Dietary Association with Midgut Microbiota Components of Eocanthecona furcellata (Wolff)

Abstract

Eocanthecona furcellata is an important predatory stinkbug that attacks many lepidopteran pests. For mass-rearing, artificial diets are used to rear this predator in the laboratory; however, the fitness of the predators is reduced, and little is known about the cause. Since gut microbiota plays vital roles in the digestion and development of many hosts and can consequently affect host fitness, an understanding of the microbial community composition of E. furcellata may help to solve this unresolved problem. We compared the development and reproduction of E. furcellata reared on an artificial diet, and a natural (Spodoptera litura) or semi-natural (Tenebrio molitor) diet, and then the midgut microbiota were assessed using high-throughput 16S rRNA. The results of the high-throughput 16S rRNA show that the bacterial richness and diversity in the artificial diet gut samples increased considerably compared with the other samples. Proteobacteria and Firmicutes were the dominant phyla in E. furcellata. At the genus level, Serratia (however, the relative abundance was lower in the artificial diet gut samples), Enterococcus, and an uncultured bacterium genus of family Enterobacteriaceae, were dominant. The midgut microbiota components significantly differed among the diets, indicating that the gut bacteria had a dietary association with E. furcellata. This study provides a better understanding of midgut microbiota and the artificial diets that might affect them in E. furcellata.

1. Introduction

Eocanthecona furcellata (Wolff) (Hemiptera: Pentatomidae), a predator to many serious pests, is widely distributed in the subtropical and tropical regions of southern China, India, and Indonesia [1]. In China, E. furcellata reportedly suppresses populations of pests such as the fall armyworm (Spodoptera frugiperda (Smith) (Lepidoptera: Noctuidae)) [1], tobacco cutworm (Spodoptera litura (Fabricius) (Lepidoptera: Noctuidae)) [2], and oriental fruit fly (Bactrocera dorsalis (Hendel) (Diptera: Tephritidae)) [3]. In large-scale agricultural settings, the use of natural or factitious foods is usually time-consuming for the mass-rearing of E. furcellata under laboratory conditions [4]. The mass production of biological control agents depending on natural or semi-natural diets could be costly and are more often than not affected by the season [5]. Therefore, the development of artificial diets is economical and possesses high biological efficiency for the mass-rearing of predators in the laboratory [6].

Attempts have been made, but they have not been adequately successful because of a slightly lower E. furcellata performance compared with more natural diets that consisted of T. molitor, even with the same functional response on the same prey [7]. The food source and quality affect the fitness of many predators and parasitoids [8,9]. By providing laboratory E. furcellata with an artificial diet, we noticed that the developmental times and preoviposition periods were extended. Since the fecundity tended to decrease, we considered that the artificial diet may have negative transgenerational effects on the life history of E. furcellata. Riddick [10] concluded that most vertebrate protein-based artificial diets are inferior to natural prey or semi-natural prey for predatory natural enemy insects. Lacewings, Chrysopa pallens (Rambur) (Neuroptera: Chrysopidae), reared on an artificial diet based on beef liver, beef powder and Chinese oak tussar moth pupal powder had comparable developmental times and better fecundity to natural preys [4]. The same artificial diet was used in the rearing of Orius strigicollis (Heteroptera: Anthocoridae) but the developmental times were longer [11]. Gut microbes influence the growth, development and reproduction of insects. Dillon and Dillon [12] reported that diets including animal-derived components may affect the physicochemical properties of insects’ gut environment, thereby inhibiting or promoting gut microbiota. Gut microbiota have the ability to metabolize or process food, thus affecting the subsequent nutrient absorption of the host [13]. Symbiotic microorganisms are known to play a pivotal role in the development of a large range of plants and animals [14,15] by breaking down otherwise unavailable nutrients to the hosts in the diet [16,17,18,19]. In some stinkbug species, the elimination of the microbiota causes delayed growth or inhibited growth and nymphal mortality [20,21]. Besides a nutritional role, gut microbiota, in some rare cases, protect hosts from predators, parasites, and pathogens, although the exact underlying mechanisms are not yet clear [22].

Previous studies have shown that insect gut microbiota are dynamic depending on the diets [23,24]; therefore, the lower performance of E. furcellata fed artificial diets may be related to their gut microbiota. We considered that an artificial diet could change the gut microbiota community components of E. furcellata to induce lower performance compared with those fed natural prey. Since gut microbiota have not been characterized in this species, E. furcellata were reared on different diets for several generations. Then, using a high-throughput 16S rRNA sequencing method, the gut microbiota was determined. Here, we present data on the fitness and gut microbes of E. furcellata fed three different diets to investigate the influence of food on fitness and gut microbiota community, which may provide a reference for future research on the gut microbiology of E. furcellata and improvements to their artificial diets.

