Lepidoptera is the second largest insect order, and many of them are the most devastating pests of the agricultural and forest ecosystems worldwide (Sree and Varma, 2015). Diverse microbial communities inhabiting caterpillar guts have important impacts on host biology (Engel and Moran, 2013; Paniagua Voirol et al., 2018), such as food digestion and nutrient acquisition (Pinto-Tomás et al., 2007; Anand et al., 2010; Xia et al., 2017), pesticide resistance (Xia et al., 2018; Chen et al., 2020), and host interactions with pathogens (Takatsuka and Kunimi, 2000; Broderick et al., 2006; Shao et al., 2017). Gut-associated microbiota are also actively involved in the interactions between caterpillars and their host plants. For example, the gut flora in caterpillars can alter plant response by secreting bioactive molecules to directly act upon plants (Spiteller et al., 2000; Indiragandhi et al., 2008), or regulating the secretion of caterpillar salivary elicitor to induce plant defense (Wang et al., 2017). On the other hand, there are many parasitic or pathogenic microbes in caterpillar gut, and Bacillus thuringiensis is probably the best-known species (Broderick et al., 2006; Mukherjee et al., 2010; Shao et al., 2017; Chevrette et al., 2019).

In contrast to the low-diversity bacterial communities in the gut of sap-feeding insects, species richness and relative abundance of gut associated bacteria are much higher in caterpillars (Jing et al., 2014; Santos-Garcia et al., 2020). So far, most research tended to characterize the most common microbial taxa associated with caterpillars, and to evaluate the fitness contribution of the most abundant bacteria to caterpillar hosts (Broderick et al., 2004; Xiang et al., 2006; Robinson et al., 2010; Pinto-Tomás et al., 2011; Xia et al., 2013; Chen et al., 2018; Jones et al., 2019). However, host fitness contributed by the associated microbiota is usually the outcome of the whole bacterial community (Gould et al., 2018). Therefore, it is essential to understand the processes and dynamics of gut microbiota assembly among individual hosts.

The neutral theory of biodiversity is a mechanistic model for predicting species coexistence and biodiversity patterns in ecological communities. It was initially developed to predict ecological communities of large organisms such as plants and animals, and recently applied to microbial community ecology (Bell et al., 2005; Sloan et al., 2006; Woodcock et al., 2007; Caruso et al., 2011; Levy and Borenstein, 2013; Morris et al., 2013; Nemergut et al., 2013; O’Dwyer et al., 2015; Venkataraman et al., 2015; Li and Ma, 2016). In contrast to the traditional niche theory (deterministic and selective processes), which predicts that species with different properties occupy niches with different characters, the neutral theory assumes that stochastic processes, independent of host traits, play important roles in shaping microbial community composition in an individual host. In other words, it assumes microbial species in a community are functionally equivalent, and the assemblages are driven by random dispersal and birth/death events (Costello et al., 2012). Therefore, taxa that are abundant in the intestinal communities of all individuals are more likely to disperse by chance and be randomly sampled by an individual host, and less abundant taxa can be lost from an individual host due to ecological drift. The microbes deviating from the neutral model can be good candidates worthy of further investigation for their potential roles in a host (Sieber et al., 2019). The structure of Drosophila gut microbiota has been extensively investigated since the early work by Wong et al. (2011), and the recent application of the neutral model further provided key insights into the processes shaping bacterial communities in these model organisms (Martinson et al., 2017; Adair et al., 2018). However, evaluation of the contribution of the neutral processes to insect microbiota community structures remains rare.

In contrast to previous research based on the pooled samples (Mishra and Tandon, 2003; Xiang et al., 2006; Priya et al., 2012; Tang et al., 2012; Ranjith et al., 2016; Shinde et al., 2017), we profiled gut-associated bacteria in individual H. armigera larvae collected from maize plants at four locations to address two aims in this study. First, the contribution of neutral processes to the gut microbiota community associated with the individual caterpillar was evaluated. As far as we know, no investigation of neutral processes has been conducted for microbial communities in caterpillars. Second, we examined biodiversity patterns in microbial communities and ecological determinants of variation among individual caterpillar hosts.

As an important agricultural pest, the microbes dwelling inside the gut of H. armigera caterpillar were previously characterized by both culture-dependent and culture-independent methods (Mishra and Tandon, 2003; Xiang et al., 2006; Priya et al., 2012; Tang et al., 2012; Ranjith et al., 2016). The top abundant phyla tended to be similar between H. armigera populations reared on different plants at different locations. For example, Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria and Acidobacteria were the major phyla whether the caterpillar was collected from tomato in India or from maize in China. In this study, we evaluated the contribution of neutral processes to the structure of caterpillar gut microbiota, and found that the distribution of a large proportion of gut-associated microbiota in the H. armigera caterpillar can be explained by stochastic processes. To our knowledge, the contribution of neutral processes to the structure of insect microbiota has only been investigated in Drosophila (Martinson et al., 2017; Adair et al., 2018).

