3.1. Changes in carbon and nitrogen
We measured the mass, moisture, temperature, and pH values of the residues to characterize the bioconversion process (Fig. S1), and the physical and chemical properties of inputs, residues, and larvae are shown in Table 1. The results showed that the degradation rate, temperature, and pH in the BSFL treatment (LT) were higher while the water content was lower, compared to the non-aeration compost (NC). The total carbon (TC) and total Kjeldahl nitrogen (KN) in the LT decreased by 13.9 g/kg and 2.76 g/kg, respectively, which were 13.9 times and 3.70 times higher than those in the NC (Fig. 1a, b). This is attributed to the lack of aeration in the NC, resulting in low microbial activity (Rincon et al. 2019). On the contrary, the peristalsis of the larvae and their own respiration ensure the oxygen supply in LT, allowing for rapid degradation of TC and KN.
Table 1
Physical and chemical properties of inputs, residues, and larvae
| Parameters | DBW Input | Fresh larvae | Residues (NC) | Residues (LT) |
| Mass (kg) | 2001 ± 4.42 | 398 ± 4.32 | 1797 ± 33.0 | 591.0 ± 11.5 |
| Moisture (%) | 72.3 ± 0.95 | 61.2 ± 1.70 | 70.4 ± 1.16 | 55.6 ± 3.14 |
| KN (g/kg) | 36.5 ± 2.55 | 62.2 ± 7.36 | 34.1 ± 1.35 | 21.3 ± 0.39 |
| TC (g/kg) | 471 ± 13.4 | 505 ± 41.6 | 458 ± 7.43 | 279 ± 16.4 |
| Ash (%) | 5.78 ± 1.42 | 2.45 ± 0.95 | 5.93 ± 1.78 | 20.9 ± 6.1 |
| Crude protein (%) | 5.61 ± 1.76 | 13.7 ± 3.1 | 4.98 ± 1.09 | 7.41 ± 1.69 |
| Crude fat (%) | 8.46 ± 0.85 | 11.5 ± 4.3 | 7.69 ± 0.23 | 1.48 ± 0.12 |
| Saltness (%) | 1.15 ± 0.65 | 1.45 ± 0.53 | 1.22 ± 0.82 | 3.09 ± 0.78 |
As two important processes, dynamic changes in ammonium (NH + 4-N) and nitrate (NO- 3-N) are crucial indicators of the nitrogen cycle, accurately reflecting the process from ammonification to nitrification/denitrification (Li et al. 2023). Due to the rapid degradation of organic nitrogen, the content of inorganic nitrogen in the residues rapidly increased. Inorganic nitrogen in the residues mainly existed as NH + 4-N and NO- 3-N, with NO- 3-N being the end-product of nitrification (Ge et al. 2018). Specifically, ammonia nitrogen and nitrate nitrogen in LT increased from 0.268 ± 0.021 g/kg and 40.9 ± 4.8 mg/kg, respectively, to 2.57 ± 0.32 g/kg and 150 ± 10 mg/kg, respectively (Fig. 1c, d). Compared with traditional composting, there was no phenomenon of NH + 4-N decreasing in the later stage (He et al. 2021). Excess NH + 4 was easily disposed of in the form of NH3 with increasing pH/temperature and the evaporation of water (Coskun et al. 2017). Compared to traditional composting, the highest temperature (< 55 ℃) and final pH (< 9) during larval treatment were lower (Fig. S1), which may reduce the conversion of NH + 4 into NH3 in the residues. In addition, nitrification was mainly dominated by mesophiles during food-manure composting (Wang et al. 2022). The temperature in the LT on days 0–4 was more conducive to nitrification, leading to a rapid increase in NO- 3-N (Fig. 1d).
