symbionts in Whiteflies
All secondary symbionts sequenced in this study were compared with bacterial reference sequences from GenBank (https://www.ncbi.nlm.nih.gov/genbank/). Sequence analysis revealed that, in addition to species diversity, several bacterial subgroups/strains may exist within the different bacterial genera tested. For Hamiltonella, the sequences from B. tabaci, T. vaporarium, and S. phillyreae formed two groups (H1 and H2) (Figure 3). Clade H1 contains Hamiltonella from B. tabaci and T. vaporarium from Croatia, Montenegro, Serbia, and Macedonia. Group H2 contains Hamiltonella from S. phillyreae and T. vaporarium from Croatia and Montenegro, respectively. 23rDNA sequences of Arsenophonus cluster into two major groups A1 and A2 (Figure 4). The A1 group contains strains from B. tabaci and T. vaporarium. Group A2 contains a distinct Arsenophonus strain from S. phillyreae from Croatia and T. vaporarium from Montenegro, which differs from the A1 group. Rickettsia was grouped into two major clusters, R1 and R2 ( Figure 5). The top clade (R1) consists of Rickettsia from both B. tabaci and T. vaporarium from Croatia and Montenegro. Clade R2 consists of Rickettsia from only T. vaporarium from Montenegro. Interestingly, none of the S. phillyreae individuals tested positive for Rickettsia. In the case of Wolbachia, to date, 16 supergroups successively named A-Q infecting a wide range of arthropods and filarial nematodes have been reported [68]. In our analysis, the B supergroup was observed in all three species tested from Croatia, Montenegro, and Macedonia ( Figure 6). 16S rDNA sequences of Cardinium from B. tabaci and S. phillyreae are grouped under a single clade (Figure 7).

Discussion
Numerous studies have shown that whitefly-associated secondary bacterial symbionts (Arsenophonus, Hamiltonella, Rickettsia, Wolbachia, and Cardinium) play major roles in whitefly biology and have developed through evolutionary history numerous mechanisms to invade and persist in the insect host [1,45,46,52,69,70]. Besides common secondary symbionts, several studies reported the presence of other bacteria in B. tabaci (e.g., Bacillus, Enterobacter, Paracoccus, and Acinetobacter) [71,72]. However, so far none of those bacteria have established a relationship with whiteflies as secondary symbionts investigated in this study. All secondary symbionts are vertically transmitted with high fidelity, but most of them are frequently transferred via horizontal transmission routes, allowing them to spread successfully within and between the species [73].
Our recent studies have extensively investigated infection and localization of secondary symbionts in the three whitefly species (B. tabaci, T. vaporariorum, S. phillyreae) in Southeast Europe [39,41]. We found unique co-infection patterns, revealing the presence of similar secondary symbionts in reproductively isolated whiteflies. This occurrence is likely to be due to highly dynamic processes of horizontal transfer of symbionts between insect hosts, possibly via sharing host plants, mating, or parasitoids [38,73,74]. These events are necessary for the symbiont to spread to novel hosts [75]. Such acquisition of bacterial symbionts can directly impact host biology [46]. For example, the

Discussion
Numerous studies have shown that whitefly-associated secondary bacterial symbionts (Arsenophonus, Hamiltonella, Rickettsia, Wolbachia, and Cardinium) play major roles in whitefly biology and have developed through evolutionary history numerous mechanisms to invade and persist in the insect host [1,45,46,52,69,70]. Besides common secondary symbionts, several studies reported the presence of other bacteria in B. tabaci (e.g., Bacillus, Enterobacter, Paracoccus, and Acinetobacter) [71,72]. However, so far none of those bacteria have established a relationship with whiteflies as secondary symbionts investigated in this study. All secondary symbionts are vertically transmitted with high fidelity, but most of them are frequently transferred via horizontal transmission routes, allowing them to spread successfully within and between the species [73].
Our recent studies have extensively investigated infection and localization of secondary symbionts in the three whitefly species (B. tabaci, T. vaporariorum, S. phillyreae) in Southeast Europe [39,41]. We found unique co-infection patterns, revealing the presence of similar secondary symbionts in reproductively isolated whiteflies. This occurrence is likely to be due to highly dynamic processes of horizontal transfer of symbionts between insect hosts, possibly via sharing host plants, mating, or parasitoids [38,73,74]. These events are necessary for the symbiont to spread to novel hosts [75]. Such acquisition of bacterial symbionts can directly impact host biology [46]. For example, the

Discussion
Numerous studies have shown that whitefly-associated secondary bacterial symbionts (Arsenophonus, Hamiltonella, Rickettsia, Wolbachia, and Cardinium) play major roles in whitefly biology and have developed through evolutionary history numerous mechanisms to invade and persist in the insect host [1,45,46,52,69,70]. Besides common secondary symbionts, several studies reported the presence of other bacteria in B. tabaci (e.g., Bacillus, Enterobacter, Paracoccus, and Acinetobacter) [71,72]. However, so far none of those bacteria have established a relationship with whiteflies as secondary symbionts investigated in this study. All secondary symbionts are vertically transmitted with high fidelity, but most of them are frequently transferred via horizontal transmission routes, allowing them to spread successfully within and between the species [73].
