Diversity and evolution of the endosymbionts of Bemisia tabaci in China

The whitefly Bemisia tabaci (Gennadius) (Hemiptera: Aleyrodidae) is a cryptic species complex, including members that are pests of global importance. This study presents a screening of B. tabaci species in China for infection by the primary endosymbiont, Portiera aleyrodidarum, and two secondary endosymbionts, Arsenophonus and Cardinium. The results showed that P. aleyrodidarum was detected in all B. tabaci individuals, while Arsenophonus was abundant in indigenous species of B. tabaci Asia II 1, Asia II 3, and China 1 but absent in the invasive species, Middle East-Asia Minor 1 (MEAM1); Cardinium presented in the Mediterranean (MED), Asia II 1 and Asia II 3 species but was rarely detected in the MEAM1 and China 1 species. Moreover, phylogenetic analyses revealed that the P. aleyrodidarum and mitochondrial cytochrome oxidase 1 (mtCO1) phylograms were similar and corresponding with the five distinct cryptic species clades to some extent, probably indicating an ancient infection followed by vertical transmission and subsequent co-evolutionary diversification. In contrast, the phylogenetic trees of Arsenophonus and Cardinium were incongruent with the mtCO1 phylogram, potentially indicating horizontal transmission in B. tabaci cryptic species complex. Taken together, our study showed the distinct infection status of endosymbionts in invasive and indigenous whiteflies; we also most likely indicated the co-evolution of primary endosymbiont and its host as well as the potential horizontal transfer of secondary endosymbionts.


INTRODUCTION
Bemisia tabaci is a cryptic species complex comprising a minimum of 40 morphologically similar species (De Barro et al., 2011;Dinsdale et al., 2010;Hu et al., 2018;Wang, Li & Liu, 2017). Among members of the complex, the Middle East-Asia Minor 1 (MEAM1) and Mediterranean (MED) groups (commonly known as the B and Q biotypes, respectively) have drawn much attention due to their global invasion and vectoring phylogenetic analyses of 16S and 23S ribosomal DNA (rDNA) (from endosymbionts) as well as mitochondrial cytochrome oxidase 1 (mtCO1) gene (from B. tabaci). Our study will not only uncover the current status of endosymbionts infection within B. tabaci in China, but also greatly provide a supplement for studies of B. tabaci endosymbionts worldwide.

Sample collection
A total of 1,510 B. tabaci individuals were collected from 71 geographical locations, including 19 provinces and four municipalities in China ( Fig. 1). At each location, B. tabaci samples were collected from different leaves of separate plants. The collection details, geographical sites and host plants were described in Table S1.

DNA extraction and gene amplification
Total DNA was extracted from individual whitefly as described in Luo et al. (2002). The primers of mtCO1 gene was used for whitefly species identification. 16S rDNA primers were used to detect P. aleyrodidarum and Cardinium, and for Arsenophonus, 23S rDNA primers were utilized. The primers, annealing temperature, and predicted PCR products size were shown in Table 1. The Polymerase chain reaction (PCR) reaction mixture contained one U Taq DNA polymerase, five ml (10Â) reaction buffer, three ml MgCl 2 (final concentration of 25 mmol/l), two ml dNTPs (10 mmol/l), two ml of forward and reverse primers (20 mmol/l each) and two ml of template DNA (Simon et al., 1994). PCR products were visualized by 1.5% agarose gels

Sequence alignment and phylogenetic analysis
Sequence fragments were assembled using ContigExpress and aligned using the Clustal X 1.83 program (Chenna et al., 2003). The GenBank database was searched for homologous sequences of mtCO1, 16S rDNA and 23S rDNA using the basic local alignment search tool. Phylogenetic trees were constructed using MrBayes 3.2.1 (Ronquist & Huelsenbeck, 2003). The best-fit substitution model for each of the aligned sequences was selected with the program Modeltest 3.7 (Posada & Crandall, 1998).
All the trees were constructed using the GTR+I+G model. The metropolis-coupled Markov chain Monte Carlo algorithm was conducted using four chains. Analyses were initiated with random starting trees, processed for 3 Â 10 6 generations, and sampled every 1,000 generations. For the burn-in period, we discarded 100,000 generations. Posterior clade probabilities obtained from the analysis were used to assess nodal support. Tree information was visualized and edited using FigTree ver. 1.3.1 (http://tree.bio.ed.ac.uk/software/figtree/).