3.2. Nutrient balances and greenhouse gas emissions
Based on the physicochemical properties of the materials and larvae (Table 1), we also calculate the flows of carbon and nitrogen in different treatments (Fig. 2). In the natural composting group, the reductions of carbon and nitrogen were entirely converted into gas emissions. During the two-week transformation, 6.68% of carbon and 12.6% of nitrogen were emitted in the form of gas (Fig. 2a and b), consistent with other studies (Parodi et al. 2021). The NC in this experiment was set for only 2 weeks, whereas traditional composting typically lasted 30 to 90 days (He et al. 2021; Li et al. 2023; Meng et al. 2022; van den Bergh et al. 2023). Literature suggested that carbon losses during food waste composting ranged from 44.9–91.8%, and nitrogen losses ranged from 26.2–53.8% (Zhang and Matsuto 2010). The carbon and nitrogen entering the residues in the LT were 28.0% and 33.1%, respectively, which were only 30.0% and 37.8% of those occurred in the NC. At the same time, the larvae recovered 29.9% of carbon and 55.0% of nitrogen internally (Fig. 2c and d). On one hand, BSFL recovered carbon and nitrogen through their metabolism and storage of organic matter, resulting in losses of only 42.1% and 11.9%, significantly lower than losses during traditional composting (Zhang and Matsuto 2010). This is likely a key reason why BSFL bioconversion technology reduced greenhouse gas emissions in organic waste treatment. On the other hand, BSFL bioconversion technology had the advantage of short cycles (Singh et al. 2023). In this experiment, the degradation efficiency of carbon and nitrogen in NC was 0.48% per day and 0.90% per day, while these were found as was 5.14% per day and 4.78% per day in LT, indicating a significantly higher degradation efficiency with BSFL bioconversion technology.
The CO2 and CH4 emission rates under all the treatments showed an upward trend, while the N2O emission rate showed a downward trend (Fig. 3a, b, and c). CO2 was the main gas produced during the bioconversion, much higher than CH4 and N2O emissions. CO2 emissions could reflect the metabolic activities of BSFL and microorganisms (Wang et al. 2018). BSFL increased CO2 emissions by 423% compared with incubating DBW without BSFL, indicating that the metabolic activity in the LT was significantly stronger than that in the NC. The emission of N2O in the LT was only 43.9% of that in the NC, although the degradation of KN in the LT was 5.33 times that in the NC. Denitrification, which is traditionally considered as the main source of N2O, is an anoxic process carried out by denitrifiers (Sanchez et al. 2015). Nevertheless, larvae movement and digestion facilitated air diffusion into the substrate, which destroyed the anaerobic regions and improved the process of ammonification and nitrification (Pang et al. 2020). This can be confirmed by the accumulation of ammonia nitrogen and nitrate nitrogen in the residues (Fig. 1c, d). Although the intervention of BSFL increased CO2 and CH4 emissions, it also accelerated the degradation of TC and KN (Fig. 1a and b). From the perspective of greenhouse gas emissions per unit TC/KN reduction, the LT produced 1.05–1.07 mg of CH4 and 0.312–0.317 g of CO2 for every 1g of TC degradation, which were only 13.1% and 38.1% of that in the NC, respectively (Fig. 3d). The reason for this phenomenon is that the BSFL recycled 29.9% of TC (Fig. 2c). Similarly, due to the 55.0% recovery of KN by BSFL, the emission of N2O in the LT was lower. In addition, for every 1 mg of methane produced, approximately 100 mg of carbon dioxide was simultaneously released in the NC, while approximately 310 mg of carbon dioxide was released in the LT (Fig. 3d). Therefore, the limited organic carbon tended to be converted into carbon dioxide rather than methane during BSFL bioconversion.
3.3. Changes in microbial community diversity
DBW with and without BSFL were mainly dominated by Actinobacteria (60.0%), Firmicutes (22.4%), Bacteroidetes (10.3%), and Proteobacteria (6.75%) (Fig. 4a). These phyla affect mineralization and the carbon/nitrogen cycle (Huang et al. 2019). Among them, Firmicutes were the most dominant bacteria in the NC (88.9–95.8%), consistent with traditional aerobic composting (He et al. 2021; Li et al. 2023). On the contrary, Firmicutes significantly decreased in the LT (3.67–19.3%) and LI (4.75–21.1%), while Actinobacteria became the most dominant bacteria (67.0-88.2% and 43.5–86.7%, respectively), consistent with our previous reporting (Jiang et al. 2019). In addition to Actinobacteriota, LT also contained a large amount of Proteobacteria (4.35–19.1%), while LI contained a large amount of Bacteroidetes (6.98–42.4%) (Fig. 4). Study showed that the percentage of N2O to total denitrification activity (N2 + N2O) was lowest in the region of the field with the highest relative abundance of Bacteroidetes (Graf et al. 2014). Therefore, the significant increase in the relative abundance of Bacteroidetes in the LT and LI, especially in the LI, may be a key reason for the decrease in N2O emissions.
