Microbial dynamics in rearing trials of Hermetia illucens larvae fed coffee silverskin and microalgae

In the present study, Hermetia illucens larvae were reared on a main rearing substrate composed of a coffee roasting byproduct (coffee silverskin, Cs) enriched with microalgae ( Schizochytrium limacinum or Isochrysis galbana ) at various substitution levels. The microbial diversity of the rearing substrates, larvae, and frass (excrement from the larvae mixed with the substrate residue) were studied by the combination of microbial culturing on various growth media and metataxonomic analysis (Illumina sequencing). High counts of total mesophilic aer- obes, bacterial spores, presumptive lactic acid bacteria, coagulase-positive cocci, and eumycetes were detected. Enterobacteriaceae counts were low in the rearing diets, whereas higher counts of this microbial family were observed in the larvae and frass. The microbiota of the rearing substrates was characterized by the presence of lactic acid bacteria, including the genera Lactobacillus , Leuconostoc and Weissella . The microbiota of the H. illucens larvae fed Cs was characterized by the dominance of Paenibacillus . H. illucens fed diets containing I. galbana were characterized by the presence of Enterococcus , Lysinibacillus , Morganella , and Paenibacillus , depending on the algae inclusion level, while H. illucens fed diets containing S. limacinum were characterized by high relative abundances of Brevundimonas, Enterococcus, Paracoccus , and Paenibacillus , depending on the algae inclusion level. Brevundimonas and Alcaligenes dominated in the frass from larvae fed I. galbana ; the predominance of Brevundimonas was also observed in the frass from larvae fed Schyzochitrium -enriched diets. Based on the results of the present study, an effect of algae nutrient bioactive substances (e.g. polysaccharides, high-unsaturated fatty acids, taurine, carotenoids) on the relative abundance of some of the bacterial taxa detected in larvae may be hypothesized, thus opening new intriguing perspectives for the control of the entomopathogenic species and foodborne human pathogens potentially occurring in edible insects. Further studies are needed to support this hypothesis. Finally, new information on the microbial diversity occurring in insect frass was also obtained.


Introduction
Since the introduction of Regulation (EU) 2015/2283 on novel foods and subsequent Regulation (EU) 2017/893 that allowed the use of insects as processed animal proteins (PAPs) in aquaculture, the exploitation of edible insects has been subjected to an increasing interest by the food and feed industry . To concretely exploit the potentialities of edible insects based on previous assessments performed by national authorities in Belgium (FASFC, 2014), the Netherlands (NVWA, 2014) and France (ANSES, 2015), the European Food Safety Authority (EFSA) published in 2015 an opinion with a list of insect species that can potentially be used as food and feed in the European Union (EU) (EFSA, 2015). Taking into account the abovementioned national risk assessments, as well as the EFSA opinion (EFSA, 2015), the following insect species were identified as those which fulfil the safety conditions for insect production for feed use: Hermetia illucens, Musca domestica, Tenebrio molitor, Alphitobius diaperinus, Acheta domesticus, Gryllodes sigillatus, and Gryllus assimilis (Regulation (EU) 2017/893).
In Western countries, the human consumption of insects represents a novelty; however, in Asia, Africa, Latin America and Australia, insects are commonly consumed as a part of the daily diet (Garofalo, Milanović, Cardinali, Aquilanti, Clementi & Osimani, 2019). Numerous published studies have demonstrated suitability of the use of insects as feed to produce a high-quality source of nutrients for livestock (Cutrignelli et al., 2018;Gasco, Biancarosa, & Liland, 2020).
Certain most reared insects, including T. molitor, A. diaperinus, A. domesticus, and H. illucens, have been demonstrated to provide high value protein intended for use as food or feed.
H. illucens is a species that can act as a bioreactor capable of extracting energy from the most diverse substrates, including animal, vegetable, and urban wastes (Allegretti, Schmidt & Talamini, 2017), thus leading to efficient insect biomass production (Zarantoniello et al., 2019). Therefore, the real earning perspective deriving from the exploitation of H. illucens has resulted in the establishment of numerous companies that have to face issues related to the optimization of the rearing cycles and the safety of the end product (De Smet, Wynants, Cos & Van Campenhout, 2018;Wynants et al., 2019).