2. Materials and Methods

2.1. Study Insects

In the small- and large-scale rearing of predators, stored product pests such as the mealworm beetle, Tenebrio molitor L. (Coleoptera: Tenebrionidae) [25], are commonly chosen for their easy access. For laboratory rearing of E. furcellata, S. litura larvae [2] and T. molitor pupae [26] are usually chosen. Wild E. furcellata nymphs were initially collected from a cowpea (Vigna unguiculata (L.)) farm in Guangzhou, China. They were fed with a mixture of T. molitor pupae and S. litura larvae at a ratio of 2:1. Tenebrio molitor pupa is not a natural prey of E. furcellata, therefore, we call it semi-natural or alternative. Although T. molitor pupae are semi-natural, the nymphal growth, survival, and reproduction of E. furcellata fed T. molitor pupae were comparable to those of E. furcellata fed natural prey [3]. Tenebrio molitor pupae were commercially available. Spodoptera litura larvae were from a population that had been reared in the laboratory on artificial diets [27] for two generations. Eocanthecona furcellata was reared this way for nine generations and the tenth-generation 2nd instar nymphs were used at the onset of the experiment.

2.2. Preparation of Artificial Diet

The artificial diet was prepared as per Gong [7]. The formula of the artificial diet was as follows: 20 g maggot powder, 10 g bovine liver, 10 g chicken egg yolk, 2 g beer yeast powder, 1 g royal jelly, 0.1 g choline chloride, 0.1 g vitamin C, 0.1 g trehalose, 0.05 g streptomycin, 5 mL hemolymph of Chinese oak tussar moth, 5 mL honey, 8 mL milk, 2 mL vegetable sap, and 100 mL sterile water. All ingredients except hemolymph of the Chinese oak tussar moth, Antheraea pernyi Guérin-Méneville (Lepidoptera: Saturniidae), were commercially available. The hemolymph of A. pernyi was collected from the pupae. The integuments of the pupae were first removed of integument by scissors and then incubated in 60 °C water bath for 10 min to prevent melanization of the hemolymph [6]. Hemolymph was then collected in a clear plastic bottle (80.00 mL) for preparation. The ingredients were blended in a magnetic mixer (JB-2, Changzhou Aohua Instrument Co., Ltd., Changzhou, China) for 20 min until homogenous. A 130 × 90 mm Parafilm membrane was stretched to about 2× its original length and width [28]. The artificial diet was placed on the membrane, which was folded and stuck tightly together to formulate 5.5 mm diameter capsules, which were poked with a chopstick. The volume of the diet itself was 0.05 mL. All materials and tools were sterilized, and all steps were completed on a clean bench to ensure no contamination. The artificial diets were prepared every week and kept in a freezer set at −20 °C.

2.3. Treatments

Three diets/treatments were prepared: an artificial diet (five packets), S. litura larvae (five 3rd instar), and T. molitor pupae (five) were established in the study (Figure 1). See Study Insects for the sources of Spodoptera litura larvae and T. molitor pupae.

2.4. Performance of E. furcellata Fed with Three Diets for Three Successive Generations

Eocanthecona furcellata were subjected to one of the three diets/treatments and reared for two generations. To accurately measure fitness, the F3 generation 2nd instar E. furcellata nymphs were then individually placed in Petri dishes (9 cm diameter × 1.5 cm height; Jiangsu Kangjian Medical Apparatus Corporation, Taizhou, China) and fed the respective diets before being individually reared. A cotton ball soaked with water was also provided. The diets and cotton balls were replaced with fresh ones daily. The developmental time from the 2nd instar to the emergence of adults were determined by monitoring molting events once a day. Emerged E. furcellata were sexed and paired to examine their fecundity and reproduction. To study the egg hatching rate, the one egg masses from all females from each diet were monitored. The whole experiment was conducted at 26 ± 1 °C, 65 ± 5% RH, in a 16:8 h L/D photoperiod in the laboratory. At the onset of the experiment, 100 E. furcellata nymphs were prepared for each diet.

