1 Introduction

Endosymbionts are common in plant-sapping insects and they have important relationships with their host species (Baumann 2005). Primary endosymbionts coevolve with their hosts, becoming key to host survival by assisting in obtaining essential nutrients (Baumann 2005; Rosell et al. 2010). Secondary endosymbionts have a facultative relationship with their hosts, providing fitness benefits such as in reproduction, host plant specialization, and increased tolerance to thermal stress and parasites (Chiel et al. 2009; Feldhaar 2011; Kaiser et al. 2010; Montllor et al. 2002; Oliver et al. 2003; Sintupachee et al. 2006; Tsuchida et al. 2004). In particular, secondary endosymbionts provide pivotal roles in virus transmission for vectoring species such as B. tabaci, which is a unique vector species of begomoviruses (Czosnek and Ghanim 2011).

Bemisia tabaci is a polyphagous pest species that feeds on various horticultural crops, ornamental crops and weed species (Cahill et al. 1996; Jones 2003). This species harbors primary endosymbiont Portiera aleyrodidarum along with at least six secondary endosymbionts, viz., Arsenophonus, Cardinium, Fritschea, Hamiltonella, Rickettsia, and Wolbachia (Bing et al. 2013a, b; Zchori-Fein et al. 2014). Recently, a new endosymbiont, Candidatus Hemipteriphilus asiaticus, was found to be a cryptic species of B. tabaci (Bing et al. 2013a, b). Secondary endosymbionts influence various biological characteristics, such as reproduction (Himler et al. 2011; Hunter et al. 2003; Zchori-Fein et al. 2001; Zchori-Fein and Perlman 2004), survival (Gottlieb et al. 2010; Kontsedalov et al. 2008; Liu et al. 2007; Thierry et al. 2011), insecticide resistance (Kontsedalov et al. 2008), and capacity for disease transmission to plants (Gottlieb et al. 2010). Given their various roles, endosymbionts are necessary for the modulation of host development and environmental impacts (Thao and Baumann 2004).

Bemisia tabaci is a species complex comprising at least 40 cryptic species that are morphologically indistinguishable, but genetically distinct in biological characteristics (De Barro et al. 2011). Among them, the Middle East-Asia Minor 1 (MEAM1, formerly B biotype) and Mediterranean (MED, formerly Q biotype) species are highly invasive and have dispersed to many countries in different continents (Dalton 2006; Horowitz et al. 2005; Pascual and Callejas 2004). In addition, many indigenous cryptic species of B. tabaci has been observed in Africa and Asia. To date, at least 23 indigenous cryptic species with different geographic distributions had been identified in Asia (Dinsdale et al. 2010; Ahmed et al. 2011; Hameed et al. 2012; Firdaus et al. 2013; Shah et al. 2013; Prasanna et al. 2015; Ellango et al. 2015; Hu et al. 2015, 2017; Götz and Winter 2016; Kumar et al. 2016; Jiu et al. 2017). Recently, Our previous study identified four indigenous cryptic species (Asia I, Asia II 1, Asia II 5 and Asia II 10) in Bangladesh (Khatun et al. 2018).

Each genetic group of B. tabaci exhibits a distinct composition of secondary endosymbionts. Several studies have investigated invasive cryptic species such as MEAM1 and MED from different geographic regions (Bing et al. 2013a, b; Chiel et al. 2007; Chu et al. 2011; Gueguen et al. 2010; Park et al. 2012; Skaljac et al. 2010, 2013). However, little information exists regarding the endosymbiont profiles of indigenous cryptic species in Asian countries.

The objective of this study was to determine the infection profiles of secondary endosymbionts in the cryptic species of B. tabaci from Bangladesh. Results from this study will improve our understanding of the composition of endosymbionts and their relationships with the Asian genetic groups of B. tabaci.

2 Materials and methods

2.1 Sample collection

Adult B. tabaci were collected from crop fields in different regions of Bangladesh from 2015 to 2017. Samples were preserved in 70% ethanol and kept at −20 °C for further analysis (Table S1). Identification and genetic diversity of four cryptic species of B. tabaci in these samples has been reported in a previous study (Khatun et al. 2018).