The massive production of edible insects relies on the use of the opportune substrates that should allow a proper development of the reared insects to obtain the end products (e.g., larvae) with suitable nutritional features (e.g., high protein and fat content) and compliance with high chemical and microbiological safety standards.
A recent review by Garofalo et al. (2019) concluded that edible insects are the vectors of a number of microbial species that can be present in their gut or on the external cuticle. Currently, commensals, spoilage or even human pathogens have been detected in edible insects (Garofalo et al., 2019;Osimani & Aquilanti, 2021). The biodiversity and dynamics of the microbiota occurring during the rearing of the insects is strongly influenced by numerous intrinsic and extrinsic factors, including: i) the transfer of the microorganisms from mother to offspring; ii) the natural microbial contamination of the rearing substrate; and iii) the hygiene of the rearing conditions (Bruno et al., 2019;. Recent studies carried out on the microbiota of the H. illucens larvae reared on various substrates, including food waste, cooked rice, calf forage and the so-called Gainesville housefly diet (a substrate containing alfalfa meal, wheat bran, corn meal, and pelleted peanut hulls) have reported the occurrence of a core microbiota and some other taxa that are likely influenced and thus modulated by biotic and abiotic factors, including the rearing substrates (De Smet et al., 2018;Wynants et al., 2019). Therefore, to obtain the insect-based ingredients suitable for feed or food production, the knowledge of microbial dynamics occurring during rearing is a key factor for the implementation of proper preventive or control strategies for microbiological risk management.
It should be noted that the rearing cycle of the edible insects can be a source of microbiological contamination of the environment, including frass (excrement from the larvae mixed with the substrate residues) highly contaminated by microbiota of the insect origin. The microbiology of frass is poorly understood; however, recent scientific studies highlighted a potential of this rearing waste for use as a soil amendment as reviewed by Schmitt & de Vries (2020), thus emphasizing the need for additional comprehensive microbiological data on this specific substrate.
In the present study, the H. illucens larvae were reared on a main rearing substrate composed of a coffee roasting by-product (coffee silverskin) enriched with various proportions of microalgae (Schizochytrium limacinum or Isochrysis galbana) as the source of polyunsaturated fatty acids (PUFAs) (Truzzi, Giorgini, et al., 2020). In more detail, S. limacinum are heterotrophic marine traustochytrids rich in docosahexaenoic acid (DHA) (35 g/100 g of their total fatty acids), whereas I. galbana are haptophyte microalgae characterized by a high content of DHA, stearidonic acid and alpha-linolenic acid (Truzzi, Giorgini, et al., 2020).
The microbial dynamics of the rearing substrates, larvae and frass were then studied by the combination of microbial culture on various growth media and metataxonomic analysis (Illumina sequencing).

Insect feeding substrates preparation
The main component of the experimental insect feeding substrates consisted of a coffee by-product (coffee silverskin, Cs) obtained from the roasting process of green coffee beans at Saccaria Caffè Srl (Marina di Montemarciano, Italy). Once collected, Cs was processed as already described by Zarantoniello et al. (2020). Briefly, the coffee by-product was grinded using an Ariete 1769 food blender (De Longhi Appliances Srl, Italy) up until particles of 0.4 ± 2 mm in diameter were obtained.
S. limacinum or I. galbana, added at different inclusion percentage to Cs, were provided in a lyophilized form by AlghItaly Società Agricola S. R.L. (Sommacampagna, VR, Italy).
Ingredients (Cs and microalgae) were mixed, homogenized, and kept 24 h at 4 • C, before use. All the experimental insect feeding substrates were moistened with distilled water to reach a final moisture of ~70% (Makkar, Tran, Heuzé & Ankers, 2014). Aliquots from diets were sampled and stored at 4 • C until analysis.

Insect rearing
Six-day-old larvae, purchased from Smart Bugs (Ponzano Veneto, Italy), were counted and divided into groups of 150 larvae per replicate (n = 5, for each of the tested feed blends) amounting to a total of 6750 specimens.
The rearing of larvae on the different feed blends was carried out in lid-covered plastic boxes (28 × 19 × 14 cm), screened with fine-mesh cotton gauze, and provided with a single ventilation hole. The boxes were placed into a climatic chamber maintained at a temperature of 27 ± 1 • C, relative humidity of 65 ± 5%, and 24 h dark photoperiod (Zarantoniello et al., 2020). Larvae were collected as soon as they showed a change in tegument color from white to black, thus attesting that the prepupal stage was reached.
