A change in the bacterial community of spider mites decreases fecundity on multiple host plants

Abstract Bacterial symbionts may influence the fitness of their herbivore hosts, but such effects have been poorly studied across most invertebrate groups. The spider mite, Tetranychus truncatus, is a polyphagous agricultural pest harboring various bacterial symbionts whose function is largely unknown. Here, by using a high‐throughput 16S rRNA amplicon sequencing approach, we characterized the bacterial diversity and community composition of spider mites fed on five host plants after communities were modified following tetracycline exposure. We demonstrated that spider mite bacterial diversity and community composition were significantly affected by host plants and antibiotics. In particular, the abundance of the maternally inherited endosymbionts Wolbachia and Spiroplasma significantly differed among spider mites that were reared on different plant species and were completely removed by antibiotics. There was an overall tendency for daily fecundity to be lower in the mites with reduced bacterial diversity following the antibiotic treatment. Our data suggest that host plants and antibiotics can shape spider mite bacterial communities and that bacterial symbionts improve mite performance.

, both primary and secondary symbionts are affected by the plant type on which the aphids feed. Host plants shape the diversity and abundance of larval gut symbiotic bacteria of the Colorado potato beetle, Leptinotarsa decemlineata, and thus affect the insect's ability to manipulate plant defenses in Solanum hosts (Chung et al., 2017). Similarly, antibiotic treatment can also alter the bacterial communities of herbivores (Lehman, Lundgren, & Petzke, 2009;Zouache, Voronin, Tran-Van, & Mavingui, 2009); in particular, antibiotic treatment significantly influences the relative abundance of Wolbachia, Spiroplasma, and/or Cardinium in the spider mite Tetranychus urticae (Staudacher et al., 2017). Antibiotics are routinely used to eliminate some endosymbionts from a wide range of insect species (Li, Floate, Fields, & Pang, 2014;Wilkinson, 1998). Although previous studies showed that host plants and antibiotic may be important factors in shaping the bacterial community of several herbivorous arthropod species, little is known about the effect of host plants and antibiotic treatment on the entire bacterial communities of spider mites.
Among spider mites, Tetranychus truncatus is a highly polyphagous species found on over 60 host plant species, including economically important crops such as bean, cotton, cucumber, tomato, and eggplant (Bolland, Gutierrez, & Flechtmann, 1998). Tetranychus truncatus is the dominant mite species in China and has diverse host plants . We previously demonstrated that T. truncatus harbor various endosymbiotic bacteria, including Wolbachia, Cardinium, and Spiroplasma (Zhang, Chen, et al., 2016), and affect host reproduction through cytoplasmic incompatibility (CI) (Zhang, Yang, Zhu, & Hong, 2018;Zhao, Zhang, & Hong, 2013). Infection prevalence of Wolbachia in T. truncatus natural populations is related to ecological factors, such as host plant, temperature, and climate , but the interaction between these factors and host fitness is not clear.
In this study, we explore host plant and antibiotic influences on spider mite symbiotic bacterial communities and performance under controlled environmental conditions. Recent developments in sequencing technologies and molecular tools have enhanced opportunities to characterize the microbial diversity associated with spider mites (Sugio, Dubreuil, Giron, & Simon, 2015). We used a high-throughput 16S rRNA amplicon sequencing procedure to investigate whether antibiotic treatment and host plant influence the composition and structure of T. truncatus bacterial communities and host performance. The results highlight roles of host plant and antibiotics in shaping the bacterial community of herbivores and highlight impacts of bacterial diversity on mite fecundity. F I G U R E 1 Overview of experimental procedure describing the different rearing condition used to compare the performance of antibiotic-untreated and antibiotic-treated spider mites after maintenance on different host plant and collection of samples for DNA extraction

| Spider mite antibiotic treatment and rearing
Spider mites were originally collected from bean (Phaseolus vulgaris L.) leaves in Hohhot, Inner Mongolia, northeast China in August 2014. Mites were reared on detached bean leaflets in a climate-controlled room at 25 ± 1°C, 60% relative humidity, and a light:dark (L:D) photoperiod of 16:8 hr. Individuals used to establish various mite strains were all derived from one adult female to minimize genetic variation between lines (Figure 1).
To investigate the effect of antibiotics on microbial communities associated with T. truncatus spider mites, lines of mites were treated with antibiotics. For the antibiotic treatment, 30 adult females mites were reared on bean leaf disks on cotton wool soaked with tetracycline solution (0.1%, w/v) for three generations. Untreated control mites were reared on leaf disks placed on water-saturated cotton wool. After treatment, lines were maintained in a mass-rearing environment without antibiotics for approximately 15 generations to establish T. truncatus lines on different host plants. The treated lines are referred to as AB-T and the untreated lines as AB-UT.