2.5. Sequencing of the 16S rRNA and Microbial Characterization

The E. furcellata were subjected to one of the three diets/treatments and reared for three generations at 26 ± 1 °C, 65 ± 5% RH, in a 16:8 h L/D photoperiod in the laboratory. Four randomly chosen F3 E. furcellata female adults (the most critical stage for mass reproduction) from each diet were dissected, and the midgut materials were collected in sterile conditions. To exclude potential differences due to E. furcellata sex, only adult females were investigated in the current study. The sample was immediately subjected to Illumina sequencing (see below for procedure) or stored at −80 °C. Each diet was replicated four times (i.e., samples).

The total DNA from the midgut samples was extracted with the NucleoSpin 96 Soi (MACHEREY-NAGEL, Düren, Germany) following the manufacturer’s protocol. After extracting the total DNA of the sample, primers (338F: 5′-ACTCCTACGGGAGCAGCA-3′; 806R: 5′-GGACTACHVGGGTWCTAAT-3′) were designed according to the conserved region V3+V4, and sequencing adaptors were added to the ends of the primers. The target sequences were amplified by PCR, and their products were purified, quantified, and homogenized to obtain a sequencing library. Then library QC was performed for constructing libraries, and qualified libraries were sequenced on an Illumina HiSeq 2500 (Beijing Biomarker Technologies, Shunyi District, Beijing, China). Sequence reads were checked for quality using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc, accessed on 13 August 2020). Low-quality, low-complexity reads or those containing N were pre-processed following Fadrosh et al. [29]. Paired-end sequences were edited by the open-source software FLASH (v1.2.11, https://sourceforge.net/projects/flashpage, accessed on 13 August 2020) following the method of Magoc and Salzberg [30]. Operational taxonomic units (OTUs) were generated based on a minimum of 97% sequence similarities to true biological sequences in the library of USEARCH (v10.0) [31]. The number of common and unique OTUs among the treatments was analyzed by the web-based tool InteractiVenn [32]. OTUs were assigned at 97% similarity and mapped to the Silva database (Release 128, http://www.arb-silva.de, accessed on 13 August 2020) [33] for functional prediction. Chimera generated by PCR amplification were removed from the OTU representative sequence using UCHIME (v4.2) [34].

2.6. α- and β-Diversity of Microbial Community

α-diversity reflects species richness and the diversity of an individual sample. Four indices—Chao1, abundance-based coverage estimator (ACE), Shannon, and Simpson were calculated using Mothur v1.30 (http://www.mothur.org, accessed on 13 August 2020). The Chao1 and ACE metrics measure species richness (i.e., the number of species). The Shannon and Simpson indexes measure species diversity and are affected by species richness and community evenness in the sample community. In the cases of the same species richness, the higher the evenness of each species in the community, the higher the community diversity is. A larger Shannon index and smaller Simpson index indicate that the species diversity of the sample is higher [35].

β-diversity measures variations in the microbial taxa between/among samples (see Section 2.7 for more details).

2.7. Statistical Analyses

The effects of diet on E. furcellata most fitness parameters, Illumina sequencing data, and α-diversity indices were analyzed using the PROC GLM procedure (SAS Institute Inc., Cary, NC, USA). Multiple mean comparisons were adjusted by Duncan’s method. The egg hatching rates were analyzed by a Chi-squared test. Multiple mean comparisons were adjusted using Bonferroni’s method if the overall null hypothesis that no significant diet effect was rejected at α = 0.05. A χ² test was conducted for the sex ratio.

The linear discriminant analysis effect size (LEfSe) algorithm [36] was used to test differences in the abundance of bacterial families among the three treatments. Principal component analysis (PCA) and a heatmap were used to characterize the β-diversity in the current study. An LDA score > 3.0 was considered to be significant by default.

The number of OTUs from the 12 samples (four per diet type) were subjected to principal component analysis (PCA). A correlation matrix was used in the PCA. The heatmap determined the relative abundance of the top 20 bacterial genera present in the samples and was generated by the function heatmap. LEfSe, PCA and heatmap were carried out by Beijing Biomarker Technologies (Shunyi District, Beijing, China).

3. Results

3.1. Performance of E. furcellata Fed with Three Diets for Three Successive Generations

The diets significantly affected development times, regardless of the nymphal stages or the total time from the egg to the 5th instar (ANOVA: 2nd instar: F_2,251 = 87.06, p < 0.001; 3rd instar: F_2,182 = 441.27, p < 0.001; 4th instar: F_2,123 = 158.55, p < 0.001; 5th instar: F_2,71 = 27.88, p < 0.001; Egg to 5th: F_2,71 = 214.10, p < 0.001;

Table 1. Developmental times (mean ± 1SE d) of the F3 generation E. furcellata fed three different diets.