Most gut bacteria fell into the 95% limits of the neutral model, suggesting that a high proportion of the gut-associated bacterial taxa was likely transient. Accordingly, Hammer et al. (2017) found that the surveyed wild caterpillars tended to lack a resident gut microbiome, and most ingested microbes (i.e., microbes in the food) did not colonize in the gut of caterpillars (Hammer et al., 2017). Similarly, the gut microbiome of five Drosophila species also fitted well with the predictions of the neutral model (Martinson et al., 2017; Adair et al., 2018). Research on the contribution of neutral assembly processes to microbiota structures in insect hosts is still very limited. Whether bacterial communities of other insects behave according to neutral expectations warrants further investigation. However, there are varying cases reported for non-insect species. For example, in the nematode Caenorhabditis elegans, a well-known bacterial feeder, a large number of microbial species deviated from the neutral expectation (both above- and below-partitions; Sieber et al., 2019). These worms naturally feed on bacteria and can likely have strong influence on their microbiota through host defense system. Interestingly, observed and expected patterns in vertebrate microbiota are more complex due, in part, to their more sophisticated immune response systems. Li and Ma (2016) concluded that human microbial communities, covering the five major body sites such as airways, gut, oral, skin, and urogenital, were not neutral in general, by observing that more than 99% of 7,437 datasets from the human microbiome project data center deviated from the neutral expectation (Li and Ma, 2016). In contrast, the intestinal microbiota of a wild Mus musculus population collected from 34 unique locations in Bavaria, Germany showed a very good alignment with the neutral expectation (Wang et al., 2015; Sieber et al., 2019). Moreover, the host status can influence the fit of the neutral model. For example, heathy lung microbiota largely followed the pattern of neutral processes but microbes in diseased lungs did not (Venkataraman et al., 2015).

A number of bacterial taxa deviated from the neutral prediction in our study, indicating that they may be more adapted to, or even selected by, the caterpillar gut environment (Sieber et al., 2019). It has been shown that caterpillars can grow well without gut-associated bacteria (Hammer et al., 2017; Phalnikar et al., 2019), but this does not preclude the potential of some gut-associated bacteria to improve host fitness under certain selection pressures (Spiteller et al., 2000; Takatsuka and Kunimi, 2000; Broderick et al., 2006; Pinto-Tomás et al., 2007; Indiragandhi et al., 2008; Anand et al., 2010; Shao et al., 2017; Xia et al., 2017, 2018; Cassone et al., 2020; Chen et al., 2020). Among the bacteria species in the above-partitions, some taxa repeatedly appeared in different caterpillar populations. Interestingly, the functions of certain species belonging to the genera, Acinetobacter, Enterococcus, Massilia and Streptomyces, have been investigated in caterpillars in other studies. A study suggested that a gut-associated strain of Enterococcus can enhance insecticide resistance of Plutella xylostella to chlorpyrifos by regulating the immune system in this insect (Xia et al., 2018). Bacteria species in the genera of Acinetobacter and Massilia can assist their caterpillar host, Galleria mellonella, to degrade the ingested polyethylene and polystyrene (Cassone et al., 2020; Lou et al., 2020; Jiang et al., 2021a). Although polyethylene and polystyrene do not contain the necessary nutrients for insect growth, the food ingested by H. armigera caterpillars inevitably contains these environmental pollutants. Some Streptomyces strains have been shown to protect hosts such as caterpillars and other insects from pathogens (Chevrette et al., 2019; Kim et al., 2019). However, not all species in a certain genus may perform similar function. Metagenomics studies may help elucidate the potential for different combinations of closely-associated bacteria taxa (the taxa in the above-partitions) to improve host fitness under different environmental challenges.

The midgut of caterpillars is highly alkaline (Dow, 1984; Johnson and Felton, 1996) and can secret antimicrobial peptides (Jiang et al., 2010). Thus, the bacteria species in the below-partitions were possibly more sensitive to this hostile environment and eliminated quickly once entering the gut lumen. Interestingly, the bacterial taxa below the prediction tended to be similar among individual caterpillars collected from different hosts in various locations, indicating that these taxa may be potentially pathogenic agents to the caterpillars. They may be recurrently building relationship with the caterpillar hosts at various environmental conditions but significantly suppressed by the host immune system in the healthy individuals.

Both the maize variety and location contributed to gut microbiota variation in H. armigera, so both plant host and location can also contribute to the structure of gut-associated microbiota. The effect of location appeared to be lower than that of diet. In agreement, H. armigera larvae collected from the same plant species at different locations tended to share more similar microbiota (Ranjith et al., 2016). Caterpillars generally acquire gut bacteria from their food and environment (Robinson et al., 2010; Tang et al., 2012; Whitaker et al., 2016; Hammer et al., 2017; Chen et al., 2018; Jones et al., 2019) and most of the bacterial species were transient (as discussed above), so variation in the phyllosphere community may also contribute to the difference seen in the gut-associated microbiota between different maize varieties and locations.

In summary, most OTUs in the H. armigera gut followed the distribution of the neutral model with a few deviating from the neutral expectation. These findings indicated that the neutral processes should be considered as one important factor when investigating the composition and structure of host-associated microbial communities in caterpillars. In the future, functions of taxa that consistently deviate from the neutral expectation can be investigated by either functional genomic analysis or culturing experiments, which can substantially advance our understanding of biodiversity, community structure and function of caterpillar microbiomes.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih.gov/, BioProject: PRJNA769459.

XJ, SZ, and SL conceived the ideas and designed methodology. SL, RT, HY, ZC, and SS conducted the sample preparation. SL and XJ analyzed the data. XJ, SL, SZ, and T-XL led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

This study was supported by the National Natural Science Foundation of China (31872299), the National Key R&D Program of China (2016YFD0300705), and the Introduction of Talent Research Start-up Fund of Northwest A&F University (2452017323).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.