Our recent studies have extensively investigated infection and localization of secondary symbionts in the three whitefly species (B. tabaci, T. vaporariorum, S. phillyreae) in Southeast Europe [39,41]. We found unique co-infection patterns, revealing the presence of similar secondary symbionts in reproductively isolated whiteflies. This occurrence is likely to be due to highly dynamic processes of horizontal transfer of symbionts between insect hosts, possibly via sharing host plants, mating, or parasitoids [38,73,74]. These events are necessary for the symbiont to spread to novel hosts [75]. Such acquisition of bacterial symbionts can directly impact host biology [46]. For example, the spread of Rickettsia in B. tabaci from the United States resulted in large fitness benefits and female bias in the tested whiteflies [46].
Intensive plant trade occurs between Mediterranean countries and this provides opportunities for introducing new whitefly species and symbionts in regions where they have not existed before [76]. Several studies reported the invasion of B. tabaci MED (Q2), previously associated only with the Middle East countries of the Mediterranean basin, into Spain, France, and Italy, where it coexisted with the native B. tabaci MED (Q1) [76,77]. Our study revealed the presence of both Q1 and Q2 groups of B. tabaci MED in the investigated region (Table 2, Figure 2). Two populations identified as Q2 were collected in Croatia (Turanj) and Montenegro (Bar). Interestingly, a B. tabaci MED population (Croatia-Zadar) collected on plants imported from Italy consisted of both Q1 and Q2 groups and they carried Arsenophonus, which was not previously detected in existing B. tabaci MED (Q1) populations (Table 2, Figure 2) [39].
Variations in the diversity and infection frequencies of bacterial symbionts between the three whitefly species in this study were substantial (Tables 2 and 3, Figure S1).
Our results revealed that Arsenophonus and Hamiltonella from both B. tabaci and T. vaporariorum were phylogenetically grouped in the same clades (with several exceptions; e.g., Hamiltonella from T. vaporariorum Podgorica), compared with those symbionts from S. phillyreae, which were placed in more distant clades (Figures 3 and 4). B. tabaci and T. vaporariorum frequently share host plants, therefore the possibility of horizontal transmission of bacterial symbionts between these two whitefly species is high [39]. S. phillyreae was only found on pomegranate plants, which was not a host for the other two whitefly species [40,42]. Placement of Hamiltonella in different clades was also reflected on its localization pattern in the whitefly hosts. Hamiltonella found in B. tabaci and T. vaporariorum was localized exclusively in bacteriosomes, while in S. phillyreae it was found to be scattered throughout the entire bodies of nymphs and adults [39,41].
Hamiltonella was present at high frequencies in the investigated species of B. tabaci MED (Q1) and MEAM1 (Table 2), suggesting it had a major impact on the host biology. The infection frequencies of Hamiltonella were in accordance with the global screening reports of bacterial symbionts in B. tabaci MED (Q1) and MEAM1 [33,38,40,78,79]. However, these studies showed absence of Hamiltonella in B. tabaci species such as MED (Q2, ASL), Sub-Saharan Africa 1-5, Asia II, and China 1 [33,78,80]. This was contrary to our results, where Hamiltonella infected 25% and 65% of individuals from MED (Q2) populations from Croatia (Turanj) and Montenegro (Bar), respectively ( Table 2). Infection of Hamiltonella varied in T. vaporariorum populations (from absence of infection to high prevalence) in this study, whereas infection with this symbiont was highly prevalent in S. phillyreae. Surprisingly, a large study by Kapantaidaki et al. (2014) [81] did not report infection of Hamiltonella in T. vaporariorum populations from different global locations.
