A continuous rearing of I. typographus was established in the laboratory in the summer of 2021. The initial population was started with adult beetles captured in a state-owned forest near Jena, Germany (50°54'28.3"N 11°39'28.9"E). A healthy Norway spruce tree (Picea abies) was felled monthly from the aforementioned location and cut into 25 cm long logs. The logs' cut ends were coated in paraffin to prevent desiccation and placed individually in plastic boxes with a metallic mesh on the lid. Each log was exposed to 40 adult beetles and kept in a walk-in rearing chamber (Viessmann) with 16 h of light and 8 h of darkness at a temperature of 25°C and relative humidity of 60%. The colony has now been kept for over 14 generations.

Three larvae and three teneral adults from a wild collection, plus four mature adults from the laboratory colony were anesthetized for 1 min at −20°C, washed individually in 0.1% SDS sterile solution inside of a microcentrifuge tube, and shaken for 30 s. Then the individuals were transferred to another microcentrifuge tube with 70% ethanol and were shaken for 30 s. Subsequently, each specimen was washed thrice in sterile 1x PBS. The gut was dissected and homogenized with a sterile micropestle in 500 μL of 1 x PBS. The sample was vortexed for 10 s and then serially diluted up to 1 × 10⁻⁴. Each dilution was plated twice (100 μL) on lysogeny broth (LB) agar and incubated at 25°C and 28°C, respectively, for 48 to 96 h. Individual colonies were picked and streaked out three subsequent times to obtain pure cultures.

To isolate microorganisms from female oral secretions, five female beetles were collected during egg-laying from the laboratory rearing. They were gently rinsed with sterile distilled water and allowed to walk in a clean Petri dish until they were dry. A 50 μL drop of sterile 1x PBS was placed in the middle of an LB agar plate. The maternal oral secretions were collected by holding the female with sterile forceps and placing its mouthparts in contact with the PBS drop in the agar plate for 10 s−15 s. The PBS drop was plated and incubated at 28°C for 48 h to 96 h. Individual colonies were picked and streaked out three subsequent times to obtain pure cultures. These pure cultures were grown in liquid LB medium at 28°C and 300 rpm for 12 h. The resulting liquid cultures were used to create glycerol stocks that were kept at −80°C for long-term storage.

Liquid cultures of all the isolates were prepared by dipping an autoclaved toothpick in a single colony and inoculating 7 mL of LB liquid medium in a 15 mL Falcon tube with a vented lid. The cultures were incubated for 24 h at 28°C and 250 rpm. Then, 1 ml of the culture was transferred to a 1.5 ml microcentrifuge tube and centrifuged at 16,000 g for 1 min. The supernatant was discarded and the pellet was re-suspended in 1x PBS by vortexing for 10 s. The sample was centrifuged again and the supernatant discarded. The washing step in PBS was repeated twice and the pellet was stored at −80°C until DNA was extracted. The MasterPure^TM complete DNA and RNA isolation Kit (Epicenter Technologies) was used following the manufacturer's protocol. The extracted DNA was used for isolate identification using the primers fD1 (5′-AGAGTTTGATCCTGGCTCAG-3′) and rP2 (5′-ACGGCTACCTTGTTACGACTT-3′) targeting the bacterial 16S rRNA gene (Weisburg et al., 1991). For amplification, 1 μl of DNA was used in a 25 μl PCR reaction with the Taq PCR Master Mix Kit (Qiagen) under the following cycling conditions: 94°C for 3 min, 35 cycles of 94°C for 60 s, 57°C for 60 s, 72°C for 60 s, and a final extension at 72°C for 10 min. The PCR products were cleaned with the DNA Clean and Concentrator-5 kit (Zymo Research) following the manufacturer's protocol and sequenced bidirectionally on a 3730XL DNA Analyser (Applied Biosystems from Thermo Fisher Scientific). Reads with minimum 700 bp and quality > 70% were used. The resulting forward and reverse sequences were trimmed by removing parts with >5% chance of error base and aligned using Geneious Prime version 2022.1 to create a consensus sequence. These consensus sequences were blasted against the NCBI database and the taxonomy was assigned to the species level when possible (>99% identity match) or to the genus level. Later, the sequences were also blasted against the 16S rRNA amplicon sequencing dataset generated in this study to confirm their presence in other sampled individuals. Isolates with no matches within the culture-independent microbiota profiling analysis were considered contaminants and removed from the isolate collection.

