Comparison of bacterial microbiota of the predatory mite Neoseiulus cucumeris (Acari: Phytoseiidae) and its factitious prey Tyrophagus putrescentiae (Acari: Acaridae)

Neoseiulus cucumeris is a predatory mite used for biological control of arthropod pests. Mass-reared predators are fed with factitious prey mites such as Tyrophagus putrescentiae. Although some information on certain endosymbionts of N. cucumeris and T. putrescentiae exists, it is unclear whether both species share bacterial communities. The bacterial communities in populations of predator and prey mites, as well as the occurence of potential acaropathogenic bacteria were analyzed. The comparisons were based on the following groups: (i) N. cucumeris mass-production; (ii) N. cucumeris laboratory population with disease symptoms; (iii) T. putrescentiae pure populations and; (iv) T. putrescentiae from rearing units of N. cucumeris. Only 15% of OTUs were present in all samples from predatory and prey mite populations (core OTUs): the intracellular symbionts Wolbachia, Cardinium, plus other Blattabacterium-like, Solitalea-like, and Bartonella-like symbionts. Environmental bacteria were more abundant in predatory mites, while symbiotic bacteria prevailed in prey mites. Relative numbers of certain bacterial taxa were significantly different between the microbiota of prey mites reared with and without N. cucumeris. No significant differences were found in the bacterial communities of healthy N. cucumeris compared to N. cucumeris showing disease symptoms. We did not identify any confirmed acaropathogenic bacteria among microbiota.

conditions and reports of horizontal bacterial transmission between trophic levels 12 suggest these environments offer an ideal setting for the comparative study of the bacterial microbiota of predatory and prey mites.
Previous studies reported the presence of endosymbiotic and pathogenic bacteria in phytoseiid mites 8,9 . Metaseiulus occidentalis (Nesbitt) is probably the best studied species 7 . Several bacteria were detected including: the pathogenic Serratia marcescens; two Rickettsia-like bacteria; the gut symbionts Enterobacter and Bacteroidetes; and the endosymbionts Wolbachia and Cardinium 7 . Enigl and Schausberger 13 screened several phytoseiid species for the presence of Wolbachia, Cardinium and Spiroplasma, they found Cardinium in Euseius finlandicus (Oudemans) and Spiroplasma in Neoseiulus californicus (McGregor). Similarly, Wolbachia and Cardinium showed different patterns of infection in the phytoseiid mite N. paspalivorus (De Leon) depending on geographic origin 14 . Interestingly, the observed postzygotic reproductive incompatibility among populations was associated with the presence of endosymbiotic bacteria. Gols et al. 15 discovered the bacterium Acaricomes phytoseiuli in several commercial populations of the phytoseiid mite Phytoseiulus persimilis (Athias-Henriot). The infected predatory mites exhibited lower fecundity and longevity and reduced attraction to plant volatiles induced by spider mites. This was designated as non-responding syndrome and rendered infected P. persimilis populations unsuitable for effective control of spider mites.
Neoseiulus cucumeris (Oudemans) is one of the most widely employed predators in augmentative biological control programs against thrips species such as Frankliniella occidentalis (Pergande) and Thrips tabaci Lindeman (Thysanoptera: Thripidae) mostly in protected crops [18][19][20] . For the mass-rearing of N. cucumeris, the acaricid mite T. putrescentiae is used as factitious prey 21 . While some information on certain endosymbionts of N. cucumeris and T. putrescentiae 2,3 exists there is no information regarding the intrinsic microbiota of these species, or the effects of predatory-prey mite interactions on the ecology of these microbiota. Although mites are kept at high densities in mass rearing units it is still unclear whether both species share bacterial communities due to horizontal transfer via predation, contact or the feces 6 . The current study, to our knowledge, is the first direct comparison of bacterial microbiota from Phytoseiidae mites and their factitious prey mites.
Understanding the composition of the bacterial microbiota in both N. cucumeris and T. putrescentiae will allow for detecting pathogenic bacteria in the mass-rearing systems and perhaps offer opportunities for the manipulation of the bacterial community to improve predatory mite health. Both of the above mentioned scenarios can potentially have a substantial impact on biological control. In the present study we use Illumina sequencing of the 16S rRNA gene region and bioinformatics tools to (i) characterize the entire bacterial microbiota (ii) identify potentially pathogenic bacteria and iii) compare bacterial microbiota to examine the effects of predator-prey interactions and different rearing conditions on microbial ecology.

