Leaves harbor diverse and abundant microbial communities–both bacterial12,38 and fungal13,41. However, the high overall richness in the caterpillar guts contradicts previous findings reporting species-poor bacterial communities7,30,39 Contrary to our previous study40, bacterial microbiome was richer than fungal microbiome, especially considering that bacterial richness was estimated at the genus level and fungal richness at the species level. The accurate quantification of the 16S and ITS2 DNA in the initial samples is necessary to assess the proportional representation of bacteria and fungi in the gut microbiota. Moreover, the extent to which microbial communities obtained by metabarcoding are composed of legacy DNA from dead or dormant cells and spores remains to be determined42.
Both the bacterial and fungal components of leaves were significantly affected by tree species, which is a typical pattern43; however, they always significantly interacted with locality, implying that trees at different localities harbored specific microbial assemblages. This contradicts a recent study suggesting low variability in leaf microbial diversity among individual sites44, although it accords with the fact that the leaf microbes are acquired from the environment, but their survival is filtered by the plant45. The microbiomes of a focal host species may be affected by local plant community composition and diversity31,37,44; these measurements were beyond the scope of our study but could be responsible for the observed spatial differences.
Irradiation significantly affected the fungal richness and bacterial composition and richness. Sun exposure is a good predictor of fungal abundance46 and diversity47. For bacteria, Stone and Jackson48 found distinct community compositions across canopy positions, with speculative attributability to radiation; however, shaded leaves are less exposed to rain, which changes bacterial composition. Moisture availability may have a substantial impact on leaf microbiota49; desiccation, especially in combination with UV radiation, strongly limits microbial populations50. However, in leaf bacteria, UV radiation has been found to have no effect on species richness51.
In caterpillars, bacterial composition and richness were primarily shaped by caterpillar species. In fungi, the role of the host species is only secondary; they are shaped primarily by spatial variables. Great interspecific differences were quite surprising, given that numerous studies have reported similar communities shared among caterpillar species, both implying the low importance of the host physiological environment in structuring microbial communities and emphasizing the dietary effect19,29,35,39. In contrast, independent of diet, there are interspecific differences in the physiochemical conditions of the gut, which exert strong selection pressure on microbiota5,17. From this point of view, the associated differences in gut microbiota were expected.
Bacterial composition was significantly affected by caterpillar body length. Changes in community composition are attributed to the increasing importance of gut filtering throughout the caterpillar life cycle7,28,52. As caterpillars grow, less oxygen penetrates the gut lumen, which promotes the development of facultative anaerobic bacteria, mainly Enterobacteriaceae, decreasing diversity23. This was not reflected here; we sampled only similarly sized caterpillars (3–4th instar) to suppress the effect of developmental stage. This protocol did not result in a dataset composed of same-instar caterpillars, and some effect of body length may be attributable to the above-mentioned effect.
Apart from the strong effect of caterpillar species, the differences in the composition and richness of gut bacterial microbiome were shaped by irradiation, spatial variables, and their interaction with caterpillar species. The climatic and ecological factors of host habitats are known to affect the insect gut bacterial composition53,54. However, as individual conditions of individual plots affect individual caterpillar species differently, the effect of environmental conditions remains unsolved. Alternatively, bacterial richness and composition may be strongly affected by biotic conditions, especially parasitoid infection26. Spatial differences in parasitism rate and parasitoid community composition are well known55,56. Although we tried to eliminate all parasitized samples before processing, parasitoid juvenile stages could have been overlooked and may have contributed to the differences in bacterial composition and richness.
Host tree species also significantly affected gut fungal and bacterial microbiomes. Bacterial gut communities are greatly influenced by diet57. Here, however, leaf bacteria were primarily affected by spatial variables. Thus, bacterial gut content may be affected not by the microbial composition of the diet, but rather by its quality (protein and carbohydrate content)20 and plant secondary metabolites25, which places major selective pressure on the gut microbiota. This pressure seems to be less important for fungal components, which are primarily affected by spatial variables, either in the diet or gut.
The compositions of both bacterial and fungal gut components significantly differed from those of leaves; however, in bacteria, this difference was much more pronounced. Caterpillars move over relatively long distances when feeding58, and Orthosia spp. are even occasional entomophages59. Thus, they probably sampled a much greater microbial pool than that reflected in the relevant leaf sample, which may have contributed to their low similarity. Bacterial richness was higher in the leaves than in the guts, whereas fungal components showed an opposite pattern, suggesting that fungi may be less filtered than bacteria. The composition of the leaf bacteria was rather balanced, whereas in the guts there was a great host-interspecific variability. Contrarily, fungal component of both leaves and guts was dominated by Aureobasidium pullulans, Ramularia, and Dothiora, again pointing out environmental acquisition greater than in the case of bacteria.
This finding was corroborated by the analysis of the factors that form individual microbial components of the gut. In fungi, the caterpillar species was only a secondary factor explaining less variability than the spatial variables (locality and sampling plot), which predominantly shaped the richness and composition of the diet. However, the bacterial component of the guts was shaped primarily by caterpillar species and was affected by the caterpillar body length, indicating a greater involvement of gut filtering. Bacteria in the diet varied greatly among individual localities; however, gut bacteria did not reflect this. The spatial variability of bacteria manifested only at the level of individual localities, which may reflect the host’s adaptation to local conditions. These results indicate that the fungal component of the gut is more transient, whereas bacteria form a core component.
The leaf–gut similarity of bacterial—but not fungal—components differed significantly among caterpillar species, suggesting that some species have lower leaf–gut similarities than others (specifically A. aescularia and O. brumata). This suggests a higher involvement of resident (core) bacterial taxa or stronger environmental filtering. This may be explained by their similar life histories and dispersal strategies. Unlike the rest of the study species, adults remain active during winter, occurring in high abundances60. For both species, ballooning dispersal, in which caterpillars use silk to move through the air, was documented61. Both strategies could contribute to sampling different microbial pools from the environment. Moreover, extreme winter conditions may alter adult gut microbiota, which may be vertically transferred to the offspring through the contamination of the egg surface23,28. Furthermore, while bacterial microbiota of some hosts (P. munda, A. aescularia, P. pilosaria, O. brumata) was quite similar to each other, other species (E. defoliaria, L. dispar) hosted more specific bacterial consortia, suggesting diversity in directions of environmental filtering.
Our study indicates the complexity of the factors shaping leaf and caterpillar gut microbiota, which makes it difficult to draw conclusions, even with a large dataset. Other important factors were not accounted for in this study, namely interactions within and among microbial groups that are abundant and of great importance, in both the leaves62,63 and gut3,64. The relatively static physiochemistry of the host as well as dynamic microbe-microbe, microbe-host and host-mediated microbe-microbe interactions are likely the drivers of microbiota community composition43, as applied to both environments. Future studies using the community-level approach may clarify the relative importance of stochastic and deterministic processes in governing gut microbiota assembly and how this importance varies through space and time. A functional approach using transcriptomics, which identifies biologically active taxa, would complement these studies and elucidate the link between the core component of the gut and its significance for the host.