To characterize the microbiota of field-collected beetles, insects were collected between June and July 2021 near Jena, Germany (50°54'28.3"N 11°39'28.9"E). Mature adults and larvae were collected from five standing spruce trees. When possible, more than one gallery was sampled from each tree, ensuring that at least three individuals per life stage per gallery were present. The samples from each gallery were labeled and kept separately for future handling. In addition, 2 cm × 2 cm squares of bark were sampled from the galleries and from adjacent unattacked bark. Each sample was stored in a 15 ml Falcon tube and placed inside a cooler with ice packs to transport them to the lab.

For the microbiota profiling of laboratory-reared samples, the gallery environment (galleries, unattacked bark, pupal chambers) and beetles (mature adults, larvae, pupae, teneral adults) were sampled from generations 12 to 14 after their introduction to the laboratory colony between January and April 2022. As in the case of wild-collected samples, individuals from at least two galleries were sampled per generation and were identified accordingly for the subsequent pooling steps. A detailed overview of the sampling and replication is provided in the Supplementary Table 1. Additionally, three logs from the F14 generation were opened 5 days after the initial beetle colonization to sample the eggs and the tightly packed bark that covered the oviposition site (henceforth referred to as “the egg plug”).

The bark pieces, eggs, and egg plugs were flash frozen in liquid nitrogen and stored at −80°C until DNA extraction. The larvae and adults were anesthetized for 1 min at −20°C, washed individually in 0.1% SDS sterile solution inside of a microcentrifuge tube, and shaken for 30 s. They were then transferred to another microcentrifuge tube with 70% ethanol and again shaken for 30 s. Subsequently, each specimen was washed thrice in sterile 1x PBS and dissected as described elsewhere (Ceja-Navarro et al., 2012). The entire gut was transferred to a microcentrifuge tube, flash frozen in liquid nitrogen and stored at −80°C until DNA was extracted. The pupae underwent the same washing steps but the specimens were not dissected, as the morphology at this life stage makes it difficult to dissect the gut. Individuals carrying mites, parasitoid larvae or visible nematodes were omitted from the study.

The insect samples were homogenized in 1.5 ml microcentrifuge tubes with liquid nitrogen and sterile micro pestles. For the larvae, teneral adults and mature adults, three guts of individuals belonging to the same gallery were pooled per tube to create one biological replicate. The pupae were processed individually (a detailed overview of the sampling and replication is provided in the Supplementary Table 1). The gallery and bark samples were ground in liquid nitrogen in autoclaved mortars and pestles. A volume of ~50 μl of homogenized bark per sample was transferred to a microcentrifuge tube. For the pupal chamber samples, a volume of ~50 μl per sample was homogenized in 1.5 ml microcentrifuge tubes with liquid nitrogen and sterile micro pestles. The MasterPure^TM complete DNA and RNA isolation Kit (Epicenter Technologies) was used following the manufacturer's protocol, with an additional incubation step with 4 μL lysozyme (100 mg/mL) at 37°C before protein digestion. A positive control (ZymoBIOMICS™ Microbial Community Standard, Zymo Research, USA) and negative controls (extraction reagents without sample tissue) were included throughout the extraction steps. The DNA was re-suspended in Low TE buffer (1:10 dilution of TE) and the concentration was measured on a Qubit fluorometer using a 1x dsDNA High sensitivity assay (Thermo Fisher Scientific).

Altogether, 55 beetle gut samples (at least 6 per life stage), 14 pupae, 3 egg samples, 3 egg plug samples, 3 pupal chamber samples, 13 bark samples, 9 gallery samples, one mock community as positive control and 4 negative extraction controls were sequenced. Bacterial 16S rRNA gene regions and fungal internal transcribed spacer 1 regions (ITS1) were sequenced by a commercial provider (StarSeq, Mainz, Germany) on a MiSeq platform (Illumina) using double indexing and a paired end approach with a read length of 300 nucleotides. The primers 341F (5′-CCTACGGGNGGCWGCAG-3′) and 806R (5′-GACTACNVGGGTWTCTAATCC-3′) were used to sequence the V3-V4 regions of the bacterial 16Ss rRNA (Klindworth et al., 2013). The primers ITS1F (5'-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2 (5′-GCTGCGTTCTTCATCGATGC-3′) (White et al., 1990; Gardes and Bruns, 1993) were used to sequence the fungal ITS1 region.