| Establishment of T. truncatus lines on different host plants
To characterize the microbial communities associated with spider mites that feed on different hosts, lines were established by transferring adult female mites (ca. 400) of both the AB-UT and AB-T lines from bean to cotton, tomato, cucumber, or eggplant. Four independent lines on each plant have been created. Lines were maintained on detached leaves from these hosts for six generations in a climatically controlled environment at 25 ± 1°C with 60% relative humidity and a 16L:8D photoperiod. More than 200 mites were transferred at each generation.
To generate mites for experiments, we used 4-to 6-day-old adult females from each strain to produce eggs on the adaxial surface of detached plant leaflets placed on water-soaked cotton. After 6 hr of egg production, all mites were removed from the leaflets. The eggs were allowed to hatch and mature in a climate-controlled room for another 13 days; this was done to obtain spider mites of the same age. The adult female mites were then collected for the performance assay and for DNA extraction (Figure 1).

| Spider mite performance assay
To investigate the effect of host plants on spider mite performance, we assessed spider mite fecundity on different plants using the method described by Staudacher et al. (2017) with minor modifications. The mite lines were maintained on detached leaves from five different plants in a climate room (25 ± 1°C, 16 hr:8 hr, light:dark, 60% RH). For the performance assay, leaf disks of bean, cucumber, cotton, and eggplant (diameter ca. 3 cm) and tomato leaflet (at least 4 cm in length) were placed on a cotton bed soaked in water. Five adult female mites (2 ± 0.25 d) were placed on each leaf disk (or leaflet), 24 leaf disks (or leaflets) per line (treated or not with antibiotics) and per plant species (n = 5 species). After 4 days, the number of eggs produced by mites was recorded using a stereomicroscope.

| DNA extraction, quantitative real-time polymerase chain reaction (qPCR), and 16S rRNA amplicon sequencing
Fifteen adult female spider mites from each of the replicated plants belonging to the five different plant species were pooled to form one sample for DNA extraction. DNA was extracted from each sample using the E.Z.N.A. ® Soil DNA Kit (Omega Bio-Tek, Norcross, GA, USA) according to manufacturer's protocols.
The PCR products of primers specific for wsp from Wolbachia and 16S rRNA from Spiroplasma were amplified by conventional PCR; then, the PCR products were purified using the AxyPrep TM DNA Gel Extraction Kit (Axygen) and cloned into a pEASY-T1 vector (TransGen Biotech, Beijing, China). A standard curve was generated using a serial dilution of plasmids containing one copy of the target sequence. Absolute quantification of wsp and 16S rRNA copy number was calculated using threshold values (C t ).
The V3-V4 region of the 16S rRNA gene was amplified from each sample using the primer pair 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACNNGGGTATCTAAT-3′). PCR was performed in a 25 μl volume that contained 12.5 μl 2× Taq Master Mix (Vazyme Biotech, China), 0.5 μl primer (20 μM each), and 1 μl of DNA, or ultrapure water for the PCR-negative controls. The PCR conditions were as follows: 95°C for 5 min, followed by 27 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 45 min, and a 72°C final extension for 10 min. PCR product quality was verified by gel electrophoresis.
Amplicons were extracted from 2% agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) according to manufacturer's instructions and quantified using QuantiFluor™-ST (Promega, USA). Purified PCR products were quantified by Qubit ® 3.0 (Life Invitrogen), and 24 amplicons with different barcodes were mixed equally. DNA concentration was adjusted to 25-35 ng/μl per sample before sequencing. All DNA samples were sent for sequencing, except for those of the treated line 4 on bean and line 3 on cucumber and the untreated line 3 on cotton, because the DNA concentration of three samples was less than required by the criteria. The pooled DNA product was used to construct an Illumina paired-end library following Illumina's genomic DNA library preparation procedure. Then, the amplicon library was paired-end (2 × 250 bp) sequenced on an Illumina HiSeq 2500 platform (Shanghai Biozeron Co., Ltd) using standard protocols.