Diet	2nd Instar	3rd Instar	4th Instar	5th Instar	Immature *
Artificial diet	4.27 ± 0.15 a	6.32 ± 0.19 a	7.96 ± 0.49 a	9.33 ± 0.80 a	36.33 ± 0.86 a
Spodoptera litura larvae	2.75 ± 0.04 c	2.58 ± 0.07 c	3.22 ± 0.12 b	6.48 ± 0.16 b	24.96 ± 0.21 b
Tenebrio molitor pupae	3.33 ± 0.50 b	3.00 ± 0.00 b	3.64 ± 0.06 b	5.11 ± 0.27 c	24.08 ± 0.29 b

Means followed by different lower-case letters for the same E. furcellata larval stage denote significant difference between diets; multiple mean comparisons were adjusted by Duncan’s method. * Duration from egg to 5th instar.

). All stages reared on the artificial diet took significantly longer to develop to the next stage than those fed S. litura larvae or T. molitor pupae. The differences in the developmental time between S. litura larvae- or T. molitor pupae-fed E. furcellata were generally small (Table 1). The adult fitness of E. furcellata reared on the artificial diet was greatly and negatively affected, compared with those reared on the other two diets. The preoviposition period of E. furcellata raised on the artificial diet was significantly longer than those raised on the other two diets (ANOVA: Preoviposition period: F_2,17 = 15.56, p < 0.001;

Table 2. Fitness (mean ± 1SE) of E. fucellata F3 adults fed three different diets.

Diet	Preoviposition Period (d)	Female Oviposition (Eggs Female−1)	Female Longevity (d)	Egg Hatching Rate	♀:(♀ + ♂)
Artificial diet	42.5 ± 8.30 a	22.00 ± 7.04 b	61.67 ± 4.44 a	36.67% b	0.46 a
Spodoptera litura larvae	16.33 ± 1.44 b	154.09 ± 25.20 a	47.74 ± 3.27 b	100.00% a	0.37 a
Tenebrio molitor pupae	13.63 ± 2.46 b	115.57 ± 42.98 a	38.73 ± 2.39 b	85.71% a	0.47 a

Means followed by different lower-case letters for the same E. furcellata fitness denote significant difference between diets; multiple mean comparisons were adjusted by Bonferroni’s method.

). The longevity of E. furcellata females fed the artificial diet was longer than those reared on the other diets (ANOVA: Female longevity: F_2,87 = 11.81, p < 0.001; Table 2). The longer E. furcellata longevity from the artificial diet treatment; however, not only did not transform into a greater oviposition rate, but resulted in less than one-fifth of the oviposition rate (i.e., number of eggs female⁻¹) in comparison to E. furcellata fed upon the other two diets (ANOVA: Female oviposition: F_2,17 = 4.05, p = 0.036; Table 2). The hatching rate of eggs laid by E. furcellata females fed the artificial diet was greatly reduced in comparison to those laid by E. furcellata females fed the other diets (Chi-squared test: Egg hatching rate: ${\chi }_2^2$ = 11.30, p = 0.004; Table 2).

3.2. Sequencing of the 16S rRNA and Microbial Characterization

The 16S rRNA sequencing of the E. furcellata midgut samples (four per diet type) of 12 individuals and the data processing yielded a total of 882,500 effective tags. The paired-end (PE) reads and clean tags did not significantly differ between the three diets. The numbers of raw tags from E. furcellata females fed the T. molitor and S. litura were lower than those fed the artificial diet (ANOVA: Raw tags: F_2,17 = 7.60, p = 0.004;

Table 3. Summary of Illumina sequencing (mean ± 1SE).

Diet	Paired-End Reads	Raw Tags	Clean Tags	Effective Tags	Average Length (bp)	Effective (%) *
Artificial diet	79,792.75 ± 113.75 a	78,148.50 ± 158.42 a	75,922.00 ± 183.27 a	71,001.50 ± 335.83 b	448.75 ± 0.75 b	88.98 ± 0.29 b
Spodoptera litura larvae	79,904.25 ± 35.14 a	76,382.00 ± 302.24 b	76,161.00 ± 259.88 a	74,643.00 ± 64.98 a	453.00 ± 0.00 a	93.42 ± 0.08 a
Tenebrio molitor pupae	79,981.50 ± 139.71 a	76,196.50 ± 584.43 b	75,826.75 ± 465.19 a	74,980.50 ± 596.86 a	453.00 ± 0.00 a	93.75 ± 0.66 a

Means followed by different lower-case letters for the same E. furcellata fitness denote significant difference between diets. Multiple mean comparisons were adjusted by Duncan’s method. * Percentage of effective tags (out of paired-end reads).