Results
Bacterial microbiota characterization. The 16S rRNA gene libraries included 818,413 sequences classified in 75 OTUs. The proportion of bacterial taxa in the groups of predatory and prey mites were visualized in Krona projection (Fig. 1). The minimal number of reads was 17,613 ( Fig. 2A). No known acaropathogenic bacteria were identified in the prey or predatory mites either from the mass rearing or the laboratory population with disease symptoms (Table S1).

Bacterial microbiota comparisons: N. cucumeris (predator) vs. T. putrescentiae (factitious prey).
Total numbers of OTUs found in predatory mites compared to prey mites were marginally non-significant (Fig. 2B). The Inverse Simpson index was 1.5 times higher in prey compared to predatory mite samples (U (1,14) = 8, P = 0.029) (Fig. 2C). Bray-Curtis dissimilarity index showed higher dissimilarity in bacterial composition within prey mite samples than within the predatory mite samples. There were significant differences between the bacterial composition of prey mites compared to predator mites (i.e. factor 1, Table S2) (two-way PERMANOVA, F = 0.947, P = 0.011). There was also significant difference between the populations of mites (i.e. factor 2, Table S2) (two-way PERMANOVA, F = 0.66, P = 0.012), but the interaction between mite species and different population factors was not significant (F = −2.19, P = 0.999). When analyzed using Jaccard similarity matrix, the results were in agreement with previous analyses. Species of mite was a significant factor affecting the bacterial diversity of microbiota (F = 0.245, P = 0.045), as was the population that they originated from (F = 0.303, P < 0.001) but their interaction was not significant (F = −1.89, P = 0.972). Sample dissimilarity was visualized by PCoA; the first axis explained 56% and the second axis 15% of variability in the data set (Fig. 3). Microbiota of predator and prey mites formed two distinct clustered when represented by PCoA. Distances between OTUs along the first axis can be attributed to differences between predator and prey microbiota, with the exception of Tyro5 and Tyro8 (T. putrescentiae from N. cucumeris mass rearing units) where the bacterial composition was similar to that of the predatory mites. The following bacterial taxa were associated with prey mites: Solitalea-like (OTU6), Bartonella-like (OTU3), Wolbachia (OTU1 and 45), Blattabacterium-like (OTU5), Brenneria (OTU9) and Xenorhabdus (OTU43). The following taxa were associated with predatory mites: Cardinium (OTU12), Bacillus (OTU21) and Staphylococcus (OTUs2, 7, 19 and 23).

Continued
Significant differences were detected between the microbiota composition of T. putrescentiae populations with and without the presence of the predator (one way PERMANOVA; F = 5.337, P = 0.024) according to Bray-Curtis distance matrix. The random forest algorithm (forest error rate = 0.33) revealed that the following OTUs were important for differences and based on METASTATS were abundant in the microbiome of T. putrescentiae from N. cucumeris rearing units compared to T. putrescentiae pure colonies.: Staphylococcus saprophyticus (OTU7), Brenneria (OTU9), Bartonella (OTU3), Kocuria koreensis (OTU25 and OTU39) ( Table 1). The Venn diagram (Fig. 4) indicated only 2 OTUs to be unique for T. putrescentiae from mass-rearing system (Leuconostoc gasicomitatum OTU48) and without predator (Lactococcus lactis OTU53). The bacterium similar to Staphylococcus kloosii (OTU2) made up a large percentage of the microbiota from both N. cucumeris populations and the T. putrescentiae population from the mass-rearing where N. cucumeris was present.

Discussion
Bacterial ecology of N. cucumeris and its prey mite T. putrescentiae. To our knowledge this is the first study comparing the entire bacterial microbiome of predatory phytoseiid mites and prey astigmatid mites used together under mass rearing conditions. We found that the predatory mite N. cucumeris and the prey mite T. putrescentiae shared 15% of core bacterial taxa. These taxa differ in relative abundance between predatory and prey mites, and among populations. Similarly, previous studies comparing the bacterial microbiota of M. occidentalis and its spider mite prey Tetranychus urticae (Koch) (Acari: Tetranychidae), provided evidence for a shared microbiota between the predator and the prey 22 . The dissimilarity in the bacterial community between lab reared and mass-reared populations of T. putrescentiae was higher in prey mites than in predatory mites. The prey mites showed higher diversity in their overall microbiota than predatory mites. The differences in diversity of microbiota among saprophagous compared to predatory groups are known in insects 23 and a similar situation is expected in these mites.