| Sequence processing and analyses
Sequences were provided as adapter-clipped fastq files and analyzed in Quantitative Insights into Microbial Ecology (QIIME), which is a standard pipeline for microbial community analysis (Caporaso et al., 2010). Raw fastq files were demultiplexed and quality-filtered using QIIME 1.17 with the following criteria: (a) The 250-bp reads were truncated at any site with an average quality score <20 over a 10-bp sliding window, and truncated reads that were shorter than 50 bp were discarded; (b) exact barcode matches, two-nucleotide mismatches in primer matching, and reads that contained ambiguous characters were removed; and (c) only sequences that overlap longer than 10 bp were assembled based on their overlapping sequences. Reads that could not be assembled were discarded.
The phylogenetic affiliation of each 16S rRNA gene sequence was analyzed with RDP Classifier (https://rdp.cme.msu.edu/) against the Silva 16S rRNA database using a confidence threshold of 70% (Amato et al., 2013). To avoid bias, OTUs (<0.1% abundance) were excluded from subsequent analysis. Rarefaction analysis was generated using Mothur 1.21.1 to determine Good's coverage, Chao 1, and Simpson and Shannon diversity indices (Schloss et al., 2009).

| Statistical analyses of bacterial community
All statistical analyses were carried out in R ver 3.3.1 (R Development Core & Team, 2016).
Diversity of the bacterial communities in the samples was determined by computing Simpson and Shannon indices, while species richness was estimated through counting OTUs or computing ACE (abundance-based coverage estimator) and Chao 1 indices (Hill, Walsh, Harris, & Moffett, 2003;Hughes, Hellmann, Ricketts, & Bohannan, 2001;Shannon, 1948;Simpson, 1949). To determine whether diversity measures were significantly different between samples from the different host/antibiotic treatments, we used two-way ANOVAs after validation of the normal distribution of the residuals.
To determine whether feeding in different host plants or/and antibiotics caused major changes in community structure, a Bray-Curtis dissimilarity matrix was calculated and analysis of molecular variance (AMOVA) was used. Multi-response permutation procedures (MRPP) analyses were also used to compare community composition between samples from the different treatments. Variation in bacterial taxonomic composition among samples was visualized using principal coordinates analyses (PCoA). PCoA was performed using the R package "vegan." To test the effect of antibiotics and/or host plant species on mite oviposition, we constructed a general linear model with two factors, antibiotics and host plant, treated as a fixed factor, and average number of eggs per females per day as response variables, which were firstly verified to follow normal distributions. If interaction terms were insignificant, mite oviposition was subjected to a one-way ANOVA. The R package "lsmeans" was used for multiple

| Overview of T. truncatus bacterial communities
Analyses of 16S rRNA amplicon sequences yielded a total of 865,578 reads after quality check, with an average of 23,394 sequences per sample. The majority of the rarefaction curves approached saturation, which indicated that our sampling depth accurately characterized the bacterial diversity of the majority of these samples (Supporting Information Figure S1).

Diversity and species richness index values are provided in
Supporting Information Table S1, and their mean (±SEM) is plotted in  Figure S2). Good's coverage for each sample was more than 99% (Supporting Information Table S1). Overall most of the sequences from the bacterial communities associated with T. truncatus lines belonged to Gammaproteobacteria (44.95%), followed by Alphaproteobacteria (33.91%), Betaproteobacteria (13.54%), and Flavobacteria (2.22%) (Supporting Information Table S2).

| Changes in bacterial communities with host plants and antibiotic treatment
The species diversity of spider mite bacterial communities, indi-  index was reduced in lines that had been exposed to antibiotics when reared on the same plant species, which suggest that the species richness of bacteria tended to decrease after antibiotic exposure ( Figure 2). Moreover, there were substantial plant-specific variations of bacterial composition at the family ( Figure 3) and phylum levels (Supporting Information Figure S2), and in the relative abundance of the 10 most common OTUs (Supporting Information

| Abundance of Wolbachia and Spiroplasma on different plants
The facultative endosymbionts Wolbachia and Spiroplasma were not detected in any of the spider mites reared on any of the plants in populations treated with antibiotics (Supporting Information Figure   S4; Figure 6). There were significant effects of host plant species

| D ISCUSS I ON
In this study, we investigated the impact of host plant and antibiotic treatment on bacterial diversity and bacterial community composition of the spider mite T. truncatus. We demonstrated that the bacterial diversity of T. truncatus was influenced by host plant species and antibiotic. In particular, the abundance of the facultative endosymbionts Wolbachia and Spiroplasma was also influenced by host plant species and was completely eliminated by the antibiotic treatment. Intriguingly, when assessing offspring production in the mites exposed to the different conditions, we found that daily fecundity