2.2 DNA extraction

Genomic DNA was extracted from a single adult B. tabaci using a pure link genomic DNA mini kit (Invitrogen, Carlsbad, CA, USA). The sample was placed in a 1.5 mL centrifuge tube containing 180 μL digestion buffer and 20 μL proteinase K (50 μg/mL) then incubated at 55 °C for 4 h. DNA samples were extracted and purified using genomic spin columns following the manufacturer’s protocol. DNA concentration was determined using a NanoPhotometer™ (Implen GmbH, Schatzbogen, Germany).

2.3 Screening of secondary endosymbionts

Five known endosymbionts (Arsenophonus, Cardinium, Hamiltonella, Rickettsia, and Wolbachia) were detected through PCR using 16S or 23S rDNA primers (Table 1). The reaction (25 μL) contained 13 μL Smart-Taq Pre-Mix (Solgent Co., Daejeon, Korea), 1 μL of each primer (10 pmol/μL), and 5 μLtemplate DNA solution (40 ng). Amplicons were separated using 1% agarose gel electrophoresis, stained with ethidium bromide solution, and visualized under Ultraviolet (UV) light. After excision from the gel, the amplicons were purified using the Wizard® PCR Preps DNA Purification System (Wizard® SV Gel, Promega Co., Madison, WI). They were then either sequenced directly or via cloning into the T-Blunt™easy plasmid vector (Promega Co., Madison, WI).

Table 1 Primers for amplifying DNA from secondary endosymbionts of Bemisia tabaci
Full size table

2.4 DNA sequence analysis

Cloned amplicons were sequenced using the BigDye® Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA) and analyzed with a 3100 Capillary DNA Sequencer (Applied Biosystems, Foster City, CA) at the Solgent Sequencing Facility (Solgent Co., Daejeon, Korea). Sequences were obtained from one individual per primer set and BLASTed against the NCBI database (Schaffer et al. 2001).

2.5 Phylogenetic analyses of secondary endosymbionts

Endosymbiont 16S and 23S rDNA sequences were aligned in Clustal Omega for the construction of a maximum-likelihood (ML) phylogenetic tree in MEGA 6.0. The model was selected based on GTR + G for Arsenophonus and Wolbachia, HKY + G for Cardinium and HKY + I for Rickettsia. Phylogeny robustness was tested with 1000 bootstraps (Felsenstein 1985).

2.6 Correlation analysis

Genetic differentiation among the endosymbionts was determined in DnaSP version 5.10 (Librado and Rozas 2009; Tajima 1989). Pairwise genetic distance was generated from 1023 permutations with the K2P model in MEGA 6.0 (Kimura 1980). Geographic distances were calculated from GPS coordinates measured during B. tabaci sample collections. The correlation between endosymbiotic genetic variability and geographic distances was analyzed using IBM SPSS software, version 23.

3 Results

3.1 Endosymbiont profiles of B. tabaci in Bangladesh

Infection profiles of five endosymbionts were determined from 110 individuals representing four Asian cryptic species of B. tabaci (Asia I, Asia II 1, Asia II 5, and Asia II 10; Table S1). Overall, the infection rates of Arsenophonus, Cardinium, Rickettsia, and Wolbachia were 93.6%, 86.4%, 31.8%, and 88.2%, respectively, but Hamiltonella was not detected in any individuals in this study (Fig. 1, Tables 2 and 3). Phylogenetic analysis of 23S rDNA sequences showed that Arsenophonus belonged to A1 and A2, but not to the A3 and A4 subgroups (Fig. 2). Sequences of both A1 and A2 exhibited a 0.19–3.18% variation and differences between A1 and A2 subgroups were 2.25–3.18% (Table 4, Table S2). The A1 subgroup was detected in Asia I and Asia II 1, while A2 was present in all four cryptic species (Figs. 1 and 2). Cardinium was present in C2 but not in the C1 or C3 subgroups (Figs. 1 and 3). The C2 subgroup was detected in Asia I, Asia II 1, and Asia II 5 with a variation of 0.25–0.74% (Fig. 3, Table 4, Table S3). Rickettsia belonged to the R1 and R2 subgroups;16S rDNA sequence variation ranged from 0.11–13.37% and differences between the R1 and R2 subgroups ranged from 10.15–13.37% (Figs. 1 and 4; Table 4, Table S4). Both R1 and R2 were detected at different rates in Asia I, Asia II 1, and Asia II 5 (Fig. 4). Wolbachia was present in W1 but not in the W2 or W3 subgroups. W1 subgroup was detected in all four cryptic species with a variation of 0.17–0.68% (Figs. 1 and 5; Table 4, Table S5).