After collection, prepupae were stored at 4 • C until further analysis. Aliquots of frass from each treatment were also collected and stored at 4 • C up until analysis.

Viable counts
To estimate viable counts, 10 g of each sample (substrate, larvae and frass) was aseptically diluted in 90 mL of a sterile water solution of bacteriological peptone (1 g L − 1 ) and homogenized in a Stomacher 400 circulator apparatus (VWR International PBI, Milan, Italy). Ten-fold dilutions of the homogenates were then inoculated on suitable solid media for the counting of various microbial groups. Total mesophilic aerobes were counted in the standard plate count agar (PCA) (Oxoid, Basingstoke, UK) incubated at 30 • C for 48 h. Aerobic bacterial spores were counted after thermal treatment of the homogenates at 80 • C for 15 min followed by cooling in iced water for 5 min and plating in the standard plate count agar (PCA) (Oxoid,) incubated at 30 • C for 48 h. Presumptive lactic acid bacteria were counted on de Man, Rogosa and Sharp (MRS) agar (Oxoid) supplemented with cycloheximide (250 mg L − 1 ) and incubated at 37 • C for 72 h under anaerobic conditions. Coagulase-positive cocci were counted on the mannitol salt agar (MSA) (VWR Prolabo Chemicals, Leuven, Belgium) incubated at 37 • C for 24-48 h. Enterobacteriaceae were counted on violet red bile glucose agar (VRBGA) (VWR Prolabo Chemicals) incubated at 37 • C for 24 h.
For each of the rearing replicates (n = 5, for each of the feed blend) the analyses were carried out in duplicate and the data were expressed as the mean of log colony forming units (cfu) per gram ± standard deviation of independent duplicate biological experiments.

DNA extraction
To obtain DNA extracts from the feed blends, larvae and frass, aliquots (1.5 mL) of each homogenate (dilution 10 − 1 ) prepared as described in section 2.3 were centrifuged for 5 min at 16,000 g and the supernatants were discarded. Total microbial DNA was extracted from the cell pellets with an E.Z.N.A. soil DNA kit (Omega Bio-tek, Norcross, GA, USA) following the manufacturer's instructions.
A Nanodrop ND 1000 (Thermo Fisher Scientific, Wilmington, DE, USA) was used to measure the quantity and purity of the DNA extracts, which were then standardized to a concentration of 20 ng μL − 1 prior to further analysis. For each sample (diets, larvae, and frass), the DNA extracts obtained from each of the five rearing trials (replicates) were pooled to reduce the intersampling variability and were subjected to 16S rRNA gene sequencing.

Metataxonomic analysis
For each of the rearing trials (n = 5, for each of the feed blend), the extracted and pooled DNA standardized at 20 ng μL − 1 was used as a template in the PCR runs for the amplification of the V3-V4 region of the 16S rRNA gene according to the procedure described by Klindworth et al. (2013). Amplicons were purified and tagged according to the 16S metagenomic sequencing library preparation instruction from Illumina. The sequencing was performed using a MiSeq instrument (Illumina, San Diego, CA, USA) with V3 chemistry, which generated 250-bp paired-end reads, following the producer's instructions.
After sequencing, the reads were initially merged using FLASH v.1.2.11 (Magoč & Salzberg, 2011) with default parameters; chimera sequences were filtered using USEARCH v. 11 and then processed with QIIME1 (Caporaso et al., 2010) according to Ferrocino et al. (2017). Briefly, after quality filtering OTUs were clustered at 97% of similarity by using UCLUST clustering method, and representative sequences of each cluster were used to assign taxonomy using the Greengenes 16S rRNA gene database v2013, by means of the RDP Classifier, with a minimum confidence score of 0.80. The operational taxonomic unit (OTU) table was rarefied at the lowest number of sequences per sample (5000) by using QIIME1 and used to build the OTU tables. Sequences were manually blasted in order to confirm the taxonomic assignment. When genus level was not reached, OTUs belonging to family level were collapsed and identified as "Others". In the OTU tables, only OTUs with an incidence ≥1% in at least one sample were shown.
