The global population is growing exponentially, with current forecasts exceeding 9 billion people by 2050 (1). To feed a population of that size, annual food production must increase by 70% and annual meat production must increase by 200 million tons. Obstacles to increased food production include climate change, depleting water sources, and restrictions on expanding farmable land. In addition, animal feed production costs from protein-rich ingredients like fishmeal and soybean meal are increasing (2). Therefore, sustainable alternative protein sources are needed to supplement food and feed, and insects are promising candidates. Insects are high in nutrients like insoluble fiber, proteins, vitamins, and minerals (3). Insects also consume less water per gram of protein produced and require lower input feeds than most traditional livestock (4). The yellow mealworm, Tenebrio molitor, is one of the most popular and well-researched insects for supplementing animal feed. Companies already mass-rear T. molitor on an industrial scale for feeding domestic birds, fish, and reptiles (5).

There is considerable research on diseases affecting T. molitor because of its notoriety as a pest of poultry houses and its use as a lab model for coleopteran biochemical research. Pseudomonas and Serratia species are opportunistic bacterial pathogens that infect T. molitor colonies under stress (6, 7). Gregarines are known apicomplexan parasites of T. molitor, but infection typically only results in a decrease in the adult lifespan, not affecting the fitness of larval stages (8, 9). Entomopathogenic fungi, including those from the genera Metarhizium and Beauveria (10), as well as nematodes (11) also have been documented as potential biological control agents for T. molitor. Our research has used T. molitor as a model for studies of the mode of action of Bacillus thuringiensis Cry3Aa toxins in coleopterans (12–16).

Often insects in rearing facilities are in population-dense environments with a high potential for pathogen spread throughout the colony (17). Insects do not have an adaptive immune system, so disease prevention methods that rely on learned immunity, like vaccines, cannot be used to prevent illness in insects as they do with traditional livestock. Although some entomopathogens may be prevented or treated with antibiotics, overuse of antibiotics in other farming industries has contributed to the development of antibiotic-resistant bacteria (18). Farmed insects represent considerable investment, and as such there is increased interest in identifying diseases that may affect insects produced for feed, and development of prophylactics and treatments for insect diseases in reared systems (19, 20). Current prevention and treatment recommendations are suitable for smaller rearing facilities or for those who rear insects at home but are not practical for larger facilities (5). Our long-term goal is to assist larger facilities in avoiding the practice of removing infected colonies and replacing with new stock, especially with the costs of genetically-selected or modified strains. More research is needed on the detection, treatment, and prevention of pathogens that may affect mass-rearing efficiency and colony health of mealworms.

This case study uses DNA sequencing to identify pathogens that may be responsible for the collapse of a T. molitor colony that had decreased survival of egg to adult and subsequent mite infestation. The data identifies presumably beneficial bacteria in a healthy mealworm microbiome and pathogenic bacteria that were associated with the T. molitor colony collapse.

This study investigated a T. molitor colony that demonstrated reduced viability and eventual collapse. In May 2021, the colony suffered a flour mite infestation that likely contributed to the collapse by competing for resources ( Figure 1A ), with only a few T. molitor larvae surviving. Larvae from the collapsed colony had reduced movement with darkening and melanization of the gut, evidence of bacterial infections in T. molitor (20, Figure 1B ). When problems were identified with the original colony, T. molitor were obtained from three commercial sources and eventually combined into a colony labeled “new”. The approximately 15 surviving larvae from the collapsed colony were transferred to a new container with fresh diet and remained quarantined from the newly established colony. With careful monitoring, the survivors of the collapsed colony, labeled “recovered”, were able to pupate, reproduce, and began to reestablish. As time passed, there were fewer larvae exhibiting signs of infection, and the mite population was brought under control with careful hygiene.