| Host plant and antibiotic treatment effects on the spider mite microbial communities
In this study, the host plants tested strongly altered the composition of the microbial community (diversity and abundance) in spider mites. Similarly, research on Colorado potato beetles (Chung et al., 2017), whiteflies (Su et al., 2016), aphids (Guidolin & Cônsoli, 2016), and pine processionary moths (Strano et al., 2017)  here may reflect the role of the microbial community in the ability of spider mites to cope with different host plants (Hansen & Moran, 2014;Jaenike, 2015;Tsuchida, Koga, & Fukatsu, 2004), but testing such hypothesis is beyond the scope of this study (i.e., fecundity was only characterized on the host plants on which the lines were maintained).
The relative abundance of maternally inherited endosymbionts should be taken into account when studying bacterial communities across host populations, as variation in bacterial density may impact host biology and ecology (Fromont, Riegler, & Cook, 2017). In the current study, both Wolbachia and Spiroplasma were detected in T. truncatus, which were also widespread in T. urticae (Enigl & Schausberger, 2007;Hoy & Jeyaprakash, 2005). We also found that host plant had effects on the relative amounts of Wolbachia and Spiroplasma. Changes in relative abundance of symbionts with host have been noted for other polyphagous insects, including pea aphids (Tsuchida et al., 2002), cowpea aphid (Brady & White, 2013), chestnut weevils (Toju & Fukatsu, 2011), and others (Pan et al., 2014).
There are several potential explanations for the effects of host plant species on symbiotic bacterial communities. First, plant secondary metabolites or phytotoxins may possess variable antibacterial activities (Harborne, 1993), which could influence population growth (Kohl & Dearing, 2012). Each plant has a different secondary metabolite profile, such as terpenoids and glycoalkaloids in tomato (Falara et al., 2011;Milner et al., 2011), glycoalkaloids in eggplant (Milner et al., 2011), cucurbitacin in cucumber (Balkema-Boomstra et al., 2003), protease inhibitors in bean (Visôtto, Oliveira, Ribon, Mares-Guia, & Guedes, 2009), and gossypol in cotton (Bottger, Sheehan, & Lukefahr, 1964). In addition to secondary metabolites, physical properties of different host plants, such as leaf toughness and trichomes, may influence bacterial communities, because they directly impact spider mite growth and physiochemical interactions between bacteria and their insect hosts (Chung et al., 2017). Alternatively, the host may manipulate its endosymbiont titer to compensate for specific F I G U R E 6 Sequence counts of two OTUs of antibioticuntreated and antibiotic-treated spider mites that were reared for six generations on different host plants. (a) Wolbachia; (b) Spiroplasma. Horizontal lines indicate the mean of biological replicates. Superscripts (a, b) above horizontal lines indicate significant differences between host plants (p < 0.05) F I G U R E 7 Fecundity (number of eggs laid/female/day) of spider mites on different plant species. Horizontal lines in the boxes represent medians, whiskers represent the 10th-90th percentiles, and dots represent data points outside of this range. "*" represents significant difference between antibiotic-untreated and antibiotictreated spider mites on the same host plant (p < 0.05) deficiencies in the nutrient profile of its host plant (Zhang, Cao, et al., 2016). However, at this stage, it is unclear whether the high relative abundances of Wolbachia and Spiroplasma in mite lines from tomato and eggplant are adaptively significant.
Previous studies have shown that antibiotic treatment alters the bacterial community of mites and other herbivores; in particular, antibiotic treatment influences the relative abundance of some facultative endosymbionts (Breeuwer, 1997;Lehman et al., 2009;Staudacher et al., 2017). Here, the number of OTUs and species diversity were reduced regardless of host plant. The facultative endosymbionts Wolbachia and Spiroplasma were completely absent from the antibiotic-treated populations, which were consistent with the effects produced by antibiotics in previous studies (Staudacher et al., 2017;Xie, Sun, Xue, & Hong, 2016).
One unexpected finding was that the relative abundance of Acinetobacter, Pseudomonas, Halomonas, Pelagibacterium, and Cloacibacterium was increased by antibiotic treatment. Prior exposure to plant toxins can enhance the diversity of gut microbes in herbivores (Kohl & Dearing, 2012 bacteria. This might correspond to de novo colonization after treatment. It has to be noted that, the symbiotic bacterial communities of arthropod herbivores may also be affected by host genotype (Brady et al., 2014), how the host impact the plant-bacterial community interaction in spider mite should thus be considered in future studies.