Fig. 1
figure 1

Infection frequency (%) of secondary endosymbionts in four cryptic Bemisia tabaci species: Asia I, Asia II 1, Asia II 5, and Asia II 10. Arsenophonus and Rickettsia were divided into subgroups, indicated by different colors: A1 (bright blue), A2 (dark blue), R1 (bright green), and R2 (dark green)

Full size image
Table 2 Infection rates of secondary endosymbionts in four cryptic species of Bemisia tabaci from Bangladesh
Full size table
Table 3 Multiple infection patterns of the secondary endosymbionts in four cryptic species of Bemisia tabaci from Bangladesh
Full size table
Fig. 2
figure 2

Maximum-likelihood phylogenetic tree of 23S rDNA in Arsenophonus (600 bp) in infected Bemisia tabaci, using a GTR + G substitution model. Evolutionary analyses were conducted in MEGA 6. Sequences from GenBank are indicated in black color and the name of the samples from this study are shown in bright and dark blue color and include the genetic group, collection site, and accession number in Bangladesh

Full size image
Table 4 Comparison of the genetic variation in secondary endosymbionts in cryptic species of Bemisia tabaci in Bangladesh
Full size table
Fig. 3
figure 3

Maximum-likelihood phylogenetic tree of 16S rDNA in Cardinium (400 bp) in infected Bemisia tabaci, using a HKY + G substitution model. Evolutionary analyses were conducted in MEGA 6. Sequences from GenBank are indicated in black color and the name of the samples from this study are shown in pink color and include the genetic group, collection site, and accession number in Bangladesh

Full size image
Fig. 4
figure 4

Maximum-likelihood phylogenetic tree of 16S rDNA in Rickettsia (900 bp) in infected Bemisia tabaci, using a HKY + I substitution model. Evolutionary analyses were conducted in MEGA 6. Sequences from GenBank are indicated in black color and the name of the samples from this study are shown in bright and dark green color and include the genetic group, collection site, and accession number in Bangladesh

Full size image
Fig. 5
figure 5

Maximum-likelihood phylogenetic tree of 16S rDNA in Wolbachia (650 bp) in infected Bemisia tabaci, using a GTR + G substitution model. Evolutionary analyses were conducted in MEGA 6. Sequences from GenBank are indicated in black color and the name of the samples from this study are shown in red color and include the genetic group, collection site, and accession number in Bangladesh

Full size image

The four endosymbionts differed in infection rates depending on B. tabaci genetic groups (Table 2). Arsenophonus was 93.6% infected on average in all four cryptic species, with a range of 90.0–95.5%, except for in Asia II 10. Asia II 10 had 100% infection, but only one individual was examined in this group. The infection rate of A1 was higher than that of A2 in both Asia I and Asia II 1, but only A2 was infected into Asia II 5 and Asia II 10 (Fig. 1, Table 2). The infection rate of Cardinium was 80.6–100%, with the highest rate occurring in Asia II 5. The infection rate of Rickettsia was 13.6–50.0% with R1 being the highest in Asia II 1 (35%) and R2 being the highest in Asia I (25.4%). The infection rate of Wolbachia was 75–100% with the highest rate observed for Asia II 5 (Table 2). Otherwise, we did not find any significant relationships between the infection rates of endosymbionts and crop species infested by B. tabaci (Table S1).