Compared to the phyla level, the genus level provided a better understanding of the functional microorganisms during mature composting (Li et al. 2023). The emission of CH4 was influenced by the activity of methanogenic and methanotrophic communities (van den Bergh et al. 2023). According to the summary of the literature (Sun et al. 2022; van den Bergh et al. 2023; Zhang et al. 2023), all methanogens and methanotrophs detected in the current study are shown in Fig. 4b. The relative abundance of methanogens and methanotrophs detected was very low, which is consistent with the low methane emissions (Fig. 3). The dominant methanogens in the NC were Methanobrevibacter, Methanospaera, and Methanomassiliicoccus, which were significantly higher than those in the LI and LT. There was almost no detection of methanotrophs in the NC and LI, while the dominant methanotroph in the LT group was Methylophaga. This seems to indicate that the intervention of the BSFL made the microbial communities in DBW more inclined to oxidize methane rather than emit methane. Previous studies have generally considered most methanogens to be mesophilic, with optimal growth temperatures of 35–45 ℃ (Ren et al. 2016), while the temperature in the LT was more suitable for the metabolism of methanogens (Fig. S1). Therefore, it seems difficult to explain the differences in methane emissions between the NC and LT from the perspective of microbial communities. Based on previous research (Graf et al. 2014), the main denitrifying bacteria in the current study were identified (Fig. 4). Data showed that the conversion of NO- 2-N to NO catalyzed by nitrite reductase (Nir) was a key step of denitrification. The Nir enzyme was reported with two types: cytochrome cd1 containing nitrite reductase (NirS) encoded by the nirS gene and copper-containing nitrite reductase (NirK) encoded by the nirK gene (Chu et al. 2022). In the current study, nirK-type denitrifying bacteria were dominant (Marinobacter, Psychrobacter, Idiomarina, and Bifidobacterium), 16.9 times higher than nirS-type denitrifying bacteria (Fig. 4c). Obviously, BSFL not only increased the relative abundance of Marinobacter in the residues (LT), but also enriched Psychrobacter in its own intestines (LI). Respiratory NO reduction to N2O in the denitrification pathway was performed by two variants of the heme-copper oxidase type NO reductases in bacterial and archaeal denitrifiers, encoded by the nor genes (norB and norC) (Graf et al. 2014). Similarly, nor-associated taxa (Campylobacter, Psychrobacter, and Idiomarina) in the LT and LI were significantly higher than those in the NC. Based on the increase of nirK-associated and nor-associated taxa, it seems that BSFL would increase the emission of N2O during the DBW bioconversion. However, the final step of denitrification was catalyzed by the nitrous oxide reductase (encoded by the nosZ gene), which reduced nitrous oxide to nitrogen (Stein 2020). nosZ-associated taxa, Campylobacter, and Marinobacter, presented remarkably higher relative abundances in LT and LI, indicating that the intervention of BSFL promoted the entire denitrification process. Previous reports have shown a positive correlation between N2O flux and the ratio of (nirS + nirK + nor)/nosZ (Chen et al. 2023). Therefore, we calculated this ratio by replacing the relative abundance of genes with the relative abundance of associated taxa. The results showed that the ratios of (nirS + nirK + nor)/nosZ in the three groups were 3.13 (NC), 0.83 (LI), and 1.40 (LT), respectively. Apparently, the ratios of (nirS + nirK + nor)/nosZ in LI and LT were remarkably higher than that of NC, indicating a potentially higher N2O reduction degree in the LI and LT. Therefore, the occurrence of a large reduction of N2O may be an important factor leading to lower N2O emissions in the BSFL bioconversion system. In summary, the intervention of the BSFL reconstructed the microbial community structure during DBW bioconversion, thereby mitigating greenhouse gas emissions. On the one hand, the decrease in the ratio of methanogens to methanotrophs decreased methane emissions, while a higher proportion of nosZ-associated taxa reduced nitrous oxide emissions.