Sequencing data were deposited at the Sequence Read Archive of the National Center for Biotechnology Information (PRJNA643230).

Statistical analysis
Shapiro-Wilk test was used to test the normality of the data (Mishra et al., 2019). The Tukey-Kramer's honest significant difference (HSD) test (level of significance 0.05) was used to evaluate the differences within viable counts of the samples by one-way analysis of variance (ANOVA). Tests were carried out using the JMP software version 11.0.0 (SAS Institute Inc., Cary, NC).

Viable counts
The results of viable counting performed on the rearing substrates are reported in Table 1.
Counts of total mesophilic aerobes were generally high from 8.3 to 7.8 log cfu g − 1 and there were no significant differences between the samples (P > 0.05). The counts of bacterial spores varied from 3.1 to 4.8 log cfu g − 1 and the highest values were recorded in the samples containing I. galbana; the lowest values were observed in the case of Cs. Presumptive lactic acid bacteria counts ranged from 8.0 to 8.6 log cfu g − 1 and Cs showed the highest values; the samples containing I. galbana had the lowest counts. In the case of coagulase-positive cocci, the values were from 4.0 to 6.6 log cfu g − 1 ; the Cs samples showed the lowest values and the samples containing S. limacinum had the highest counts. In all samples, Enterobacteriaceae counts were <1 log cfu g − 1 . Finally, eumycetes counts ranged from 4.3 to 7.7 log cfu g − 1 ; the Cs samples showed the lowest values whereas the samples containing S. limacinum showed the highest values.
The results of viable counting performed on H. illucens larvae reared on various feed blends showed significant differences (P < 0.05) and are reported in Table 2.
In the case of total mesophilic aerobes, the counts were from 7.7 to 9.1 log cfu g − 1 ; the highest values were detected in the samples containing I. galbana and the lowest counts were in the samples containing S. limacinum. Bacterial spore counts ranged from 3.1 to 6.7 log cfu g − 1 ; the highest values were observed in the samples containing I. galbana and the lowest counts were in the Cs samples. The values of presumptive viable lactic acid bacteria were from 6.5 to 7.8 log cfu g − 1 ; the highest values were detected in the samples containing I. galbana. In the case of coagulase-positive cocci, viable counts ranged from 6.2 to 8.4 log cfu g − 1 , with the highest values detected in the samples containing I. galbana and the lowest counts detected in the samples containing S. limacinum. Enterobacteriaceae counts were from 2.5 to 4.9 log cfu g − 1 ; the Cs samples showed the highest values whereas the samples containing S. limacinum had the lowest values. Values from 5.4 to 7.4 log cfu g − 1 were detected in the case of eumycetes, with the highest values found in the samples containing I. galbana and the lowest in the Cs samples.
The results of viable counting performed on frass from H. illucens larvae reared on the different feed blends are reported in Table 3.
The counts of total mesophilic aerobes were comprised between 9.3 and 9.9 log cfu g − 1 , with no significant differences among samples. Bacterial spores ranged between 7.2 and 7.6 log cfu g − 1 , with no significant differences among samples. As for presumptive lactic acid bacteria, values from 7.6 to 8.2 log cfu g − 1 were detected, with no significant differences among samples (P > 0.05). Coagulase-positive cocci were comprised between 7.0 and 7.3 log cfu g − 1 , again with no significant differences among samples (P > 0.05). The counts of Enterobacteriaceae were comprised between 3.7 and 4.8 log cfu g − 1 , with the highest values for the samples containing I. galbana. Finally, eumycetes counts ranged from 6.3 to 6.8 log cfu g − 1 and there were no significant differences between the samples.
Listeria monocytogenes and Salmonella spp. were never detected in 25 g sample.

Microbiota of the rearing substrates
A total of 98,358 sequences were analyzed and the results showed that Lactobacillus, Leuconostoc and Weissella were stably detected in the basal diet (Cs), where they constitute the core microbiota (Fig. 1).
