Introduction

Insects can derive diverse benefits from their intestinal microbiota. The best example is the termite, whose digestive track is specifically adapted for colonization by various microbes essential to the digestion of wood biomass [1]. The microbial community of the termite intestinal microbiota consists largely of Proteobacteria, Spirochetes, Bacteroidetes and Gram-positive bacteria with a low G+C content [2]. In contrast, bacterial symbionts are apparently not essential for digestion of the plant cell wall in many other insect species that secrete digestive enzymes encoded in their own genomes [3]. However, there are other benefits associated with midgut bacteria, including enhanced resistance to pathogens [4], promotion of intestinal epithelium development [5], neutralization of toxins [6] and production of chemical communication molecules [7]. Lepidopteran insects feed mostly on plants and possess a very simple digestive system devoid of any known adaptation for the maintenance of microbes. The majority of bacteria composing their microbiota are acquired from their food plants [8], making diet an important factor in microbiota composition. The microbiota has no defined role in the Lepidoptera, but it is often dominated by Enterobacteriaceae, which could play a role in carbohydrate metabolism [9].

Each year in Canada, defoliating insects are responsible for important timber volume losses for the forest industry. Controlling these pests is a significant challenge and scientists are constantly looking for novel strategies that are more cost-efficient and/or eco-friendly. As chemical pesticides are no longer looked upon favorably for use in forest pest management, molecular and biological targets that can be perturbed in a pest-specific manner have become the focus of many current research efforts aimed at developing alternative pest-control tools. The intestinal flora of these insects could be one such target, particularly where it plays a significant role in host health.

It is in this context that we undertook the present work on the midgut microbiota of the spruce budworm, Choristoneura fumiferana, a major pest of spruce and fir in Canada. Our primary objective was to conduct a survey of this insect’s gut microflora using a next-generation sequencing (NGS) approach. A pioneering study that examined the intestinal microbiota of C. fumiferana indicated a dominance of Enterococcus and Staphylococcus bacteria among microorganisms that could be cultured from its midgut, and showed that antibiotics present in a synthetic diet [10] were not potent enough to eliminate gut bacteria at the concentrations used [11]. Another study also showed that intestinal juices from larvae fed balsam fir foliage could inhibit bacterial growth [12], suggesting that the host plant could have a negative impact on the budworm midgut microflora. For this reason, the second objective of our study was to conduct a preliminary assessment of the effect of rearing conditions, including diet (synthetic, black spruce, balsam fir) and insect sources (laboratory colony versus field-collected) on budworm midgut microbiota composition. Our results point to a significant effect of food and insect source on the taxonomic diversity and structure of the budworm midgut microbiota.

Materials and Methods

Insects were collected by government scientists in permanent research plots located on crown land; no permission was therefore required. The sampling was considered non-destructive and did not involve protected or endangered species.

Results

Variation in richness among microbiotas (α-diversity)

For each of the three treatment groups, the OTU rarefaction curves began to plateau by the time all reads had been analyzed, indicating that the sequencing depth was sufficient to capture most of the biodiversity found in the larval midguts (Fig 1). Interestingly, far more OTUs were detected in larvae reared on a synthetic diet containing Aureomycin (>1000) than in larvae collected in the field on either balsam fir or black spruce (<600).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. OTU rarefaction curves for each experimental group.

Each group represents the aggregate of 12 individuals, at 2,460 reads each (total 29,520 reads per group). Note that the balsam fir and black spruce curves have overlapping 95% confidence intervals (i.e., not significantly different).

https://doi.org/10.1371/journal.pone.0144077.g001

Overall composition of spruce budworm midgut microbiota

Overall, the budworm midgut microbiota was composed primarily of Proteobacteria (Fig 2). The dominant genus was Pseudomonas, which constituted the majority of the Proteobacteria (Fig 3), whereas Bradyrhizobium was essentially specific to larvae reared on synthetic diet (915 reads vs 30–62 reads for each of the other two diet groups). An examination of the top 10 OTUs within the three diet groups (Table 1) clearly showed that the genus Pseudomonas was indeed the most abundant taxon in the budworm midgut microbiota, as the majority of dominant OTUs belonged to this genus. These Pseudomonas OTUs could not be assigned to species using the NCBI 16S database, since each typically had several hits with an E-value of zero and an identity of 100%. This being said, species such as P. fluorescens and P. paea were often among these hits. Table 1 also shows that the communities were highly skewed with respect to OTU distribution, with the 10 top OTUs representing a very large proportion of the total reads obtained (~60–80%), indicating that the remaining OTUs represented low-abundance, less frequent taxonomic groups. Surprisingly, the two taxa–Staphylococcus and Enterococcus–that were identified in an earlier study [11] as the most abundant in the budworm midgut, using a bacterial culture approach, were among the lower abundance OTUs in this study, represented by 249 and 4 reads, respectively, among the 88,560 reads of the dataset. At the phylum level, the microbiotas of larvae fed either spruce or fir were very similar, except for a higher proportion of Actinobacteria (majority Propionibacterium) in larvae collected and reared on balsam fir (Fig 2; p < 0.01, Mann-Whitney). In larvae reared on artificial diet, the proportion of Bacteroidetes was much higher than in larvae collected and reared on the other two food types (p < 0.01, Kruskal-Wallis). More than 97% of these Bacteroidetes were assigned to the family Chitinophagacaea, but we could not assign them to specific genera using the GreenGenes database. However, some of the most abundant OTUs found in larvae fed artificial diet belonged to this family and were identified as Hydrotalea flava or Sediminibacterium gingensoli using the NCBI 16S database (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Taxonomic distribution of reads at the phylum level for each experimental group.