We sequenced DNA from an adult from the collapsed colony (C-Adult), as well as their frass (C-Frass) and diet (C-Diet), and DNA from a recovered larva (R-Larva) ( Table 1 ). We also sequenced DNA from a new colony adult (N-Adult) and their frass and diet combined (N-Diet). Sequencing provided from 3.8 to 9.3 million reads for downstream analysis ( Table 2 ). Additionally, we sequenced over 23 million reads from fresh diet (F-Diet). Reads were analyzed for similarity to sequences in microbial databases. The majority of reads in all samples were unclassified, as they were likely host (T. molitor, Additional File S1A ). At the superkingdom level, most reads were bacterial ( Additional File S1B ), the focus of this study. Diets had a higher percentage (16.4-29.3%) of eukaryotic reads than insect samples (0.99-9.93%) due mostly to contribution from reads similar to Saccharomycetes and likely due to the yeast in the diet ( Additional File S1A ). Some reads were assigned to Mammalia, but likely those were insect reads aligning to orthologs in the database. Fewer reads were associated with archea and viruses.

The lowest percentage of classified reads was from R-Larva DNA (2%), whereas 28% of reads from the C-Frass sample were classified ( Table 2 ). We compared reads from the major bacterial classifications among the samples, focusing on of those were assigned to families at 1% or more of the total reads ( Figure 2 ). Overall, the taxonomic distribution of the bacterial reads from the new colony samples were more similar, and those of the collapsed colony were similar. The taxonomic distribution of the bacterial reads from the R-Larva sample was intermediate to the distributions of those from the new colony and the collapsed colony samples.

Reads derived from the family Staphylococcaeceae were predominant in the microbial reads from the collapsed colony samples, constituting 53, 76, and 88% of the total classified reads in C-Adult, C-Frass, and C-Diet, respectively ( Figure 2 ). In contrast, reads associated with Staphylococcaeceae in the new colony samples were 12, 13, and 30% for N-Diet, N-Adult, and N-Pupa, respectively. Reads from the bacterial family Streptococcaceae also were prevalent in the collapsed colony, constituting 4-23% of classified reads compared to 0-6% in the new colony. The newly established colony had higher percentages of bacteria from the families Lactobacillaceae (21-47%), Paenibacillaceae (4-9%), and Petrotogaceae (4-19%) than the collapsed colony (2-10, 0, 0-2%, respectively).

At the genus level, large increases in reads from the collapsed colony samples were observed from the genera Mammaliicoccus, Lactococcus, Staphylococcus compared to those from the new colony samples ( Additional File S2 ). However, the collapsed colony samples had fewer reads from the genera Lactobacillus and Lacticaseibacillus than those from the new colony samples, in agreement with observations at the family level.

We were particularly interested in the family Bacillaceae since the phenotype of some of the collapsed insects suggested an active B. thuringiensis infection ( Figure 1 ). Therefore, we examined sequences in the genus Bacillus through a targeted blast search to the B. thuringiensis database at NCBI. There were a larger number of reads from the Bacillaceae family in the samples from the new colony, but most of these were from different Bacillus species and other Bacillaceae genera ( Additional File S3 ). More reads with high scoring blastn matches to B. thuringiensis were found in the collapsed colony samples, especially in the C-Frass sample ( Table 3 ). The number of reads in the new colony samples were too low (1, 2) to confirm the presence of B. thuringiensis. However, 240 reads from C-Frass aligned to Bacillus spp, and 74 aligned to B. thuringiensis serovar morrisoni str. 4AA1 (taxid ID 1461024) with evalues < 10^-40 ( Additional File S3 ).

Fresh samples of rearing diet (F-Diet) were sequenced to confirm that was not the original source of B. thuringiensis in our collapsed T. molitor colony. The number of F-Diet sample reads were more than 4-fold greater than those in the other samples to ensure that B. thuringiensis would be identified if present ( Table 2 ). However, none of the reads in F-Diet were identified as B. thuringiensis, indicating that the rearing diet likely was not the source of B. thuringiensis found in the other samples ( Table 3 ).

A comparison of microbial sequences from diets demonstrated the progression of loss of microbial diversity in pre-rearing diet to diet collected from the collapsed colony ( Figure 3 ). The pre-rearing diet (F-Diet) only had 0.4% of microbial reads but from a more diverse bacteria flora, whereas the bacterial profile in diet from the new colony (N-Diet), also 0.4% of total reads, had shifted to primarily Lactobacilleceae. In contrast, microbial sequences in the diet from the collapsed colony (C-Diet), with 2.5% of the classified sequences as bacteria, had shifted to Staphylococceae with fewer other bacterial sequences found.