3.2 Multiple infections of B. tabaci endosymbionts

All B. tabaci individuals were multiplied infected in various combinations by the four endosymbionts (Table 3). We identified 12 types of co-infection patterns, with multiple infections of 2, 3, or 4 species of endosymbionts. Single infections were identified only in A1 in 4.5% of Asia I and C2 in 10.0% of Asia II 1 cryptic species. Arsenophonus was the most common endosymbiont found in multiple infections. Infection patterns were more diverse in Asia I (8 patterns) and Asia II 1 (6 patterns) than in Asia II 5 (3 patterns). The most common combination was A1 + C2 + W1, which was the highest in Asia I (41.8%) and Asia II 1 (35.0%), whereas the A2 + C2 + W1 combination was only present in Asia II 5 at a high rate (86.4%). Among Arsenophonus and Rickettsia co-infections A1 was co-infected with R2 but A2 was co-infected with R1 (Table 3).

3.3 Correlation between genetic variation of endosymbionts and geographic distance of cryptic species

The variations in ribosomal DNA nucleotide sequences of Arsenophonus, Cardinium, and Wolbachia did not correlate with the geographic distribution of B. tabaci in Bangladesh. However, rDNA sequences of Rickettsia differed significantly and had a high distribution frequency (r = 0.428, P = 0.023; Table 5). Namely, the R1 subgroup was mostly present in the north, while R2 was found in the southern regions of Bangladesh (Fig. 6). Thus, Rickettsia subgroups had different distributions in Bangladesh, but the distributions of the other endosymbionts were not related to geographic distribution.

Table 5 Correlations between genetic variation in endosymbionts and geographic distance of cryptic species of Bemisia tabaci from Bangladesh
Full size table
Fig. 6
figure 6

Geographic distribution of Arsenophonus (a) and Rickettsia (b) subgroups of Bemisia tabaci in Bangladesh. Two subgroups of both Arsenophonus and Rickettsia are shown in bright blue (A1), dark blue (A2), bright green (R1), and dark green (R2)

Full size image

4 Discussion

Our study showed that profiles of the five species of endosymbionts in the indigenous Asian cryptic species of B. tabaci identified in Bangladesh were highly variable. Overall, the infection rate of Arsenophonus was the highest (93.6%), followed by Cardinium and Wolbachia (86.4% and 88.2%, respectively). Rickettsia only showed a moderate level of infection (31.8%). Hamiltonella was not detected in any individuals. These infection rates were complex and varied among the four Asian cryptic species (Asia I, Asia II 1, Asia II 5 and Asia II 10) of B. tabaci.

Arsenophonus was the most prevalent among the four Asian cryptic species. This is consistent with the findings of previous studies (Bing et al. 2013a, b; Ansari et al. 2017; Hashmi et al. 2018). Arsenophonus was detected in all indigenous genetic groups identified from China and India, such as Asia 1, Asia II 1, Asia II 5, and Asia II 7. In India, among all Asian cryptic species, Arsenophonus infection rate was higher in Asia 1 (70.0%) (Ansari et al. 2017; Hashmi et al. 2018). The infection rate of Arsenophonus was higher in Asia II 1 (78.9%) than in Asia II 3 and Asia II 7 from China (Bing et al. 2013a, b). Arsenophonus was also abundant in China I, but had not been previously detected in China (Bing et al. 2013a, b; Tang et al. 2018). Many studies have shown that Arsenophonus infection rates in invasive species are low in MEAM1 (Chiel et al. 2007; Chu et al. 2011; Gottlieb et al. 2008; Gueguen et al. 2010; Marubayashi et al. 2014; Skaljac et al. 2010; Thierry et al. 2011) or absent in MEAM1 and MED (Bing et al. 2013a, b). However, some MED populations from the USA, Montenegro (Q1), Israel (Q2), and Burkina Faso (Q3) were highly infected with Arsenophonus (Bing et al. 2013a, b; Gueguen et al. 2010; Skaljac et al. 2010; Skaljac et al. 2013). Therefore, these findings suggest that Arsenophonus is the most prevalent endosymbiont in Asian cryptic species, which is unlike the profiles of invasive species MEAM1 and MED of B. tabaci.