). In contrast, the numbers of effective tags, the average gene length, and the effective tag percentages from E. furcellata fed the natural or semi-natural diets were greater than those from E. furcellata fed the artificial diet (ANOVA: Effective tags: F_2,17 = 30.86, p < 0.001; Average length: F_2,17 = 32.11, p < 0.001; Effective percentage: F_2,17 = 40.35, p < 0.001; Table 3). The tags generated a total of 271 OTUs at a 97% threshold. The number of OTUs generated from the midguts of E. furcellata females raised on the artificial diets was greater than those raised either on T. molitor pupae or S. litura larvae and the difference was statistically significant (ANOVA: OTU: F_2,9 = 39.85, p < 0.001; Figure 2). of the 271 OTUs, 118 were shared by all three diets, 74 were unique to the artificial diet, and seven were unique to the other two diets (Figure 3 and Appendix A (

Table A1. OTUs unique to midgut of female E. fucellata fed three different diets.

Phylum	Family	Genus	Species (OTU)
OTUs unique to Spodoptera litura larvae and Tenebrio molitor pupae diets (7)
Bacteroidetes	Prevotellaceae	Prevotellaceae NK3B31 group	uncultured species 1 (OTU60)
		Prevotellaceae UCG-003	uncultured species 2 (OTU141)
Proteobacteria	Sphingomonadaceae	Sphingobium	uncultured species 3 (OTU387)
	Burkholderiaceae	Burkholderia-Caballeronia-Paraburkholderia	Paraburkholderia tropica (OTU309)
	Enterobacteriaceae	Serratia	uncultured species 4 (OTU337)
Spirochaetes	Spirochaetaceae	Treponema 2	uncultured species 5 (OTU198)
unassigned	unassigned	unassigned	unassigned species 1 (OTU27)
OTUs unique to the artificial diet (74)
Acidobacteria	Solibacteraceae Subgroup 3	Bryobacter	uncultured species 6 (OTU120)
			uncultured species 7 (OTU121)
			uncultured species 8 (OTU152)
		Candidatus Solibacter	uncultured species 9 (OTU168)
			uncultured species 10 (OTU273)
			uncultured species 11 (OTU491)
		Paludibaculum	uncultured species 12 (OTU161)
			uncultured species 13 (OTU163)
	Holophagaceae	Holophaga	uncultured species 14 (OTU472)
Bacteroidetes	Bacteroidaceae	Bacteroides	uncultured species 15 (OTU26)
			uncultured species 16 (OTU79)
			uncultured species 17 (OTU135)
			uncultured species 18 (OTU528)
	Barnesiellaceae	unknown genus 1	uncultured species 19 (OTU43)
	Muribaculaceae	unknown genus 2	uncultured species 20 (OTU28)
			uncultured species 21 (OTU54)
			uncultured species 22 (OTU56)
			uncultured species 23 (OTU104)
			uncultured species 24 (OTU173)
			uncultured species 25 (OTU220)
			uncultured species 26 (OTU246)
			uncultured species 27 (OTU249)
	Prevotellaceae	Alloprevotella	uncultured species 28 (OTU29)
		Prevotellaceae NK3B31 group	uncultured species 29 (OTU289)
		Prevotellaceae UCG-001	uncultured species 30 (OTU86)
	Rikenellaceae	Alistipes	uncultured species 31 (OTU74)
	Tannerellaceae	Macellibacteroides	uncultured species 32 (OTU45)
			uncultured species 33 (OTU94)
			uncultured species 34 (OTU122)
Ignavibacteria	BSV26	unknown genus 3	uncultured species 35 (OTU127)
Firmicutes	Carnobacteriaceae	Carnobacterium	uncultured species 36 (OTU13)
	Streptococcaceae	Lactococcus	Lactococcus garvieae subsp. Garvieae (OTU5)
	Clostridiaceae 1	Clostridium sensu stricto 10	uncultured species 37 (OTU87)
	Eubacteriaceae	Acetobacterium	uncultured species 38 (OTU91)
		Eubacterium	bacterium NLAE-zl-P455 (OTU131)
	Lachnospiraceae	Blautia	uncultured species 39 (OTU50)
		Epulopiscium	uncultured species 40 (OTU97)
		Lachnospira	uncultured species 41 (OTU178)
		Lachnospiraceae NC2004 group	uncultured species 42 (OTU133)
		Lachnospiraceae NK4A136 group	uncultured species 43 (OTU495)
			uncultured species 44 (OTU541)
			uncultured species 45 (OTU589)
			uncultured species 46 (OTU606)
		unknown genus 4	uncultured species 47 (OTU137)
			uncultured species 48 (OTU232)
	Ruminococcaceae	Ruminiclostridium 9	uncultured species 49 (OTU155)
		Ruminococcaceae UCG-014	uncultured species 50 (OTU126)
		Ruminococcus 1	uncultured species 51 (OTU132)
		Subdoligranulum	uncultured species 52 (OTU449)
		unknown genus 5	uncultured species 53 (OTU71)
			uncultured species 54 (OTU281)
			uncultured species 55 (OTU488)
	Erysipelotrichaceae	[Anaerorhabdus] furcosa group	uncultured species 56 (OTU33)
		Allobaculum	uncultured species 57 (OTU73)
Fusobacteria	Fusobacteriaceae	Cetobacterium	uncultured species 58 (OTU225)
Gemmatimonadetes	Gemmatimonadaceae	Gemmatimonas	bacterium enrichment culture clone auto9_4W (OTU174)
		unknown genus 5	uncultured species 59 (OTU58)
			uncultured species 60 (OTU75)
			uncultured species 61 (OTU229)
Nitrospirae	Unknown family 1		uncultured species 62 (OTU226)
			uncultured species 63 (OTU578)
Patescibacteria	Unknown family 2		uncultured species 64 (OTU88)
Proteobacteria	Kiloniellaceae	Pelagibius	Pelagibius litoralis (OTU102)
	Unknown family 3		uncultured species 65 (OTU138)
	Desulfovibrionaceae	Bilophila	uncultured species 66 (OTU171)
	Unknown family 4		uncultured species 67 (OTU571)
	Nannocystaceae	Nannocystis	uncultured species 68 (OTU65)
	Polyangiaceae	Pajaroellobacter	uncultured species 69 (OTU244)
	Burkholderiaceae	Paenalcaligenes	uncultured species 70 (OTU426)
		Pandoraea	uncultured species 71 (OTU41)
	Rhodocyclaceae	Zoogloea	uncultured species 72 (OTU37)
	Enterobacteriaceae	Morganella	uncultured species 73 (OTU90)
	Pseudomonadaceae	Pseudomonas	uncultured species 74 (OTU19)
	Rhodanobacteraceae	Rhodanobacter	uncultured species 75 (OTU433)