Diets with the inclusion of 5, 10, 20 or 25% of microalgae were clearly different compared to Cs (Fig. 1). Specifically, as I. galbana was added at 5, 10, 20, or 25% inclusion level, an increase in Weissella (57, 52, 62 and 62% of the relative abundance, respectively) was observed together with a reduction of Lactobacillus (from 33% of the Cs to about 2%). Leuconostoc was rarely observed (below 1%). The microbiota of the samples CsS5, CsS10 and CsS20 containing S. limacinum was characterized by the presence of Leuconostoc (with a relative abundance of 50, 28 and 57%, respectively) and Lactobacillus (11, 13 and 6%, respectively). Moreover, Weisella was stably detected in all samples (with a relative abundance of approximately 20%), reaching the maximum level on sample CsS25 (Fig. 1).

Microbiota of H. illucens larvae
The results of the 16S rRNA gene sequencing using 60,704 reads ( Fig. 2) showed the presence of Paenibacillus (22% of the relative abundance) and Lysinibacillus (9%) in larvae fed Cs (HCs). In addition, Solibacillus and Brevundimonas were found at 7 and 4%, respectively, in the same samples.
A separation of the samples based on the microbiota composition emerged from the heat plot (Fig. 3). In more detail, the inclusion of S. limacinum was associated with the presence of Bacillus, Brevundimonas, Devosia, Paenibacillus, and Paracoccus, whereas the inclusion of I. galbana was associated with the presence of Clostridium, Alcaligens, Lysinibacillus, and Morganella.

Discussion
H. illucens can use a number of organic materials for its growth; however, not all substrates are able to assure the proper provision of the nutrients for the larval development (Wang & Shelomi, 2017). It is known that similar to other edible insects, the microbiological quality of H. illucens larvae strictly depends on the contamination of the rearing substrate and the rearing environment. It should be noted that H. illucens larvae are able to reduce the microbial load of the substrates containing Gram-negative bacteria of the Enterobacteriaceae family (Wang & Shelomi, 2017). Hence, the study of the microbial dynamics during the rearing of this edible insect is crucial for a proper selection of the most suitable rearing substrate and the implementation of the proper preventive measures for microbiological risk management.
According to the microbiological analyses performed on Cs, high counts were detected for all microbial groups assessed, except Enterobacteriaceae. The data for total mesophilic aerobes, bacterial spores and presumptive lactic acid bacteria are in agreement with the results reported by Osimani et al. (2019) in the same rearing substrate. Cs is the tegument of green coffee beans and a byproduct of the roasting process. This substrate is a natural source of numerous bioactive compounds, such as chlorogenic acid, caffeic acid, melanoidins and dietary fiber, that can modulate the growth of microorganisms (Fernandez-Gomez et al., 2016;Truzzi, Giorgini, et al., 2020). The microbiological quality of Cs can vary depending on the roasting process and storage conditions before disposal. In the present study, viable counts were detected in the feed blends containing various levels of either I. galbana or S. limacinum and were generally comparable with the counts detected in the unblended Cs.
As a general trend, larvae were characterized by higher counts than those detected in the feed blends. Loads of total mesophilic aerobes, bacterial spores, lactic acid bacteria and eumycetes were in agreement with the loads reported by Wynants et al. (2019) for H. illucens larvae reared on organic waste streams. In detail, Wynants et al. (2019) reported values up to 9.0 log cfu g − 1 for total mesophilic aerobes, whereas Fig. 3. Heat plot of the abundances of OTUs and samples (upper red bar). Rows are clustered by means of Ward linkage hierarchical clustering methods. The intensity of the colors represents the degree of correlation between the samples and OTUs with dark red positive association and dark blue negative association. HCs, larvae fed with 100% coffee silverskin; HCsI5, larvae fed with 95% Cs plus 5% Isochrysis galbana, HCsI10, larvae fed with 90% Cs plus 10% I. galbana; HCsI20, larvae fed with 80% Cs plus 20% I. galbana; HCsI25, larvae fed with 75% Cs plus 25% I. galbana; HCsS5, larvae fed with 95% Cs plus 5% Schizochytrium limacinum; HCsS10, larvae fed with 90% Cs plus 10% S. limacinum; HCsS20, larvae fed with 80% Cs plus 20% S. limacinum; HCsS25, larvae fed with 75% Cs plus 25% S. limacinum. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) values up to 7.5 log cfu g − 1 were reported by the same authors for bacterial spores; in the case of lactic acid bacteria and eumycetes, values up to 8.5 and 6.8 log cfu g − 1 were reported by Wynants et al. (2019), respectively.