The OTU similar to S. kloosii accounted for a large proportion of the total bacterial sequences isolated from both N. cucumeris (72% mass-reared, 61% lab-reared) and T. putrescentiae populations (40%) the latter from  Table S1 provides the extensive identification of OTUs. The random forest analyze was calculated for predatory/prey and populations as the factors (see Table S2), separately. The forest error rates were 0.33 and 0.53 respectively, aOTU -total number of sequnces in dataset, P-values < 0.05 are indicated by bold. putrescentiae laboratory culture without the presence of predatory mite (pure) and Tyro4-9 = from N. cucumeris mass-rearing The samples are described in Table 1, the OTUs are identified in Table S1. the mass rearing where the predator was present. In the T. putrescentiae pure population, where no N. cucumeris was present, this bacterium was still present but yet not so prevalent. A similar pattern was seen for other Staphylococcus OTUs found in this study. This suggests that bacteria might be transmitted from N. cucumeris to its factitious prey, possibly via direct contact or through the feces. Alternatively, the rearing environment of T. putrescentiae is not as optimal for Staphylococcus growth as the habitat in the presence of the predator. Our suggestion is that the predatory mites can alter the rearing environment in such a way that promotes the proliferation of certain bacterial taxa. Taking into account the pattern observed here as well as the previously demonstrated effects of diet and environment on Astigmata mites microbiota 17,24,25 , it is possible that the microbiota of T. putrescentiae is more influenced by diet and environment than by the microbiota of N. cucumeris. The bacterial community of T. putrescentiae exhibited higher diversity than that of the predatory mites. Acquisition of bacteria through the prey mites' diet might explain these results. It is important to highlight that the analyses of differences in the bacterial composition (PCoA) and population level analyses showed that the differences between predatory and prey mites were lower than among the two prey mite populations examined. This is supported by differences between predator and prey in relative numbers of Blattabacterium-like and Solitalea-like symbiotic bacteria, with a higher relative abundance in prey mites. This suggests that the microbiota of different T. putrescentiae populations are more variable than the microbiota of different N. cucumeris populations.
While feeding, it is suggested that T. putrescentiae ingests debris of plant cells, fungi or yeasts which are already colonized by environmental bacteria [26][27][28] . The environmental bacteria enter the gut with the food and are passed through the gut or the bacteria adhere to the mite integument in the manner that fungal spores do 29 . Both modes of transmission resulted in the presence of environmental bacteria in the microbiota of astigmatids 17,25 . In the present study the mites were surface sterilized with sterilized phosphate saline buffer and tween 20, thus the identified bacteria species should be regarded as colonizing the mites internally. However, we cannot explicitly exclude either the possibility of some levels of surface contamination or random ingestion of environmental bacteria.
Based on previous studies with stored product mites we can identify the bacterial taxa which could be considered as environmental 17 . For example, Brevibacterium, Staphylococcus, Kocuria and Stenotrophomonas are present in mite laboratory habitats e.g. rearing diets and the feces of mites. However, some bacterial taxa isolated in this study (Table S1) are known to belong to either skin or gut communities in other animals. For example, the genus Corynebacterium includes species that are widely found in the microbiota of vertebrates 30,31 , Anaerococcus and Finegoldia spp. are commonly found among microbial skin communities in humans and can occasionally become pathogenic 32 . Bacterial taxa characteristic of mammal gut communities were also found in this study, particularly bacteria of the genera Escherichia, Lactococcus, Leuconostoc, and Propionibacterium. The latter genus contains species often found to have beneficial probiotic and nutritional effects within the gut microbiota of many vertebrate hosts such as humans and ruminants 33 . It is possible that Propionibacterium has a similar nutritional effect on its mite host, i.e prey mites. Conversely, some Propionibacterium species belong to skin microbiota and may cause disease in some cases 34 . Endosymbionts in predatory mite and prey populations. The observed microbiota of the predatory mite N. cucumeris and the prey mite T. putrescentiae consisted of intracellular symbionts as well as gut/environmental bacteria as it is reported for other species of predatory or saprophagous mites 35,36 . The presence of core bacterial taxa in prey and predatory mites poses the question whether these bacteria are autochthonous or allochthonous, i.e. horizontally transmitted from predator to prey. For example, the intracellular symbiont Wolbachia has been shown to be transmitted across trophic levels. Aedes aegypti (Diptera: Culicidae) larvae infected with the wMelPop strain did not transfer Wolbachia to predators, including copepods (Crustacea: Copepoda) and mosquito species 37 . In contrast, T. putrescentiae feeding on Wolbachia infested corpses of Drosophila melanogaster Meigen (Diptera: Drosophilidae) resulted in establishment of Wolbachia population in both mites and Drosophila and suggested horizontal transfer 38 . Generally, the presence of prey DNA is common in predators. For example, aphid nuclear and mitochondrial DNA had a detectability period longer than 23 hours in the harlequin ladybird Harmonia axyridis 39 . Likewise, the bacteria of prey mites and their DNA can be ingested by the predator mites and may be detected in the predator's digestive tract 13,40,41 . Therefore, the sole detection of bacterial DNA in predators does not necessarily mean that these bacteria form part of the predator's microbiota, e.g. Wolbachia in Phytoseiulus persimilis 42 and M. occidentalis 40 occurs due to infected prey. Presence of bacteria in the predator's eggs (and concomitant vertical transfer) would unambiguously confirm the bacterial presence in the microbiota 13,25,41 . Further studies screening the presence of symbiotic bacteria in the eggs and other life stages are needed in order to solve the question of autochthonous or allochthonous origin of symbiotic or parasitic bacteria in phytoseiid mites.