)).

The midgut microbial distribution showed that Proteobacteria (Artificial diet: 57.12%; S. litura: 79.27%; T. molitor: 83.98%) and Firmicutes (Artificial diet: 34.57%; S. litura: 20.56%; T. molitor: 15.50%) were the two dominant bacterial phyla in the E. furcellata female midguts (Figure 4A), collectively constituting over 90% of the 16S rRNA gene sequences, regardless of diet type. The other phyla in the artificial diet treatment constituted 8.29%, as presented in Figure 4A, but less than 1% in the S. litura and T. molitor treatments (Gemmatimonadetes even absent in the S. litura treatment). Enterobacteriaceae (Artificial diet: 54.94%; S. litura: 79.17%; T. molitor: 83.83%) and Enterococcaceae (Artificial diet: 22.91%; S. litura: 20.07%; T. molitor: 15.07%) were the two most abundant families in the E. furcellata female midguts, but these two families constituted only about 77.85% in the artificial diet treatment (Figure 4B). Serratia (Artificial diet: 9.59%; S. litura: 58.34%; T. molitor: 47.39%), and an uncultured bacterium genus of family Enterobacteriaceae (Artificial diet: 44.64%; S. litura: 20.53%; T. molitor: 36.37%), and Enterococcus (Artificial diet: 22.91%; S. litura: 20.07%; T. molitor: 15.07%) were the three most abundant genera detected in the E. furcellata female midguts (Figure 4C). The percentages of Serratia from the artificial diet treatment were numerically over 30% lower. A unique genus in the artificial diet treatment was Lactococcus (2.7130%), which was not observed in the S. litura and T. molitor treatments. These results indicated that there were considerable differences in the midgut microbiota components between the artificial diet treatment and the others.