To the best of our knowledge, there is a paucity of the available data regarding microbial counts in the insect frass. Notwithstanding, the data obtained in the present study are in agreement with a previous study by Wynants et al. (2019) that indicated that in frass produced during the rearing of H. illucens larvae, detected counts of total mesophilic aerobes were up to 10.0 log cfu g − 1 . Moreover, Wynants et al. (2019) reported bacterial spore counts up to 7.0 log cfu g − 1 ; in the case of lactic acid bacteria and eumycetes, the counts were up to 9.8 and 7.8 log cfu g − 1 , respectively.
To the best of our knowledge, no previous reports on the counts of coagulase-positive cocci in the H. illucens rearing chain are available in the scientific literature, and thus, the data cannot be compared.
The results of the metataxonomic approach identified the major and minor taxa.
With regard to the rearing diets, lactic acid bacteria were massively present in the coffee byproduct, thus confirming the results of viable counting. Pothakos et al. (2020) have recently reported that Lactococcus and Enterococcus together with Weissella, Leuconostoc and Lactobacillus constitute the dominant lactic acid bacteria genera in the last phase of the coffee mucilage fermentation, thus explaining the occurrence of the latter three genera in the Cs used as a rearing substrate for H. illucens larvae. Lactic acid bacteria are known to be responsible for coffee fermentation and for the production of organic acids, esters and alcohols that directly influence the sensory properties of coffee (Martins et al., 2020).
With regard to the microbiota of the H. illucens larvae, a complex microbiota was detected and hence only the more relevant taxa will be discussed. Interestingly, only a few bacterial taxa previously detected in the feed blends were found to be dominant OTUs in the larvae, thus suggesting that a stable resident gut microbiota is present in the larvae irrespective of their diet. In addition, a selective effect of microalgae inclusion in the rearing substrates on the composition of the larvae microbiota was observed. Regarding the latter aspect, Zheng, Chen & Cheong (2020) have recently reported that microalgae are a potential source of polysaccharides that can selectively enhance or reduce the viability of the microorganisms.
In detail, Paenibacillus was detected at variable abundances in the larvae, except for the larvae that were reared on the diet with the highest microalgae inclusion, irrespective of the type of algae. The genus Paenibacillus encompasses over 150 species of spore-forming and plant growth-promoting rhizobacteria (Eastman, Heinrichs & Yuan, 2014). In edible insects, Paenibacillus has been reported as an entomopathogen of the coconut rhinoceros beetle (Gahukar, 2020). The same genus has also been reported as a causative agent of a lethal intestinal infection of honeybee (Apis mellifera) (Morrissey et al., 2015). Currently, Paenibacillus has been detected in numerous edible insects, including giant water bugs, black ants, winged termites, rhino beetles, mole crickets and silkworm pupae . Members of the Paenibacillus genus are known to inhibit other bacteria and even fungi due to their production of biologically active substances, such as antibiotics, bacteriocins and/or small active peptides (Singh, Ghodke & Chhatpar, 2009). Given the latter property, these microorganisms might have contributed to the modulation of the viable gut microbiota of the analyzed H. illucens larvae. In the present study, Paenibacillus was detected at the lowest relative abundance in the samples of larvae fed on the substrates with the highest substitution level of either I. galbana or S. limacinum, thus suggesting a negative effect of the algae on the occurrence of this bacterial genus. Fig. 4. Incidence of the major taxonomic groups detected by sequencing of frass samples. FCs, frass from larvae fed with 100% coffee silverskin; FCsI5, frass from larvae fed with 95% Cs plus 5% Isochrysis galbana, FCsI10, frass from larvae fed with 90% Cs plus 10% I. galbana; FCsI20, frass from larvae fed with 80% Cs plus 20% I. galbana; FCsI25, frass from larvae fed with 75% Cs plus 25% I. galbana; FCsS5, frass from larvae fed with 95% Cs plus 5% Schizochytrium limacinum; FCsS10, frass from larvae fed with 90% Cs plus 10% S. limacinum; FCsS20, frass from larvae fed with 80% Cs plus 20% S. limacinum; FCsS25, frass from larvae fed with 75% Cs plus 25% S. limacinum. Only OTUs with an incidence ≥1% in at least one sample are shown. The OTUs with relative abundance below 1% are grouped together in "Other". (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) With regard to the presence of Lysinibacillus in larvae fed 100% Cs, members of this genus can have pathogenic activity in mosquitoes, harlequin fly larvae, German cockroach, common cutworm and water scorpions (Berry, 2012). Moreover, Lysinibacillus has already been detected in a number of species of edible insects, thus confirming its adaptation to the gut of these organisms (Garofalo et al., 2019;Grabowski & Klein, 2017). In the present study, Lysinibacillus was detected at the highest relative abundance in larvae fed the substrate with the highest substitution level of I. galbana, whereas the lowest abundance was observed in larvae fed the highest substitution level of S. limacinum. It should be noted that the absence of Lysinibacillus in all substrates used for the rearing of the larvae suggests that occurrence of this genus in the gut of the H. illucens larvae can be ascribed to a direct contamination of the eggs from the mother to the offspring during egg-laying. Moreover, no effects of this microorganism on the viability of larvae were observed, thus suggesting the absence of any pathogenic activity of Lysinibacillus in this insect species.