Genetic diversity of B. tabaci and its endosymbionts
Aligned sequences from B. tabaci (813 bp, mtCO1 gene), P. aleyrodidarum (886 bp, 16s rDNA), Arsenophonus (551 bp, 23S rDNA), and Cardinium (460 bp, 16S rDNA) were used to analyze the genetic variation of whitefly and its endosymbionts. The results showed that 21 haplotypes were identified in whiteflies based on mtCO1 sequences, while 17, seven and eight symbiont haplotypes were defined based on analysis of P. aleyrodidarum, Arsenophonus and Cardinium sequences, respectively (Table 2). These haplotypes sequences were used to construct the phylogenetic trees.

Phylogenetic analysis of B. tabaci and endosymbionts
Phylogenetic trees were constructed for B. tabaci, P. aleyrodidarum, Arsenophonus, and Cardinium based on mtCO1 gene, 16s rDNA, 23S rDNA, and 16s rDNA sequences of haplotypes, respectively. We also pulled the related sequences from NCBI to explore the phylogenetic status of our haplotypes. The GenBank number of those related sequences could be found in phylogenetic trees. Based on the mtCO1 gene, we found five distinct genetic groups of B. tabaci, which was corresponding with the five cryptic species including the MEAM1, MED, Asia II 1, Asia II 3, and China 1 (Fig. 3). We can also find that all of our MED individuals belonged to Q1 subclade (Chu et al., 2011).
For phylogenetic trees of endosymbionts, several distinct bacterial strains existed within individual bacterium. It is interesting that the phylogenetic tree of P. aleyrodidarum were similar to the mtCO1 tree and exhibited four groups corresponding to MEAM1+MED, Asia II 1, Asia II 3, and China 1 clades to some extent (Fig. 4). While there are still several differences. For example, the sequences of P. aleyrodidarum of China 1 were not well clustered and P. aleyrodidarum from B. tabaci species MED (KP201113) was also present in that China 1 clade. In addition, the MEAM1+MED clade (Fig. 4) contained sequences of P. aleyrodidarum from both MEAM1 and MED; however, it was separated into two distinct clades in the B. tabaci tree (Fig. 3). In contrast, the phylogenetic trees for Arsenophonus and Cardinium were totally incongruent with the mtCO1 tree, exhibiting four (A1-A4) and three groups (C1-C3), respectively (Figs. 5 and 6).