Cardinium infection rates were high in Asia I, Asia II 1 and Asia II 5 from Bangladesh. In particular, Cardinium infected all investigated individuals (n = 22) of Asia II 5. High infection rates of Cardinium were also reported in Asia II 1 from India (Ansari et al. 2017; Hashmi et al. 2018) and Asia II 3 and Asia II 7 from China (Bing et al. 2013a, b). Otherwise, Cardinium infection is rare in MEAMI and MED (Chu et al. 2011; Gueguen et al. 2010; Thierry et al. 2011). Similar to Arsenophonus, Cardinium had high infection rates in the Asian genetic groups, but low infection rates in the invasive genetic groups.

Hamiltonella was absent in all Asian cryptic species from Bangladesh, and these results are consistent with the reports of previous studies (Bing et al. 2013a, b; Singh et al. 2012). Hamiltonella is known to infect only two cryptic species, MEAM1 and MED and these species are highly invasive worldwide (Bing et al. 2013a, b; Chu et al. 2011; Gottlieb et al. 2008; Gueguen et al. 2010; Marubayashi et al. 2014; Skaljac et al. 2017; Thierry et al. 2011). Hamiltonella infection in MED (previously Q-biotype) is variable in the subgroups of this genetic group. For example, Hamiltonella has high infection rates in Q1 populations in China, but is absent in Q2 and ASL of MED (Bing et al. 2013a, b; Chu et al. 2011; Gottlieb et al. 2008; Gueguen et al. 2010; Marubayashi et al. 2014; Skaljac et al. 2017; Thierry et al. 2011). Hamiltonella is also absent in the sub-Saharan Africa 1–5 cryptic species of B. tabaci (Ghosh et al. 2015).

Rickettsia was identified in MEAM1 of B. tabaci, which is the first recorded occurrence in an insect (Gottlieb et al. 2006). Analysis of its genome sequence revealed that it grouped together with R. bellii (Rao et al. 2012). Our study showed that Rickettsia was detected in three cryptic species in Bangladesh. Its infection rate was moderate, but variable with the highest infection rate in Asia II 1 (50.0%), a moderate infection rate in Asia I (32.8%), and the lowest infection rate in Asia II 5 (13.6%). However, its infection rate was higher in Asia I than in Asia II 1 and Asia II 7 in Indian populations (Hashmi et al. 2018). Rickettsia was also detected in Asia II 3, Asia II 7, China 1, and MEAM1, but was absent from Asia II 1 and MED in China (Bing et al. 2013a, b; Chiel et al. 2007; Chu et al. 2011; Gueguen et al. 2010; Thierry et al. 2011). This suggests that the infection rate of Rickettsia is variable in different genetic groups and geographic regions. For example, our study showed that its infection rate in Asia II 1 in Bangladesh had a significantly different geographic distribution than those in China and India.

Wolbachia is one of the most common insect endosymbionts, with various infection rates for different genetic groups (Correa and Ballard 2016). Our study showed that Wolbachia was highly prevalent in all cryptic species of B. tabaci from Bangladesh, with a 100% infection rate in Asia II 5. High infection rates of Wolbachia were reported in Asia II 1, Asia II 7, and China 1, but Wolbachia was absent in Asia II 3 from China (Bing et al. 2013a, b). However, its infection rate was moderate in Asia 1 and Asia II 7 and low in Asia II 1 from India (Hashmi et al. 2018). This suggests that Wolbachia infection rates vary regardless of the genetic groups of B. tabaci. Otherwise, Wolbachia infection was high in Bangladesh and China, but low in India. However, this difference may not be associated with geographic distribution. In fact, the Wolbachia infection rate is highly variable across regions and populations. For example, the Wolbachia infection rate in both MED and MEAM1 ranges from 0% to 100% throughout China (Bing et al. 2013a, b, 2014; Pan et al. 2012). In Europe and western Africa, Wolbachia infection of MEAM1 and MED was 0–8.3% and 0–33% (Chiel et al. 2007; Chu et al. 2011; Gnankine et al. 2013; Gueguen et al. 2010; Nirgianaki et al. 2003; Skaljac et al. 2010; Thierry et al. 2011).