The overall differences in the occurrence of Lysinibacillus and Paenibacillus in larvae fed various diets suggest an influence of the two tested algae on these two bacterial genera. These preliminary results suggest a need of further investigations on the inclusion of microalgae in the substrates for rearing of edible insects. To the best of our knowledge, no previous studies on the effect of algae on the microbiota of edible insects are available in the scientific literature and the data cannot be compared.
Moreover, an increase in the inclusion of algae resulted in a reduction in the relative abundance of Solibacillus detected in the larvae microbiota. Solibacillus belongs to the Planococcaceae family; this genus of spore-forming bacteria is generally of environmental origin (e.g., sediment of estuarine environment) (Markande, Acharya & Nerurkar, 2013). To the best of our knowledge, this is the first report of Solibacillus in edible insects. Interestingly, Wan, Zhao, Guo, Asraful, & Bai (2013) described a bioflocculanting activity of Solibacillus silvestris towards microalgae, thus suggesting that a potential interaction between the two algae included in the H. illucens diets and Solibacillus should be investigated in the future.
In the case of Brevundimonas, its occurrence in insects has already been described in scorpions and in both larvae and adults of lepidoptera (Grabowski et al., 2017;Garofalo et al., 2019). Interestingly, Brevundimonas was detected as one of the major OTUs in the mucosa-associated bacteria of rainbow trout (Oncorhynchus mykiss) fed a H. illucensenriched diet, thus confirming the strict association of this insect with H. illucens (Bruni, Pastorelli, Viti, Gasco & Parisi, 2018). In the present study, Brevundimonas was massively found in H. illucens fed S. limacinum, irrespective of the level of algae inclusion.
An increase in the inclusion of I. galbana was associated with an increase in the relative abundance of Morganella in the larvae, whereas no appreciable differences in the relative abundance of Morganella were detected in larvae fed S. limacinum-enriched diets. Morganella is a genus of the Enterobacteriaceae family, which includes Gram-negative rodshaped aerobic and facultatively anaerobic bacterial species. This genus is usually isolated from freshwater, soil and intestinal tracts of animals (Oh et al., 2020). Morganella has already been detected in insects, including flies, spiders (Grabowski & Klein, 2017), and edible insects, such as fresh and processed mealworm larvae and H. illucens larvae (Garofalo et al., 2019;Liu, Yao, Chapman, Su & Wang, 2020).
Finally, the abundance of Enterococcus was increased concomitantly with an increase in the inclusion levels of S. limacinum. The presence of Enterococcus in insects is common and has been documented by numerous reports on its detection in mealworms, moths, ants, butterflies, flies and even H. illucens larvae (Garofalo et al., 2019;Grabowski & Klein, 2017;Osimani et al., 2019). As reported by Zheng et al. (2020), algae polysaccharides may stimulate the growth of enterococci, thus explaining an increase in the relative abundance of this microbial group in larvae fed diets containing the highest inclusion level of S. limacinum.