DISCUSSION
In this study, we conducted an extensive screening of B. tabaci for the presence of one P-endosymbiont and two common S-endosymbionts, along with phylogenetic analyses of these symbionts to compare with host species from the cryptic B. tabaci complex. The reason we chose Arsenophonus and Cardinium as representatives of S-endosymbionts because they are the very common S-endosymbionts in whiteflies, and Wolbachia has been thoroughly investigated in the study of Bing et al. (2014). The results showed that P-endosymbiont P. aleyrodidarum was detected in all whitefly individuals while S-endosymbionts infection were varied among species. The variation in the prevalence of endosymbionts could be influenced by numerous factors such as host, environmental conditions, geographical location or even climate (Chu et al., 2011;Karut & Tok, 2014;Morag et al., 2012;Skaljac et al., 2010). In our study, Arsenophonus was abundant in Asia II 1, Asia II 3, and China 1 species but absent in the invasive species MEAM1, which is exactly consistent with previous studies (Bing et al., 2013;Karut et al., 2017); Cardinium was present in the MED, Asia II 1 and Asia II 3 species (20.3-40.7%) but was rarely detected in MEAM1 and not detected in China 1. Taken together, it seemed that these two S-endosymbionts had high prevalence in native species rather than invasive species, which is consistent with another S-endosymbiont Wolbachia but in contrast to Hamiltonella; Hamiltonella was found abundant in invasive species rather than native species (Bing et al., 2013). Previous studies showed that B. tabaci could be co-infected with particular pairs of S-endosymbionts, including Rickettsia and Hamiltonella, Hamiltonella and Cardinium, or Rickettsia and Arsenophonous; others were less common, such as Cardinium and Rickettsia, Hamiltonella and Arsenophonous, Cardinium and Wolbachia, and even three or four endosymbionts together (Gueguen et al., 2010;Karut & Tok, 2014;Marubayashi et al., 2014;Pan et al., 2012;Skaljac et al., 2010). In our study, we found evidence for a low rate (5.6%) of co-infection with Arsenophonous and Cardinium in four B. tabaci species (MEAM1 was the exception since Arsenophonous was not detected in this species), which has also been reported before (Bing et al., 2013;Chu et al., 2011;Parrella et al., 2014;Zchori-Fein, Lahav & Freilich, 2014). However, the reason of so few rate of co-infection by Arsenophonous and Cardinium is that Arsenophonous and Cardinium are potential reproductive manipulators that compete for resources inside the bacteriocytes, thus compromising the fitness of host (Gottlieb et al., 2008). We have one plausible explanation for the co-infection status, that is Cardinium is not restricted to the bacteriocytes (Skaljac et al., 2010), and perhaps the non-overlapping niche makes co-infection of Arsenophonous and Cardinium possible. In addition, the co-infection symbiont system in whiteflies may indicate the roles of dual endosymbionts: work as important mutually dependents to provide full complement of nutrients to their host (Rao et al., 2015) or affect the fitness and biology of the B. tabaci (Ghosh et al., 2018).
Our phylogenetic analyses indicated that B. tabacia mtCO1 sequences could be assigned to five distinct clades, which conformed to existing MEAM1, MED, Asia II 1, Asia II 3, and China 1 clades. Similarly, the sequences of the P-endosymbionts P. aleyrodidarum were assigned to their own clade and the phylogeny was similar with that of B. tabaci genetic groups to some extent. This may potentially indicate an ancient infection followed by vertical transmission and subsequent co-evolutionary diversification (Baumann, 2005). Meanwhile, it is important to note that the correlation was not perfect: sequences of P. aleyrodidarum from MEAM1 and MED were assigned to the same clade instead of the two distinct clades presented in the mtCO1 tree. The reason could be the dissemination of the MEAM1 and MED species; furthermore, the spread of these two invasive species in China has been frequently associated with founder effects that fix specific mtDNA variation(s) along with particular endosymbionts (Chu et al., 2011;Gueguen et al., 2010). Taken together, although there was similarity between the two trees of P. aleyrodidarum and B. tabaci, the genetic variation of primary symbiont might not be an ideal reflecting the genetic variation of the cryptic B. tabaci.
The S-endosymbionts, Arsenophonus and Cardinium, both showed a lack of congruence with the B. tabaci mtCO1 phylogram. This is consistent with the finding from Ahmed et al. (2013), who provided evidence for horizontal transmission of S-endosymbionts in the B. tabaci cryptic species complex based on phylogenies studies. There are substantial phylogenetic evidences showing that S-endosymbionts such as Wolbachia and Arsenophonus, undergoing horizontal transfer among host arthropod species (Ahmed et al., 2013;Chrostek et al., 2017;Kolasa et al., 2017;Li et al., 2017;Russell et al., 2003;Vavre et al., 1999). In some cases, the mechanisms for horizontal transmission of S-endosymbionts are already known, including transferring through parasitoid wasps (Ahmed et al., 2015;Gehrer & Vorburger, 2012), plants (Caspi-Fluger et al., 2012;Li et al., 2017) or even sexual transmission (Moran & Dunbar, 2006). Therefore, the potential horizontal transfer of S-endosymbionts in our samples could be one or combination of the above ways.
In summary, this study reported the varied prevalence of three endosymbionts within five B. tabaci cryptic species. The P-endosymbiont P. aleyrodidarum was detected in all whitefly individuals; the S-endosymbionts Arsenophonus was abundant in native species while Cardinium was common in the invasive species. In addition, the phylogenetic relationships between endosymbionts and their hosts B. tabaci probably indicated the vertical transmission and co-evolution of P. aleyrodidarum and B. tabaci; meanwhile, horizontal transfer of Arsenophonus and Cardinium may happen in our collecting samples. Our study not only reported current infection status of endosymbionts within B. tabaci populations in China, but also demonstrated that S-endosymbionts genetic variation may not reflect host genetic variation and should not be used to infer taxonomic relationships within the host species complex. If funding allows, more endosymbionts should be investigated and future investigations could be the contribution of endosymbionts to invasiveness, population expansion, or even competitiveness of whitefly species.