In this study, Rickettsia was present in nearly all populations of B. tabaci MED. However, infection rates were much lower independently of Q1 and Q2 groups compared to Israeli MED and MEAM1 populations ( Table 2). This suggests recent introduction of Rickettsia in B. tabaci populations in the investigated region. Other studies have shown that Rickettsia was present at high infection rates in the MEAM1, MED (Q2), and China 1 groups of B. tabaci, while lower infection rates were observed in Asia II and the MED groups of B. tabaci from Burkina Faso (Q3) and China [33,38,40,78,79]. This symbiont was rarely detected in T. vaporariorum, and no infection was found in S. phillyreae populations (Table 2) [41,81].
Wolbachia is one of the most commonly present symbionts in insects [82]. In this study, nearly all B. tabaci and S. phillyreae populations were infected with Wolbachia (10-60% and 15-50%, respectively), while it was rarely found in T. vaporariorum (Table 2). Several studies reported highly diverse infection rates of Wolbachia in B. tabaci MED (Q1, Q2, ASL), while it was not observed in other genetic groups [33,40]. Bing et al. (2013) [78] found that native genetic groups of B. tabaci (Asia II, China 1) were highly infected with Wolbachia, whereas this symbiont was not detected in invasive MED and MEAM1 groups. Cardinium, another known reproductive manipulator, was rarely detected among populations of the three whitefly species in this study (Table 2) [39,41,43]. Other studies have revealed large differences in rates of Cardinium infection in populations from the B. tabaci species complex [33,38,40,78,79]. Phylogenetically, Wolbachia and Cardinium present in the three whitefly species from this study appeared in the same clades, together with other reported sequences of the same symbionts (Figures 6 and 7).
Mixed infections and heterogeneity among populations in our study could be a result of competition for space and resources among the symbionts or the tolerance of the host to maintain diverse bacterial community [39,83]. This can contribute to multiple events of horizontal transmissions of secondary symbionts within the same whitefly species and reproductively isolated B. tabaci and T. vaporariorum (Tables 2 and 3). In our study, horizontal transmission seems likely for Hamiltonella, Arsenophonus, Wolbachia, and Cardinium between B. tabaci and T. vaporariorum. Rickettsia was mostly associated with B. tabaci, and there were rare individuals from T. vaporariorum that were positive for this symbiont, which could reflect isolated events of horizontal transmission [41]. Chiel et al. (2009) [75] demonstrated horizontal transmission of Rickettsia in B. tabaci using parasitoid wasps, while transmission of Hamiltonella failed in this system. In addition, Rickettsia was successfully shown in B. tabaci to be transmitted via host plants [73]. There are evidences for horizontal transmission of bacterial symbionts such as Arsenophonus, Cardinium, and Wolbachia either among whiteflies, between whiteflies and arthropods present in the environment, or, as previously mentioned, via host plants [62,[84][85][86]. Phylogenies of secondary symbionts are largely incongruent with those of their hosts, which is not the case with the phylogeny of primary symbionts [86]. This suggests an ancient infection of a host with primary symbiont followed by vertical transmission and consequent cospeciation with the insect host. In contrast, incongruent phylogenies suggest multiple infections and horizontal transfer between different hosts [86][87][88].
In the investigated bacterial community of the three whiteflies, we have found more than 20 different symbiotic combinations in a range from no infected individuals to those that were infected with four symbionts (Table 3). In populations of B. tabaci and T. vaporariorum single infections dominated (43% and 56%, respectively), while in S. phillyreae single (25%), double (38%), and triple (31%) infections were present in similar range. Double infections were present in 25% and 38% of individuals of B. tabaci and T. vaporariorum, respectively. Interestingly, T. vaporariorum in this study had no infections with three or four symbionts, while those infections existed in B. tabaci in 14% and 7.5%, respectively. In B. tabaci, T. vaporariorum, and S. phillyreae, individuals with no infections were present in 11%, 5%, and 3%, respectively. Our results were slightly different from study of Gueguen et al. (2010) [33], where infections of global populations of B. tabaci with two or more symbionts were common and double infections dominated (59%). Furthermore, several studies on sub-Saharan populations of B. tabaci report increasing number of secondary symbiont-free whiteflies [88,89]. It is likely to conclude that heterogeneity of symbiotic community of the whiteflies depends on location, population, and time [88].

Conclusions
The present study found substantial diversity of bacterial symbionts in the three whitefly species, especially between those that shared habitat (B. tabaci and T. vaporariorum) and in S. phillyreae, which was exclusively found on pomegranate plants. Dynamic processes of horizontal transmission could bring phylogenetically distant symbionts together in previously uninfected whiteflies (e.g., via parasitoids). This finding can contribute to the discovery of new functions that symbionts such as Hamiltonella or Arsenophonus may fulfill in their whitefly hosts.  Table S1. The whitefly populations used in this study.