The reads were demultiplexed onboard in MiSeq Reporter software, allowing for one mismatch. The demultiplexed reads were processed following the DADA2 pipeline (version 1.28.0) (Callahan et al., 2016). For the bacterial sequences, the last 30 base pairs of all forward reads and the last 50 base pairs of the reverse reads were trimmed to remove nucleotides with low quality scores. For the fungal sequences, Cutadapt (version 4.0) (Martin, 2011) was used to remove the primer sequences and then a minimum sequence length of 50 base pairs was enforced to remove spurious short sequences. This different choice in the trimming approach was due to the large variability in the sequence lengths obtained in the fungal dataset (Schoch et al., 2014). For both datasets, the maximum expected error was set to 2 and the minimal overlap for merging the sequences was set to 12 nucleotides. Chimera removal was done with the “consensus” method. The resulting bacterial sequences were further filtered to remove any consensus sequence of length below 300 bp, as shorter reads were likely resulting from sequencing artifacts. The databases used for taxonomy assignment were SILVA trainset v138.1 for bacteria (Quast et al., 2013) and the UNITE general FASTA release for Fungi 2 version 8.3 (Abarenkov et al., 2021). The identities of the most abundant taxa were verified by blasting the sequence against the NCBI Nucleotide database (National Center for Biotechnology Information, 1988).

The “phyloseq” R package (version 1.44.0) (McMurdie and Holmes, 2013) was used to analyse the amplicon sequence variant (ASV) tables. Single reads and taxa that could not be classified as “bacteria” (e.g., Chloroplasts) or “fungi” were removed from the dataset. The data were not rarefied to improve the detection of differentially abundant species (McMurdie and Holmes, 2014). The alpha diversity was estimated with the Shannon and Simpson indices, and the beta diversity among groups was visualized with principal component analysis (PCoA) of the Bray-Curtis dissimilarity index. Significant differences in alpha diversity between samples were assessed with the Wilcoxon or Kruskal–Wallis rank-sum tests. Dunn's test was used to calculate pairwise comparisons between the sample types (population of origin, tissue, insect life stage) using Bonferroni's correction for multiple sample comparison. A permutational multivariate analysis of variance (PERMANOVA) was performed to evaluate the differences in beta diversity among samples. The absolute abundance data were transformed to relative counts to create the ordination plots. Core taxa shared by different life stages were identified and visualized with the R package “microbiome” (Lahti and Sudarshan, 2012). The core sequences were defined as ASVs that were present in at least 50% of the insect samples with a relative abundance of more than 0.1% per sample, respectively.

The DNA obtained for the microbial community analysis was also used to estimate bacterial and fungal titers across different beetle life stages. Additional DNA extracts from eggs and egg plugs were included for this analysis: the eggs and their respective egg plugs were sampled from five galleries from two logs of the laboratory colony as described before. The eggs of each gallery were pooled together in a 1.5 ml microcentrifuge tube to create a biological replicate (~20 eggs). Their matching plugs were pooled in a separate tube. Five additional galleries were sampled to create pooled samples containing both the eggs and the plugs. DNA was extracted and measured following the same kits and protocols as the previous samples.

Quantitative PCR (qPCR) was carried out in 20 μl reactions using Blue S'Green qPCR mix (Biozym), 1 μl template DNA and 0.4 μM of each primer. The bacterial 16S primers EUB338mod (5′- TCCTACGGGAGGCAGCAG-3′) and EUB518 (5′- ATTACCGCGGCTGCTGG-3′) and fungal 18S primers FR1(5′- AICCATTCAATCGGTAIT-3′) and FF390 (5′-CGATAACGAACGAGACCT-3′) were used (Fierer et al., 2005; Chemidlin Prévost-Bouré et al., 2011).