The co-occurrence of intracellular symbiotic bacteria such as Cardinium and Wolbachia in predatory and prey mites is quite common 6,14,22,43,44 . Double infections within the same individuals were common in tetranychid mites 45 . It is likely that similar occurrences are found in N. cucumeris and T. putrescentiae. It is interesting to mention the presence of the symbiotic Blattabacterium-like bacteria found in all populations of mites examined in this study. Blattabacterium-like symbionts were recently identified in some T. putrescentiae populations 25 . Bayesian analyses of the 16S rRNA gene sequences showed that Blattabacterium-like symbionts clustered as a monophyletic lineage. This cluster is outside Blattabacterium 46 , Cand. Brownia rhizoecola 47 , Cand. Uzinora diaspidicola 48 and Cand. Sulcia muelleri 49 . Blattabacterium species are obligate endosymbionts found in all cockroaches. Genome sequencing of these bacteria showed that they have a nutritional role in their host including vitamin synthesis and nitrogen recycling 50,51 . Given their similarity to Blattabacterium, the bacteria found in N. cucumeris and T. putrescentiae populations in this study may fulfil a similar role in their host mites. These Blattabacterium-like bacteria may be a unique mite specific lineage of Flavobacterium and their presence warrants further study to elicit their role in mites. Given that they were more prevalent in T. putrescentiae populations and have been previously isolated from this mite 25 , it is likely that this bacterium is found in predatory mites due to ingestion during feeding rather than an obligate interaction.
Solitalea-like bacterium was found in the reproductive tract and parenchymal tissues of A. siro L. and in five populations of T. putrescentiae 24,25 . The clones formed a new distinct cluster separate from Solitalea and other genera of the Sphingobacteriaceae family. Based on their localization in the gut, fat body, reproductive tissues and eggs of A. siro a symbiotic mode of action has been suggested 24 . Bartonella-like bacteria were previously identified in Dermatophagoides spp., A. siro and T. putrescentiae 16,17,52 . However, Bartonella-like bacteria were never amplified from eggs, eliminating the possibility of a vertically transmitted symbiont 25 . Herein, Bartonella-like bacteria were found in both predatory and prey mites, showing similar proportions to the predator-prey system of Cheyletus eruditus (Schrank) (Acarina: Cheyletidae) -A. siro 53 . Finally, Spiroplasma a commonly known bacteria from some Phytoseiidae 13 was not isolated from either mite species in this study. The original screening never isolated Spiroplasma 13 from N. cucumeris.
In this study we found significant differences in themicrobiome of T. putrescentiae from mass rearing units (with predator) and laboratory culture (without predator). The OTUs predominantly responsible for these differences were similar to S. saprophyticus, Brenneria and K. koreensis. These taxa had higher relative abundance in the bacterial microbiota of T. putrescentiae from mass rearing units (predator present) compared to laboratory cultures with no predators present. These taxa are suggested as environmental 17 and we can only speculate that the mass rearing conditions with the presence of predators are more favorable for their development, possibly due to the buildup of feces or husks of predated prey mites. Surprisingly we found Bartonella-like bacteria of higher relative abundance in prey mites from the mass rearing units compared to the laboratory cultures. The higher proportion of Bartonella-like bacteria in mass rearing culture with the presence of predators may indicate better conditions for T. putrescentiae.