3.4. Pathways of nitrogen and methane metabolism
Subsequently, the mechanisms of BSFL on CH4 and N2O emission mitigation were investigated at the level of biotic factors. Firstly, we summarized the functional genes related to methanogenesis and carbon metabolism and calculated the ratio of methane synthesis to carbon metabolism (Fig. S5a). The ratio of methanogenesis to carbon metabolism in LT and LI was significantly lower than that in NC group. This confirms the previous conclusion that BSFL reduced the proportion of organic carbon converted to methane. Then, the CH4 net emission depended on the intensity of CH4 production and CH4 oxidation (Sun et al. 2022). Therefore, we focused on four pathways for biogenic methane production, namely methanol to methane, acetate to methane, carbon dioxide to methane, and methylamine/dimethylamine/trimethylamine to methane (Fig. 5a). These four pathways all required methyl-CoM reductase (EC: 2.8.4.1) to catalyze the production of CH4 from acetic acid, which is a direct potential cause of CH4 emissions. In addition, particulate methane monooxygenase, soluble methane monooxygenase, and methanol dehydrogenase, encoded by pmoA, mmoX, and mxaF (EC: 1.14.13.25, and 1.14.18.3), were the main drivers of methane oxidation (Sun et al. 2022). Therefore, we summarized the gene abundance corresponding to the enzymes involved in methane production and oxidation in three groups (Fig. 5c). Among the four pathways for methanogenesis, the gene abundance of the pathway for methanogenesis form acetate is the highest. However, almost no genes related to methyl-CoM reductase was detected in all samples, which is consistent with the previous conclusion of low methane production (Fig. 3a). In addition, the gene for methane monooxygenase was not detected. Therefore, from the perspective of functional genes, it cannot explain the reduction of methane emissions by BSFL. Microbial-mediated denitrification process was considered as the most important contributor to N2O production in bioconversion systems (Guo et al. 2020). During the reduction of nitrite, one part was reduced to ammonia (EC: 1.7.1.4, 1.7.7.1, 1.7.1.15, and 1.7.2.2) (He et al. 2021), while another part was first reduced to nitric oxide (EC: 1.72.1) then further reduced to nitrous oxide (EC: 1.7.2.5 and 1.7.1.14), and finally was reduced to nitrogen gas (EC: 1.7.2.4) (Graf et al. 2014) (Fig. 5b). Therefore, we further depicted the significant differences in the proportions of enzymes associated with nitrogen metabolism between the NC, LI, and LT (Fig. 5d). The abundance of nitrite reductase in the LI (EC: 1.7.7.2) and LT (EC: 1.7.1.15) was 5.69 times and 4.00 times higher than that in the NC, respectively. This suggests that microorganisms in the LI and LT were more inclined to reduce nitrite to ammonia. Meanwhile, the abundance of nitrous-oxide reductase (EC: 1.7.2.4) in LT and LI was significantly higher than that in NC (increased by 21.3 times and 5.13 times, respectively), which is likely one of the important reasons for the lower emissions of N2O in the LT (Fig. 3c). In addition, to further characterize the emission intensity of N2O, we calculated (nirK + nirS + nor)/nosZ (Chen et al. 2023). The ratio of (nirK + nirS + nor)/nosZ continuously decreased in all samples with the progress of bioconversion (Fig. S5b), which is consistent with the previous result of continuously decreasing emissions of N2O (Fig. 3c). Meanwhile, this ratio in the NC was similar to that in the LI and significantly higher than that in the LT, indicating that BSFL indeed reduced the emission of nitrous oxide by altering the abundance of functional genes in the material. The ratio of (nirK + nirS + nor)/ (nirA + nirB + nirD + nrfA) was used to characterize the ratio of nitrite reduction to N2O and ammonia (He et al. 2021). Except for day 2 and day 4, the ratio in LI was significantly lower than that in NC and LT (Fig. S5c), indicating that more nitrite was converted into ammonia in the BSFL intestines, thus reducing the emission of N2O. Therefore, the intervention of the BSFL directly reduced the emission of N2O by reducing the ratio of (nirK + nirS + nor)/nosZ in the material, and indirectly reduced the emission of N2O by increasing the ratio (nirK + nirS + nor)/ (nirA + nirB + nirD + nrfA) in the BSFL intestines. We constructed heatmaps by calculating the Spearman correlations between genes related to nitrogen reduction and denitrifying bacteria genera (Fig. 6). Marinobacter and Campylobacter were strongly positively correlated with the abundance of genes encoding EC: 1.7.2.4, indicating that these two genera further reduced nitrous oxide to nitrogen, consistent with previous researches (Graf et al. 2014). Therefore, correlation analysis further proved that the BSFL enriched bacteria (e.g. Marinobacter and Campylobacter) to reduce nitrous oxide emissions by reducing nitrous oxide to nitrogen.