Standard curves with defined copy numbers of the 16S rRNA gene were created by amplifying the fragment first, followed by purification and determination of the DNA concentration via NanoDrop1000 (Peqlab, Germany). After determination of the DNA concentration, eight serial 1:10 dilutions were prepared to generate the standards. One μL of each dilution was included in the qPCR reaction to standardize the measurements across reactions. The number of copies of DNA in the standard dilutions was calculated using the formula N=DNA amount (ng) × 6.022 × 1023 (numbermole)fragment length (bp)× 1×109 (ngg)×330gmole . The DNA copy numbers in the samples were estimated taking into account the efficiency of each reaction, using the formula: X0=10(Cq−b)·m−1where X is the estimated DNA copy numbers, Cq is the quantification cycle of the qPCR, b is the intercept of the reaction's standard curve and m is the slope.

Modified phloem sandwiches (Bedard, 1933) were used to study maternal gallery construction and egg laying behavior. In short, a 15 × 8 cm strip of phloem was peeled from a freshly cut P. abies log and placed between two 18 × 10 cm Plexiglas panels. The panels were sealed together with electrical insulation tape and secured with four binder clips. A male bark beetle was placed in an opening in the top panel and covered with a piece of plastic mesh. After 24 h, a female beetle was added if the male had initiated a mating chamber. Twelve successfully mated pairs were observed with an EOS 600D (Canon) camera mounted on a Stemi 2000-C microscope (Zeiss). The different life stages were photographed and egg laying behavior was recorded whenever possible.

Significant differences in alpha diversity between samples were assessed with the Wilcoxon or Kruskal–Wallis rank-sum tests using the “biostat” package (Gegzna, 2020). Dunn's test was used to calculate pairwise comparisons between the sample types (population of origin, tissue, insect life stage) using Bonferroni's correction for multiple sample comparison. A PERMANOVA was performed to evaluate the differences in beta diversity among samples using the R packages “vegan” and “pairwiseAdonis” (Martinez Arbizu, 2017; Oksanen et al., 2022). For the microbial titers, the differences in estimated DNA copy numbers normalized by individual were assessed with a negative binomial model to correct for the data's over dispersion using the package “MASS” (Venables and Ripley, 2002). A least-squares means test (LS means) was used to assess the significant differences among sample types using the “emmeans” package (Lenth, 2023). All analyses were carried out in R version 4.3.0 (R Core Team, 2023). Figures were created with “ggplot2” (Wickham, 2009) and illustrations with BioRender.com (2023).

Bacterial isolates were obtained from the guts of larvae, teneral adults and mature adults, and from maternal oral secretions and cultivated on LB medium. These predominantly belonged to the class Gammaproteobacteria (Table 1). Erwinia typographi was the most frequently isolated bacterium, Rahnella species were isolated only from teneral adults, and Pseudoxanthomonas isolates were obtained from mature adults and maternal oral secretions. Bacilli were the second most abundant group of bacteria, with Paenibacillus being the dominant genus, followed by Lactococcus. Most of the Paenibacillus isolates were obtained from teneral adults. Bacteria belonging to Actinomycetia, Alphaproteobacteria, Sphigobacteriia, and Streptomycetales were also present to a lesser extent in mature adults, larvae, and female oral secretions.

In the culture-independent high-throughput profiling, Gammaproteobacteria largely dominated the bacterial communities associated with the beetle gut across all the life stages, as well as the eggs and the beetle-modified environment. The most abundant taxon was E. typographi, followed by Pseudoxanthomonas spadix and Rahnella sp. All three taxa were present in both wild-collected and laboratory-reared specimens, but the wild-collected samples had a higher prevalence of Rahnella sp., while P. spadix was more common in the laboratory population. Additionally, a further Erwinia ASV (not identifiable to the species level) was abundant in most of the beetle and the beetle-modified environment samples. Even though these Gammaproteobacteria were also present in the unattacked bark samples, they were not as abundant as in the beetle guts, eggs, and gallery environment (Figure 1A). E. typographi, P. spadix, Rahnella sp. and Erwinia sp. formed the core gut bacterial community shared by all life stages across the wild-collected and laboratory-reared samples (Figure 2A). Serratia sp., Pseudomonas bohemica, Ochrobactrum sp., and an unidentified ASV belonging to Rhizobiaceae were present to a lesser extent in the galleries, larvae, pupae, teneral adults and mature adults of both populations. An unidentified Chitinophagaceae ASV was present in the galleries and some of the adults. The mollicute Spiroplasma sp. was identified in variable abundances in all the mature adults captured in the wild, but it was almost absent from the laboratory-reared individuals. Overall, the composition and structure of the bacterial communities associated with I. typographus and the gallery environment were largely different from the ones in the unattacked bark. The unattacked bark harbored taxa belonging mostly to Alphaproteobacteria, Gammaproteobacteria and Verrucomicrobiota (Supplementary Figure 2A).