https://doi.org/10.1371/journal.pone.0144077.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Taxonomic distribution of reads assigned to the Proteobacteria (taxonomic breakdown within purple bars of Fig 2) for each experimental group.

Only genera comprising more than ~1% of the phylum are shown.

https://doi.org/10.1371/journal.pone.0144077.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. List of top 10 OTUs within each experimental group, along with the names of the species or genera they most likely represent (from BLAST analysis).

https://doi.org/10.1371/journal.pone.0144077.t001

Compositional variation among microbiotas (β-diversity)

The Bray-Curtis analysis, which is based on the presence/absence and relative abundance of OTUs in each sample, showed a strong clustering of individual larval microbiotas according to rearing conditions (Fig 4). Interestingly, the microbiotas of larvae reared on the two host trees clearly formed two compositional sub-clusters. To further compare the composition of the microbiotas among the three food sources, we calculated abundance-based Sørensen similarity values among and within the groups (Fig 4). The results of this analysis support the conclusions drawn from the Bray-Curtis similarity analysis, i.e., strong separation as a function of food source, with very little overlap (only ~1–5%) among diet groups. However, microbiotas of individual larvae reared on artificial diet all displayed very high similarity among one another (92%). For the two groups of larvae reared on host trees, the Sørensen similarity values calculated for each sub-cluster were higher than those calculated for the whole group, supporting the conclusion drawn from the Bray-Curtis analysis and pointing to the greater variation in the composition of the midgut microflora among larvae collected and reared on natural foliage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Individual and group diversity patterns.

Dendrogram depicting the Bray-Curtis distances between each larva (left) and Sørensen similarities between or within the three groups (right). Note that individuals within the fir and spruce experimental groups form two subclades, each displaying high within subclade similarity (92–98%, in parentheses) compared to each group taken as a whole (62 and 67%).

https://doi.org/10.1371/journal.pone.0144077.g004

Discussion

The present study shows that the vast majority of bacteria found in the spruce budworm midgut belong to the phylum Proteobacteria (Fig 2), with the genus Pseudomonas constituting the dominant taxon, irrespective of rearing conditions (Fig 3). Indeed, more than 60% of the reads generated from larvae reared on artificial diet belonged to this genus, while this proportion increased > 80% in field-collected insects. A previous study using culture-dependent methods to identify bacteria from spruce budworm midguts also identified Pseudomonas as one of the taxa present in laboratory-reared insects, although in much lower proportions than those reported here [11]. This difference could result from taxon-specific differential growth of midgut bacteria on the tryptic soy agar medium used by these authors. Differential growth may also explain why the same study reported Enterococcus and Staphylococcus as the two dominant bacterial genera in budworm midguts, while these two Firmicute genera represented a low proportion of the culture-independent experiment reported here. It should also be pointed out that notable differences in the presence/absence of midgut bacterial taxa have been reported for different laboratory cultures of the same insect (e.g., Enterobacter in Lymantria dispar; [8,11]), suggesting the possibility of important differences in host genotypes controlling gut bacterial communities among different sources of laboratory-reared insects [17]. A study combining culture-dependent and culture-independent approaches for the characterization of the gypsy moth (L. dispar) midgut microbial community also identified Pseudomonas as representing a significant proportion of the microbiota [8]. Pseudomonas fluorescens, which is very closely related to the dominant bacterial OTUs in our samples, has been shown to degrade pectin, one of the components of plant cell walls [9]. Whether this bacterium plays a role in foliage digestion remains to be determined, but it could explain its possible dominance in the budworm midgut bacterial community.

The main difference between the communities of larvae reared in the laboratory and sampled in the field was the presence of a higher proportion of bacteria belonging to the phylum Bacteroidetes in insects reared on artificial diet (Fig 2). Prominent among these bacteria were the Chitinophagacea (more than 97% of Bacteroidetes reads), which could not be assigned to specific genera using the GreenGenes97 reference database; however, they were identified as Hydrotalea flava or Sediminibacterium gingensoli using the NCBI 16S database (see Table 1). These bacteria have also been reported in the gut of Nyssomyia neivai sandflies, which feed on the blood of other insects [18]. Their role, if any, in the biology of C. fumiferana is unknown, but their absence in field-collected insects suggests they are not essential and that they may be repressed by antibacterial compounds present in conifer leaves [12] or their growth may be favored in insects fed on artificial diet.