Potential acaropathogens in mite populations. Not known acaropathogenic bacteria were detected in the natural microbiota of the mites examined in this study. Of special interest is the occurrence of B. cereus in all samples of predatory and prey mites. Some Bacillus taxa such as B. spahericus and B. thuringiensis (which are not distinguishable by 16S rRNA) are known to be acaropathogenic 54,55 . While B. cereus is suggested to be opportunistic on insects, B. sphaericus, B. papillae and B. thuringiensis are known pathogens 56 . B. cereus was previously isolated from the feces of a laboratory population of T. putrescentiae reared on dog kernels 27 ; the identification was confirmed by cloning and sequencing of the motB gene 57 . The addition of B. cereus to this mite diet led to a substantial reduction of population growth. Exo-enzymes of this bacterium have been proposed as aiding mite digestion 27 altogether with inhibition of the population growth of mites suggesting a sort of opportunistic mode of pathogenesis 56 . The presence of B. cereus in all mite samples, coupled with the fact that no differences were detected in the microbiota between healthy and sick populations of predatory mites, suggests that symptoms of illness of the predatory mites in the laboratory colony were not caused by bacteria or that the method we used could not detect the acaropathogenic bacteria. Often, some pathogens make up part of the natural microflora of its host and opportunistically cause disease due to subtle changes in environmental factors, host immune system and even microbiota 58 . Presence of disease symptoms in the lab-reared N. cucumeris could not be explained by the presence of harmful bacteria. Instead, perhaps it was the result of the lack of beneficial protective microbes as seen in other animals 58 . Alternatively, the disease symptoms in the laboratory population of the predatory mites might have been caused by viruses, fungi or protozoan pathogens 8 or accumulation of toxic metabolites such as guanine 10 . Conclusion. The microbiota of the prey mite T. putrescentiae and the predatory mite N. cucumeris consisted of core bacterial taxa present in all prey and predatory mite populations.This core microbiota comprised Wolbachia, Cardinium, Bartonella-like, Blattabacterium-like, Solitalea-like, Brevibacterium, Staphylococcus spp. and B. cereus. Among them Brevibacterium, Staphylococcus and Bacillus were the most abundant in predatory mites, while Blattabacterium-like and Solitalea-like bacteria were the most abundant in prey mites. Significant differences were detected between the bacterial communities of prey mites without predators and prey mites reared with N. cucumeris. S. saprophyticus, Brenneria and K. koreensis were more abundant in the presence of predators. The occurrence of no acaropathogenic bacteria was examined in predatory mite populations with and without disease symptoms. Interestingly, bacterial microbiome of healthy and with disease symptoms predatory mites showed similar diversity, without significant differences between the two groups. This suggests disease symptoms in this case are caused by non-bacterial entities such as other types of microbes, unsuitable environmental conditions or genetic factors. Further study is required to confirm the cause of disease. Ultimately, characterization of predatory mite and prey microbiota may help inform mass rearing practices, but this understanding also required knowledge on the effects of certain microbial taxa on the health of both species of mites. This study can serve to influence the subsequent study of mechanistic studies of the effects of some bacterial taxa on mite hosts of economic importance.

Materials and Methods
Origin of mites. The predatory N. cucumeris and prey T. putrescentiae mites originated from the Biobest rearing facilities in Belgium (Westerlo). T. putrescentiae was reared on a mix of yeast flakes, wheat germ and dried yeasts at a ratio 10:10:1 (w/w) 59 in rearing units of 0.2 L kept at 27 °C and 85% relative humidity (r.h.). N. cucumeris was reared on the aforementioned mix plus T. putrescentiae (treatment 4 see below) at a ratio of approximately (1:15; N. cucumeris: T. putrescentiae) in rearing buckets of 5L (mass rearing) or 0.2 L (laboratory population) kept at 25 °C and 70% r.h. The trial included the following treatments: (i) N. cucumeris from a mass-production population (ii) N. cucumeris from a laboratory population with disease symptoms (slower movement and lighter coloration of the mites); (iii) pure T. putrescentiae (with no predators); (iv) T. putrescentiae from rearing units with predatory mites. Three samples were examined for treatments i-iii (see below for sample definition), and six samples for treatment iv (Table S2). Predatory and prey mites from treatment 1 were sampled one week after the addition of the prey mites.