Schmitt & de Vries (2020) have recently reported that frass produced during the rearing of H. illucens larvae can be used in a strained agricultural system. The environmental impact of application of this byproduct as a soil amendment is lower compared to the impact of existing products. However, the effectiveness of H. illucens frass varies depending on the diet used during the rearing of the insects and its physicochemical and microbiological features are thus variable (Schmitt & de Vries, 2020). With regard to the latter aspect, insect frass can undoubtedly represent a rich culture medium for saprophytic microorganisms and a vehicle for transmission of the bacterial plant pathogens . Moreover, the metabolic activity of the microorganisms in the frass can lead to the production of the molecules that can be perceived by other insects and influence their oviposition (Anbutsu & Togashi, 2002).
In the present study, an enrichment in some of the species detected in the larvae was observed in the corresponding frass. To the best of our knowledge, only a few published reports have described the microbiota of frass, thus making comparison of the data difficult.
As a general trend Alcaligenes had the highest abundance in the frass from larvae fed Cs enriched with I. galbana. Species of the Alcaligenes genus have a potential detoxifying activity that still requires a clarification; Alcaligenes have been isolated from cerambycid beetle (Lemoine, Engl & Kaltenpoth, 2020).
A decrease in the relative abundance of the Trichococcus genus was detected in the frass from larvae fed increasing levels of I. galbana. Trichococcus belongs to the Carnobacteriaceae family and to the best of our knowledge, there are no previous reports in the scientific literature on the occurrence of this genus in the rearing chain of edible insects; thus, the data cannot be further compared.
With regard to the influence of microalgae on the microbial growth, Han, Pang, Wen, You, Huang & Kulikouskaya (2020) have recently reported that a sulfated polysaccharide produced by Rhodophyta algae can have a positive effect on the growth of Bacteroidetes and Proteobacteria and lead to a reduction in Firmicutes and Actinobacteria in the human feces. Moreover, Shang et al. (2017) reported that fucoidan (a sulfated polysaccharide) from the Laminaria and Ascophyllum algae can stimulate the growth of Bacteroides and selectively modulate the growth of other taxa present in the gut of mice. Finally, laminarin (a polysaccharide produced by brown algae from glucose) can negatively impact the growth of Firmicutes, while positively influencing the multiplication of Bacteroides (Nguyen et al., 2016).
With regard to the algae used in the present study, Jin et al. (2020) reported that in a simulated in vitro digestion and colonic fermentation, the supplementation of microalgae, including Schizochytrium spp., leads to a significantly increase in the abundance of Bacteroides and Dialister involved in propionate metabolism. Moreover, Lyons, Turnbull, Dawson, & Crumlishm (2017) have reported a notable increase in the relative abundances of Streptococcus, Leuconostoc, Lactobacillus, Lactococcus and Weissella in the distal intestinal microbiome of rainbow trout fed a whole cell S. limacinum meal. With regard to I. galbana, Nuño, Villarruel-López, Puebla-Pérez, Romero-Velarde, Puebla-Mora & Ascencio (2013) observed an increase in the counts of intestinal lactic acid bacteria in experimental rats fed a diet including Isochrysis spp., thus suggesting a prebiotic effect of these microalgae. Moreover, members of the Isochrysis genus are known to produce fucoxanthin (Gullón, Gagaoua, Barba, Gullón, Zhang & Lorenzo, 2020), a marine carotenoid, which can significantly promote the growth of Bifidobacterium lactis (Hwang, Phan, Lu, Ngoc Hieu & Lin, 2016).

Conclusions
Numerous scientific studies reported that the rearing process together with the rearing substrates can positively or negatively influence the final microbiological quality of edible insects. Moreover, a fraction of microbiota of larvae and adults can be difficult to control, being directly transmitted from the mother to the offspring. A novel approach to the latter issue can be based on the use of the feeding substrates that can selectively modulate the prevalence of potentially pathogenic microbial taxa in the edible insects. Based on the results of the present study, an effect of algae nutrient bioactive substances (e.g. polysaccharides, high-unsaturated fatty acids, taurine, carotenoids) on the relative abundance of some of the bacterial taxa detected in larvae may be hypothesized, thus opening new intriguing perspectives for the control of the entomopathogenic species and foodborne human pathogens potentially occurring in edible insects. Further studies are needed to support this hypothesis. Finally, new information on the microbial dynamics occurring in insect frass was also obtained.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.