3.3. α- and β- Diversity of Midgut Microbiota

Among the four α-diversity indices, all except the Simpson index in the S. litura and T. molitor treatments were lower, compared with their corresponding indices in the artificial diet treatment (ANOVA: ACE: F_2,9 = 42.97, p < 0.001; Chao1: F_2,9 = 43.46; p < 0.001; Figure 2). The Simpson index in the T. molitor treatment was greater than that in the S. litura treatment, which was further greater than that in the artificial diet treatment (ANOVA: Simpson: F_2,9 = 20.47, p = 0.004; Shannon: F_2,9 = 59.86, p < 0.001; Figure 2). The results indicate that the artificial diet’s bacterial community was more diverse.

The PCA resulted in a clear separation into two groups (group 1: samples of artificial diet treatment, group 2: samples of S. litura and T. molitor treatments; Figure 5). The heat-map shows that the samples from the artificial diet were all clustered together and far away from the other samples, and there was greater relative richness of the genus Serratia in the S. litura and T. molitor treatments than that in the artificial diet treatment (Figure 6). The LEfSe showed no statistical difference between the S. litura and T. molitor treatments in any of the genera (data not shown). Comparing the S. litura and the artificial diet treatments, the LEfSe indicated that only two genera Staphylococcus and Proteus were enriched in the S. litura treatment. Conversely, 28 biomarkers had statistic difference in the artificial diet treatment (Figure 7). These results further illustrate that E. furcellata feeding on the artificial diet obviously had an affected gut bacterial community, whilst those feeding on a semi-natural diet did not.

4. Discussion

In the present study, the fitness parameters such as developmental times, the preoviposition period and the female oviposition of E. furcellata reared on the artificial diet, showed worse performance compared with those fed the natural diets (i.e., S. litura larvae) and the semi-natural diet (i.e., T. molitor pupae) (Table 1 and Table 2). Previous studies have shown that an animal diet may affect the physical and chemical properties of the intestinal environment [12], which may discourage or promote the growth of certain gut microorganisms. Different gut microbiotas possess different capacities in metabolizing or processing food and, subsequently, can impact the absorption of hosts [13]. The substantial differences in the developmental time and fitness between E. furcellata fed the artificial diet and the natural or semi-natural diet led us to investigate differences in the gut microbial community due to feeding upon different diets.

The results of the 16S rRNA sequencing and microbial characterization demonstrate that the midguts of the E. furcellata reared on the artificial diet had a richer bacterial community and exhibited greater α- and β- diversity than the midguts from the other two diets. Whether the richer bacterial community in the midgut of E. furcellata was brought about by the artificial diet needs to be further studied. The variations in the performance of E. furcellata fed the artificial diet might be related to the midgut bacterial community. In Acrosternum hilare Say (Hemiptera: Pentatomidae) and Murgantia histrionica (Hahn) (Hemiptera: Pentatomidae), lower survivorship and reproduction were linked to a loss of intestinal bacteria [37]. An imbalance diet in mirids bugs Adelphocoris suturalis Jakovlev (Hemiptera: Miridae) induced significantly higher mortality and increased the total gut bacterial load [38]. These studies have shown that diet-modified gut microbiota may promote changes in bugs’ fitness.