Saccharomycetales yeasts largely dominated the fungal communities across all beetle life stages, as well as the gallery environment (Figure 1B). Wickerhamomyces bisporus was ubiquitous, followed in abundance by Kuraishia molischiana, Ogataea ramenticola and Cyberlindnera amylophila. These four yeasts formed the core gut community of all life stages across the wild-collected and laboratory-reared samples (Figure 2B). Nakazawaea ernobii and Saccharomycopsis lassenensis were also widespread among different life stages but their relative abundances were lower when compared to the core yeasts. The second most abundant order of fungi were the Sordariomycetes. Interestingly, the abundance of Endoconidiophora polonica, one of the best known fungal ectosymbionts of I. typographus, was higher in the samples from the wild population than in the laboratory-reared beetles and galleries. Another ASV, assigned to Grosmannia aenigmatica, was abundant in some of the larvae, pupae, pupal chambers and teneral adults of the laboratory-reared samples. A search of this ASV against the NCBI database revealed multiple 100% sequence identity hits with Leptiographium piceaperdum (the anamorph of G. piceaperda). A third sordariomycete fungus, Rhexographium fimbriasporum, was present mostly in the laboratory-reared insects. Additionally, Sistotrema brinkmannii and Cylindrobasidium torrendi (Basidiomycetes) were very abundant in the wild-caught teneral adults. An unidentified Basidiomycete appeared in several gallery and insect samples from the wild and, in lower abundances, in the laboratory-reared samples. As observed in the bacterial communities, the composition and structure of the fungal communities associated with the unattacked bark were remarkably distinct from the ones present in the beetle guts, eggs, and gallery environment. The unattacked phloem harbored taxa belonging mostly to Sordariomycetes, Dothideomycetes, Leotiomycetes and Lecanoromycetes (Supplementary Figure 2B).

The 16S rRNA dataset had an average of 14107 reads with a minimum of 6 and a maximum of 48820 across samples. The ITS dataset had an average of 48030 reads with a minimum of 12 and a maximum of 334941 across samples. The bacterial alpha diversity of the unattacked bark was significantly higher than that of all other samples (Kruskal–Wallis test: Shannon χ² = 40.03, p < 0.0001; Simpson χ² = 38.196, p < 0.0001; Supplementary Figure 1B) and did not differ significantly between wild-collected and laboratory-reared samples (Wilcoxon test: Shannon p = 0.093, Simpson p = 0.079; Supplementary Figure 1A). Similar to the case of bacteria, fungal alpha diversity was also higher in the bark and in the galleries than in the beetle guts, eggs, pupal chambers, and bark plugs placed over the eggs immediately after oviposition (Kruskal–Wallis test: Shannon χ² = 48.673, p < 0.0001; Simpson χ² = 37.234, p < 0.0001; Supplementary Figure 1D). In contrast, the wild-collected samples showed a significantly higher fungal alpha diversity than the ones from the laboratory colony (Wilcoxon test: Shannon index p = 0.002, Simpson index p = 0.02, Supplementary Figure 1C).

The beta diversity analysis revealed significant differences in the microbial community composition across beetle life stages. The principal component analysis (PCoA) of both the bacterial and fungal Bray-Curtis distances showed the unattacked bark clustering separately from the rest of the samples (Figures 3A, C), and the PERMANOVA generated significant results for the bacterial (F = 3.215, p = 0.001) and fungal (F = 4.227, p = 0.001) Bray-Curtis metrics. Further inspection of the data revealed a clear separation of the bark samples from the beetle-modified environment (galleries, egg plugs, and pupal chambers) (Figures 3B, D, Supplementary Tables 2, 3). The beta diversity analysis across life stages did not reveal significant differences in community composition except for the pupae, both in the bacterial (F = 2.636, p = 0.001) and the fungal communities (F = 2.902, p = 0.001) (Supplementary Figures 5A, B, Supplementary Tables 4, 5). PERMANOVA analysis showed significant differences between the bacterial (F = 6.1837, p = 0.001) and fungal (F = 12.100, p = 0.001) communities of the samples from the wild collection and the laboratory rearing, but no clear patterns were observed in the PCoA (Supplementary Figures 5C, D).