We found twice as many OTUs in larvae reared on artificial diet as compared to those collected in the field (Fig 1). Whether the presence of antimicrobial secondary metabolites in fir and spruce foliage is responsible for the lower taxonomic richness assessed in larvae collected on host plant materials is unclear at this point, as several other factors could account for this difference. Although the antibiotic present in the synthetic diet apparently did not inhibit midgut bacterial growth, it may nonetheless have affected the taxonomic composition of the intestinal microbiota. For example, rearing of budworms on an Aureomycin-laden diet over several generations may have led to the elimination of some bacterial taxa, leaving their niches open to the establishment of others. Similarly, the greater diversity of midgut bacteria in artificial-diet-reared insects could be an adaptive response to the diversity of components in this wheat-germ-based diet, thus favoring the growth of gut Bacteroidetes which have been associated with degradation of high-molecular-weight proteins and carbohydrates [19]. Finally, various other factors, including genetic differences between field-collected and laboratory-reared insects, could play a role in shaping the observed differences. An experiment involving the rearing of budworms from a single genetic stock on different diets over several generations would likely shed light on this issue. Interestingly, earlier studies focusing on lepidopteran midgut microbiota have reported a poorer microbial diversity in insects reared on artificial diet as compared to insects reared on natural host plants [20,21]. This discrepancy may simply be related to differences in experimental design, although the possibility remains that there are fir- and spruce-borne compounds that interfere with budworm midgut bacterial growth [12], whereas the opposite may be true for species where the insect’s gut microbiota depends largely on the microbial community present on the surface of the host plant [20].

In comparing the effects of diet on features of the budworm midgut microbiota, we observed a greater bacterial taxonomic similarity among larvae reared on artificial diet (92% Sørensen similarity), as compared to larvae collected on either host tree (62–67% Sørensen similarity; Fig 4). Although this difference could be due, at least in part, to the normalizing/homogenizing effects of Aureomycin (in the synthetic diet) and multi-generation rearing in the laboratory, it seems largely attributable to the presence of two clearly distinct Bray-Curtis clusters in each group of field-collected larvae (Fig 4). Since all larvae obtained from the field were processed in the same fashion, it appears that the deep within-host-tree taxonomic subdivision results from factors inherent to the larvae themselves. Given a sex ratio of ~50:50 in the spruce budworm [22] and an observed segregation of larvae in the same proportions within each host group (Fig 4), we speculate that the sex of the larvae (not recorded in this study) may be the factor responsible for the observed clustering.

Relatively minor differences were observed between the midgut microbiotas of larvae collected on balsam fir as opposed to those collected on black spruce, at least at the phylum (Fig 2) and genus (for Proteobacteria; Fig 3) levels. The main difference was a slightly greater proportion of Actinobacteria in the gut of larvae collected on balsam fir. Additional differences are likely present at the species level, as suggested by the presence of different phylotypes within each of these two diet groups (Table 1). However, the significance of such differences is unclear at this point and its assessment will require more extensive sampling of larvae on host trees from different geographical regions and habitats.

Interestingly, the present analysis suggested the absence, in the spruce budworm, of a distinct, core midgut microbiota, i.e., a set of bacterial taxa present in all processed larvae. Although some genera were present in all three diet groups (Table 1), none of the phylotypes were detected in all sampled larvae. This observation is in contrast with earlier studies where the authors reported the presence of a few bacterial species in all insects examined (e.g., [8,20]). It has been pointed out, however, that a lack of core gut microbiota among healthy individuals of a given species may not be critical so long as the microbial phylotypes present in the gut collectively provide a core set of bacterial genes required by the host [23], a situation that may apply to the spruce budworm gut microbiota.

In summary, the present work provides baseline data on the spruce budworm midgut microbiota, as determined using a culture-independent NGS approach. It also provides evidence that rearing conditions, including diet, can have a significant effect on budworm midgut microbiota species diversity. As such, the findings reported here should serve as a starting point for more in-depth studies examining the possible functions of the budworm intestinal microbiota; such functions could include processing of secondary plant metabolites, nutrition/digestion, immunity and communication, among others [4]. To this end, metagenomics and transcriptomics approaches will likely be required.

Supporting Information

Acknowledgments

We express our gratitude to members of the next generation sequencing platform and bioinformatics platform at the Institut de biologie integrative et des systèmes (IBIS), Université Laval, and to Catherine Beliveau for technical assistance with the PCR work. We thank Dr. D. Pureswaran (Natural Resources Canada, Quebec City) for providing the field-collected larvae. This research was funded by grants to R. C. Levesque, M. Cusson (NSERC Discovery grant #171350–2012 and in part by Génome Québec).