Phylogenetic analyses showed that both Arsenophonus and Rickettsia were detected in two subgroups, but Cardinium and Wolbachia were detected in a single subgroup in Bangladesh. Arsenophonus was infected into all four cryptic species. Our study showed that A2 infected all four cryptic species, but A1 infected only in Asia I and Asia II 1. This result suggests that Arsenophonus infection depends on the genetic group of B. tabaci. Similarly, two Rickettsia subgroups (R1, R2) were detected in three cryptic species in Bangladesh with different infection rates; R1 was higher in Asia I and R2 was higher in Asia II 1 and Asia II 5. The presence of two Rickettsia subgroups in B. tabaci has been reported in previous studies (Singh et al. 2012; Bing et al. 2013a, b; Ghosh et al. 2015). Singh et al. (2012) observed RI1 and RI2 strains in Indian populations. Bing et al. (2013a, b) reported that one subgroup detected in MEAM1 and Asia II 3 is widely separated from the other group found in Asia II 7 and China 1. Ghosh et al. (2015) described two clusters that had more than 8.5% nucleotide distance in various genetic groups of B. tabaci infesting cassava in Africa. Our results showed that 16S rDNA sequence variation was low within subgroups but subgroups were highly distinct, with differences of 10.15–13.37%. This suggests that there are at least two genetically distinct Rickettsia species in the B. tabaci species complex. Further, the two subgroups were present in the same cryptic species of B. tabaci, although the ratios were different. This suggests that each subgroup of Rickettsia is not confined within certain genetic groups of B. tabaci. This characteristic is also found in Rickettsia infection in the African genetic groups of B. tabaci (Ghosh et al. 2015). For example, sub-Saharan Africa 1- subgroup 3 (SSA 1-SG3) is infected by two groups of Rickettsia at different rates.

Our study also showed that there is a relationship between genetic variation of Rickettsia and geographic distribution, contrary to other endosymbionts. Namely, R1 subgroup was mostly present in the northern region, while R2 subgroup was distributed in the southern region of Bangladesh. Similarly, Singh et al. (2012) identified two subgroups of Rickettsia in B. tabaci collected from various regions of Central and Northern India. R1 was detected in the northern region, but R2 was detected in the central region of India. Our phylogenetic analysis showed that the R2 subgroup in Bangladesh was clustered with R2 strains collected from Akola and Indore, which are in the central region of India. Thus, the geographic distributions of the two Rickettsia subgroups are distinct in these two countries. Results from this study also showed that the distribution of the two Rickettsia subgroups was not associated with genetic groups of B. tabaci, because each Asian cryptic species was infected by both subgroups of Rickettsia at various rates. Further analysis is required to determine the influence of geographic distribution.

Multiple infections of various secondary endosymbionts are common in most host species of insects (Zchori-Fein et al. 2014). We identified 12 types of co-infection patterns among four endosymbionts, particularly among two subgroups of Arsenophonus and Rickettsia. Every individual in this study was infected by at least one endosymbiont. Arsenophonus was the dominant endosymbiont in most co-infections. Interestingly, the combination of Arsenophonus and Rickettsia subgroups was always present as A1 + R2 or A2 + R1. Similarly, data from Israel and Burkina Faso demonstrated A1 + R co-infection in MED (Q2) and A2 + R in MED (Q3), respectively (Gueguen et al. 2010). Arsenophonus in MED (Q1 and Q2) is the R1 subgroup (Skaljac et al. 2017). Together, these data suggest that Asian cryptic B. tabaci may have a unique characteristic that is different from other cryptic species such as MED.

In conclusion, our results revealed the complexity of endosymbiont profiles in four cryptic B. tabaci species from Bangladesh. Specifically, each cryptic species exhibited characteristic patterns of infection rates and combinations of endosymbionts. Notably, Arsenophonuswas a major endosymbiont but Hamiltonella was absent from all Asian cryptic species of B. tabaci. Two genetically distinct subgroups of Rickettsia were associated with geographic distribution. This study provides important insights that can help improve the control techniques of whiteflies.