| Host plant and antibiotic treatment effects on T. truncatus performance
Many facultative endosymbionts are not essential for host survival but can have an important impact on insect life history traits (Giron et al., 2017;O'Neill, Werren, & Hoffmann, 1997). Our previous study showed that reproductive parasites, such as Wolbachia, Cardinium, and Spiroplasma, are widespread in T. truncatus (Zhang, Chen, et al., 2016), and those facultative endosymbionts influence host reproduction via various phenotypic effects (Engelstädter & Hurst, 2009 (Chung et al., 2013(Chung et al., , 2017 or enzymes, such as hydrolases, glucosidases, phosphatases, and glutathione transferases (Dowd & Shen, 1990;Shen & Dowd, 1991). For instance, the presence/absence of the bacterial endosymbionts Wolbachia, Cardinium, or/and Spiroplasma in the spider mite T. urticae has also been previously reported as altering distinct plant defense parameters and affecting mite performance, but there were no indications of a causal link between the two (Staudacher et al., 2017). Western corn rootworms (Diabrotica virgifera) infected with Wolbachia suppressed defense-related genes in maize roots and altered host performance compared with uninfected rootworms (Barr, Hearne, Briesacher, Clark, & Davis, 2010). Another explanation is that microbes could improve the nutritional properties of the herbivore's diet for certain plant species, which allows females to allocate more resources to egg production (Douglas, 1998;Feldhaar et al., 2007). Buchnera synthesize essential amino acids and other substances that are absent from their host aphid's diet, and if the symbiont is removed, the host grows very slowly and cannot reproduce (Douglas, 1996;Koga, Tsuchida, & Fukatsu, 2003). Wolbachia genome analysis revealed that they lack many essential biosynthetic pathways (Wu et al., 2004). Therefore, it is likely that Wolbachia impose a nutritional burden on their hosts, and host-symbiont competition for key resources, such as amino acids (Caragata, Rances, O'Neill, & McGraw, 2014), sugars (Markov & Zakharov, 2006), or iron (Gill, Darby, & Makepeace, 2014), could influence host performance.
Although the antibiotic treatments resulted in the complete removal of Wolbachia and Spiroplasma from the mites, various other bacterial strains were also affected, making attribution to specific bacteria difficult. Future experiments might focus on the performance of spider mites with or without symbionts from natural populations with multiple host plants. Polyphagous arthropod-associated microbes could have played a role in adaptation to new hosts and host range expansion (Chu, Spencer, Curzi, Zavala, & Seufferheld, 2013;Jaenike, 2015). Because T. truncatus is a polyphagous pest and harbors different species of bacterial symbionts, an association of reproductive bacteria with spider mites might help them adapt to new host plants.
Host plant-dependent impacts of symbiotic microorganisms on the fitness of herbivorous insects may be a widespread and currently unrecognized dimension in insect-plant interactions (Chandler et al., 2008). Here, we found that host plant and antibiotic treatment influenced the T. truncatus symbiotic bacterial community; in particular, there were effects on the relative abundances of the endosymbionts Wolbachia and Spiroplasma in T. truncatus, which in turn may have altered host performance. Further studies are needed to determine whether spider mite-associated microbes alter spider mite fitness by mediating plant defenses and detoxifying phytochemicals. Documenting the presence of the microbial community and identifying their effects on hosts can have important implications for the management of this pest species (Crotti et al., 2012;Oliver, Degnan, Burke, & Moran, 2010).

ACK N OWLED G M ENTS
This study was supported in part by a grant-in-aid for Scientific

CO N FLI C T O F I NTE R E S T S
The authors state that there are no conflict of interests.

AUTH O R S CO NTR I B UTI O N
YXZ and XYH conceived and designed the experiments; YXZ and YLS performed the experiments; YXZ and PYJ analyzed the data with guidance from AAH; YXZ, AAH, and XYH wrote the manuscript.
All authors discussed the results and commented on the manuscript.

E TH I C S S TATEM ENT
No specific permissions were required for the collection because T. truncatus is a pest on the bean. Tetranychus truncatus is not an endangered species in China and is not protected by law. No ethical approval was required to work with this species in our study.

DATA ACCE SS I B I LIT Y
Illumina sequence reads are available on the NCBI Sequence Read Archive (SRA) under accession number SRP158961.