High throughput sequencing can be an effective tool to understand the microbial profiles in insects, and this approach can be even more effective in diagnostics of diseased colonies or routine monitoring of farmed insects. Successful results are dependent on sequencing depth and the accuracy of microbial databases. We were limited by sample size in the collapsed colony, as only a few individuals survived. While we were able to identify meaningful and actionable data from the small number of samples in this study, improvements could be made with additional samples from different developmental stages and increased sequencing depth. As more insect and microbial sequences are acquired by data depositories, the accuracy of predicting microbial profiles will be improved.

In this study, we found that reads from the families Staphylococcaeceae and Streptococcaceae collectively constituted 76% of microbial families in the adult sequenced from the collapsed colony, with the bacterial communities of the frass and diet from that colony also containing large numbers of reads (93% and 86%, respectively, of the total bacterial reads in each sample). This narrow microbial diversity represented a shift from the more diverse community found in the adult from the new colony, with only 21% of bacteria from Lactobacilleceae, and the remainder of microbial reads spread over more than 15 different families. Similar profiles were found in the diet from the new colony. There also was a difference in the overall relative abundance of bacterial reads found in the two colonies, as the total number of microbial reads in the new colony was 62-68% for the sequenced larva and adult, whereas more microbial reads (85%) were found in the collapsed adult.

The phenotype of the diseased adult in the collapsed colony suggested there was an active B. thuringiensis infection, and our sequencing data support this observation. There were 74 reads from the collapsed colony diet that aligned to B. thuringiensis serovar morrisoni, and this serovar typically expresses Cry3Aa toxins that are insecticidal to Coleoptera (22, 23). However, it is unclear how the infection occurred, although finding a small number of reads aligning to B. thuringiensis in the new colony adult and diet suggests that these bacteria were present, possibly via the sweet potatoes that are introduced to larvae and adults during rearing.

These data do not inform us as to whether the bacteria identified in the collapsed adult contributed to the colony collapse, but more likely they were a result of the sepsis that occurs in the intestinal tract during infection with the insecticidal toxins from B. thuringiensis. Staphylococcus aureus was found in the intestinal tract of Musca domesticus and is speculated to be contained in that insect via excretion, digestion, and antimicrobial peptides, indicating that insects may have an active system to contain harmful bacteria (24). Streptococci, specifically enterococci, have been identified in more than half of insects surveyed, and it was speculated that insects may serve as an overwintering host for these bacteria (25). Given that the adult sequenced from the adult colony was dead, we cannot know if the microbiota populations were a result of postmortem changes in the insect. On the other hand, our data suggests that Lactobacilli may be beneficial to mealworms as they were found in 21% of the microbial reads from the healthy adult in the new colony. In Hymenoptera, evidence supports that hypothesis, in that some bees have host-specific Lactobacilli, and may be important for bee health (26). Lactobacilli can check the growth of other bacteria and are often components of probiotics, and these bacteria were among those suggested as beneficial for farmed insects (27). Our data supports the hypotheses that all of these bacteria are part of the normal flora of the mealworm gut tract and only become pathogenic when other factors disrupt their balance in the microbiome.

A healthy microbiome has a large diversity of species over highly diverged phylogenetic taxa (28). Divergent taxa fill unique roles within a system, but when closely related taxa increase in a biological system, these roles break down and a dysbiosis event occurs. Our data indicates that during the bottleneck event of the collapsed colony, a dysbiosis occurred where there was a major shift in the microbiome, either leading to the colony collapse or a factor of the collapse.

While the high mite infestation was competing for resources and emphasizes the need for frequent culturing of reared insects, we never found the mites directly feeding on T. molitor developmental stages. There are no references that indicate that mites can pass on pathogens to insects, but that possibility remains.

More data are needed on the natural and diseased profiles of insects reared for downstream food applications. Our efforts are in developing insect-specific pathogen databases that can rapidly and accurately identify insect pathogens harmful to farmed insects (disease diagnosis) as well as those that can identify potentially harmful pathogens in insect food and feed for downstream applications (disease prevention).

The datasets presented in this study can be found in online repositories. The name of the repository and accession number can be found below: NCBI; PRJNA883751.

DL, MO, BO were involved in all aspects of the project and manuscript preparation. All authors contributed to the article and approved the submitted version.