DNA extraction from mites. One sample consisted of about 50 (predator or prey) mites collected in ethanol (90%). Ethanol was removed and samples were washed 3 times in Phosphate Buffered Saline and Tween-20 (PBST). The mites were homogenized in 500 μL of PBST using a Radnoti tissue grinder (Cat No. 440613, Monrovia, CA, USA). Total DNA was extracted using the Wizard ® Genomic DNA Purification kit (Promega, Madison, WI, USA) following the manufacturer's instructions. The extracted DNA was stored at −20 °C until further analyses.
Amplification, cloning and sequencing. The quality and presence of bacterial DNA in every sample was tested by PCR amplification using eubacterial primers and routinely using protocols 60 . When amplicons were not obtained, the samples were replaced by new samples positive for amplicons. The DNA samples were sent to MR DNA (http://mrzaqqqsadnalab.com, Shallowater, TX, USA). The sequencing of V1-V3 part of 16S rRNA gene was based by the universal primers 27Fmod and 519Rmod in the Illumina MiSeq platform and the bTEFAP ® process 61 . The length of the read was 300 bp, the reads were forward and reverse. The sequencing reads were preprocessed in MR DNA to contingencies. The sequences in bioproject PRJNA321085 were deposited in GenBank SRP074673, barcodes and biosample codes are given in Table S2.
Data analyses and sequence processing. The contingencies were demultiplexed using MR DNA binning software (http://mrdnalab.com), sequences were renamed and trimmed in MOTHUR v. 1.36.1 software 62 according to the MiSeq standard operation procedure MiSeq SOP 63 . The actual commands used in MOTHUR are available at http://www.mothur.org/wiki/MiSeq_SOP (accession date 2/19/2016). The demultiplexed, renamed and trimmed sequences were processed in UPARSE and USEARCH 64 , singletons were removed and the sequences were classified using a naive Bayesian classifier with a training set (version 15) made available through the Ribosomal Database Project (RDP) (http://rdp.cme.msu.edu) 65 . Operational taxonomic units (OTUs) were defined by the clustering of sequences at ≥97% identity. The representative sequences obtained from UPARSE were processed in MOTHUR again, aligned against SILVA reference database 66 . The aligned sequences were screened for chimeras with UCHIME 67 . Sequences from chloroplasts, mitochondria, Archaea and Eukaryotes were removed. The representative sequences were analyzed via BLASTin GenBank 68,69 . Solitalea-like, Blattabacterium-like and Cardinium bacteria were identified by aligning OTUs to previously identified almost full length 16S rRNA sequences of these taxa 17,24,25 in Codone Code Aligner (CodonCode Corporation, Centerville, MA, USA). Taxonomic diversity and relative proportions of bacterial taxa were visualized using Krona tools 70 .
Statistical analyses. The shared file was generated from UPARSE data and was processed in MOTHUR, and PAST 3.06 71 . The results were visualized by XLSTAT (http://www.xlstat.com/en/, Addinsoft, New York, NY, USA). The subsample data set was constructed in MOTHUR on 17,613 sequences. Alpha-diversity, (OTUs richness) in predatory or prey mites was assessed using the Inverse-Simpson index, the number of OTUs was calculated in MOTHUR from a subsample data set. The Inverse-Simpson index and species level OTUs (sOTU) of the predator and prey samples were compared using a nonparametric Mann-Whitney test. Beta-diversity, (similarity of samples) was assessed using the Bray-Curtis and Jacquard indices and visualized by principal coordinate analyses (PCoA) of the subsample data set. The contribution of OTUs was calculated using Pearson correlation coefficient. We analyzed the effects of two factors on bacterial diversity: (i) mite species (i.e. prey and predator) and (ii) origin of mite population (i.e predator with and without sick symptoms and prey from mass rearing facility and without any predator) ( Table S2). The effects of both factors were evaluated with two-way PERMANOVA 72 with 100,000 permutations. Calculations were based on Bray-Curtis and Jaccard matrices of the subsample data. The different populations of predatory mite (sick and without disease symptoms) and T. putrescentiae (from mass production and without predatory mites) were also compared separately. Venn diagrams were used to highlight shared bacterial taxa in mites from different treatments, to the analyses we included only those OTUs, which were presented in all three replicates per treatment. The population-level analysis was calculated using METASTATS 73 based on 100,000 permutations to compare the effects predator/prey to distribution of bacterial taxa in the subsample data set. The random forest analysis in MOTHUR was applied to compare the differences in bacterial microbiome of sick and healthy predator or T. putrescentiae from mass rearing and laboratory culture.