The bacterial loads in the gut did not differ significantly across life stages except for the pupae, where the entire individual including the gut was sampled. The copy number of the 16S gene was higher for the pupae than the mean copy number for the first instar larvae, teneral and mature adults (Negative Binomial GLM p = 1.98 × 10⁻¹¹, LS means p < 0.0001, Figure 4A). Due to the pooling of egg samples, it is difficult to assess the microbial load on individual eggs accurately. Assuming an average of 20 eggs per sampled gallery, the bacterial load of an individual egg was lower than that in the gut of the other life stages, which was not statistically significant except for the teneral adults (Negative Binomial GLM p = 0.00718, LS means p < 0.0081). When compared to their matching plugs, the eggs themselves had higher numbers of 16S rRNA gene copies (Negative Binomial GLM p = 0.0046, LS means p < 0.0127, Supplementary Figure 4A). The bacterial titers in the additional samples where the eggs and plugs were collected together were not statistically different from the egg-only samples, suggesting that the contribution of the plugs to the bacteria in the oviposition site is lower than the contribution from the bacteria present on the egg surface (Negative Binomial GLM p = 0.0.094, LS means p < 0.215). The 16S rRNA gene copy numbers were an order of magnitude higher in the galleries than in the unattacked phloem (Negative Binomial GLM p = 0.00104, LS means p < 0.003, Supplementary Figure 3A). The same trend was observed for the pupal chambers, but was not statistically significant (Negative Binomial GLM p = 0.05507, LS means p < 0.1335).

The fungal load in the gut remained stable as the life stages progressed (Figure 4B). As in the case of bacteria, the pupae were the exception, as they had higher titers than the rest of the life stages (Negative Binomial GLM p = 5.1 × 10⁻⁸, LS means p < 0.0001). As opposed to bacteria, the egg plugs carried significantly higher ITS gene copy numbers than the eggs (Negative Binomial GLM p = 2.99 × 10⁻⁹, LS means p < 0.0001, Supplementary Figure 4B). The fungal titers in the samples where eggs and plugs were collected together were also higher than in the egg-only samples (Negative Binomial GLM p = 2.75 × 10⁻¹⁰, LS means p < 0.0001), which indicates that the plugs make a major contribution to the presence of fungi in the oviposition site. As for the bacteria, the fungal load was an order of magnitude higher in the galleries than in the unattacked phloem (Negative Binomial GLM p = 0.00289, LS means p < 0.0081, Supplementary Figure 3B). This was also the case for the pupal chambers, which had a significantly higher number of ITS gene copies than the unattacked phloem (Negative Binomial GLM p = 0.00594, LS means p < 0.0164) and did not differ in fungal titers from the galleries (Negative Binomial GLM p = 0.00594, LS means p < 0.9260).

To explore the likelihood of vertical transmission of microbes from parents to offspring, we first used phloem sandwiches to observe parental behavior, offspring development, and feeding habits of the different life stages (Figure 5). The gallery excavation and developmental stages have been extensively documented in previous works [as summarized by Schebeck et al. (2023)]. Earlier literature indicates that the parent beetles “groom” the eggs, but a detailed description of this behavior is missing. We observed that, while boring the maternal gallery, the female removed most of the bark debris by pushing it toward the mating chamber with its elytral declivity. However, it stored part of the chewed bark at the farthest end of the tunnel. Every 2–3 millimeters, the female stopped boring the main gallery and carved a niche with its mouthparts on one of the sides. Once the niche was large enough to harbor an egg, the mother traveled back to the mating chamber to make a 180° turn. With the abdomen facing the end of the maternal gallery, it laid an egg in the niche. It then rapidly moved back to the mating chamber to turn around, and returned to the freshly laid egg at a considerable speed. Using the chewed bark stored at the end of the gallery, the mother quickly created a bark plug to cover the egg (Supplementary Video 1, Figures 5A, B).