Irrespective of food type, the E. furcellata midgut microbes were dominated by the phyla Proteobacteria and Firmicutes (Figure 4A), which is consistent with other studies on insect gut bacteria [13]. Midguts from the artificial diet, however, had greater percentage of Bacteroidetes and Fusobacteria than the other two diets. Bacteria from both Bacteroidetes and Fusobacteria were found in the omnivorous American cockroach, Periplaneta americana L. (Blattodea: Blattidae), which is associated with plant polysaccharides degradation [39,40]. This was somewhat unexpected since none of the ingredients of the artificial diet had high levels of polysaccharide, although many ingredients, such as vegetable sap and honey, contains monosaccharide. At the family level, Enterobacteriaceae (in phylum Proteobacteria) and Enterococcaceae (in phylum Firmicutes) were the two dominant families and they constituted over 77%, regardless of diet type (Figure 4B). The bacterial families that accounted for a higher proportion in the artificial diet treatment than the other two diets are related to fermentation [41] and carbohydrate degradation [42]. The most abundant genus in the E. furcellata midguts from the natural and semi-natural diet treatments (over 45%) was Serratia, which is widely found in insects from orders such as Hymenoptera, Lepidoptera, Neuroptera, and Hemiptera [43]. Moreover, the most predominate species identified were an uncultured bacterium of Enterobacteriaceae, and an uncultured bacterium of Serratia, S. marcescens, and Enterococcus faecalis (data not shown). In these bacteria, two Serratia in the guts of the artificial diet had a lower abundance than the other two diets. Serratia was thought to be a harmless environmental bacterium until one of the most common species S. marcescens was discovered to be an opportunistic pathogen of many animals [44]. In early studies, strains of Serratia were isolated from ailing insects of Orthoptera, Coleoptera, Hymenoptera, Lepidoptera and Diptera, and the genera Serratia were found as insect pathogenic bacteria [45,46]. Another aspect is that many species in the genera Serratia are also symbionts of insects ranging from supplying hosts with vitamins and amino acids in aphids [47,48] to defense against parasitic wasps, high temperature stress and suppressing plant defenses [49,50,51]. In pea aphids Acyrthosiphon pisum Harris (Hemiptera: Aphididae), Serratia promoted the development and growth of its host through enhancing fatty acid biosynthesis [52]. Therefore, further research is needed to elucidate whether Serratia may have a similar role in the nutritional effects responsible for the slower growth and lower fitness of E. furcellata. The most abundant genus in the midguts of the E. furcellata feeding on the artificial diet was a bacterium in the family Enterobacteriaceae. Although this bacterium, on average, accounted for a greater percentage in the artificial diet treatment than the other two more natural diet treatments, the differences was small (Figure 4C). As such, this bacterium is probably a weak candidate for the observed differences in the growth and fitness of E. furcellata.

Of the genera in which the seven OTUs unique to S. litura and T. molitor diets were observed, Serratia were also present here (Appendix A). As mentioned above, Serratia seemed to be associated with E. furcellata growth and fitness. Another bacterium genus Burkholderia-Caballeronia-Paraburkholderia (Appendix A), is heavily involved in facilitating nutrient uptakes and eventually growth. Burkholderia-Caballeronia-Paraburkholderia also have been found in the gut of triatomines [53], which may be a kind of biological probiotic [54]. Burkholderia also increases the body size and developmental rate of Riptortus clavatus (Thunberg) (Hemiptera: Alydidae) [13]. In contrast, the other genera from the seven unique OTUs such as Sphingobium and Treponema (Appendix A), are linked to the metabolism of the plant cell wall by herbivorous insects [55,56,57]. Since E. furcellata is a predator and does not feed on cellulose-rich prey, the genus is likely transferred from its preys and may not significantly contribute to faster growth and greater fitness in E. furcellata reared on these two diets. Surprisingly, the E. furcellata reared on the artificial diet had 74 unique OTUs (Appendix A). Compared with the natural and semi-natural diet, the artificial diet might be an imbalanced diet for E. furcellata, but it exhibited greater α-diversity. Consistent with our present study, previous studies have shown that diets in nutritional richness may lead to a decrease in α-diversity, although changes in diet were also associated with increased species abundance [38,58,59].

In summary, E. furcellata raised on S. litura larvae and T. molitor pupae grew faster and had greater fitness in comparison to counterparts raised on the artificial diets. The microbial community from the midguts of the E. furcellata raised on S. litura larvae and T. molitor pupae, on the contrary, had lower diversity. Some midgut microbial species unique to the E. furcellata raised on S. litura larvae and T. molitor pupae reside in genera that are known to be involved in metabolism and the uptake of nutrients. For our present study, possible mechanisms for the lower performance and fitness of E. furcellata fed the artificial diet can be categorized into three possible pathways: (1) The artificial diet is an imbalanced diet that lacks the essential ingredients contained in the natural or semi-natural diet for development and reproduction, although the artificial diet could maintain a successive generation of E. furcellata. (2) Insect endosymbionts may compete with each other [60], so the artificial diet may lead to the expansion of non-dominant microbiota and interferes with the functions of dominant microbiota. (3) Some unfavorable bacteria (may be one or more from the 74 unique OTUs in the artificial diet treatment) may be introduced into the midgut by the artificial diet. Future research will be conducted to confirm the above pathways and elucidate the roles of specific microbiota.

5. Conclusions

In this study, the midgut bacterial community in E. furcellata was identified. The bacterial diversity and richness in artificial diet samples were significantly higher, even though they had worse fitness. We found that Proteobacteria and Firmicutes dominated the bacterial community in the midgut of E. furcellata. At the genus level, Serratia and Burkholderia-Caballeronia-Paraburkholderia might be associated with E. furcellata growth and fitness. To elucidate the roles of specific microbiota in the midguts of E. furcellata, further research is needed.