A schematic diagram of an inbuilt MBBR in the household aquarium is shown in Figure 1. As shown in Figure 1A, the inbuilt MBBR was immersed in the aquarium. The device’s length, width and heigh were 50 cm, 28 cm, and 38 cm, respectively. The depth of water was 35 cm. There were 88 guppies with a body length of 2–4 cm in the aquarium. As illustrated in Figure 1B, the shell of the equipment is a transparent cylindrical plastic, including an upper layer and a lower layer. The upper layer is the microbial reaction zone with a diameter of 6 cm and a height of 20 cm and was filled with plastic carriers with a fraction of 50%. For esthetic reasons, the carriers came in four colors, including equal amounts of white (without dyestuff), red (with red dyestuff), green (with green dyestuff), and blue (with blue dyestuff). The carriers in the reactor were the type of K1 with a depth of 7 mm, a diameter of 9 mm, and a specific surface area of 500 m²/m³ (Water Environment Federation, 2010). It should be noted that the density of carriers was close to that of water (0.94–0.97 g/cm³). The air diffuser was also inside the reactor, providing air to the water, and it was connected to the aeration pump with a nominal voltage of 10 W through a hose. The carriers were held in suspension throughout the reactor by turbulent energy imparted by aeration. There were eight side rectangular holes evenly arranged on the bottom of the upper layer shell with a width of 5 mm and a length of 7 mm. There were also eight round holes evenly arranged on the top of the shell with a diameter of 8 mm. Under the action of gas lifting, water in the aquarium flowed into the reactor through the rectangular holes on the bottom side and then was discharged through the upper round holes with bubbles. There was a suction cup at the bottom of the equipment to attach it to the bottom of the aquarium. At the same time, the lower layer was filled with colorful stones to make the equipment more stable in the water. The inbuilt equipment in the aquarium seemed beautiful due to the colorful suspended carriers dancing, increasing the ornamental of the household aquarium.

At the beginning of the experimental operation, all the water in the aquarium was tap water after exposure to the sun for 2 h, and simultaneously the fish food was first fed with an amount of ~3.0 g. After that, it was provided once a week. The temperature of the water in the aquarium was always ~25°C. The composition analysis of the fish food is shown in Table 1. During the operating time, the water samples were taken every day. COD, NH₄⁺-N, NO₂⁻-N and NO₃⁻-N concentrations were tested using the standard methods (The State Environmental Protection Administration of China, 2022). The blank contrast aquarium was operated under the same condition as the experimental aquarium, except that there was no inbuilt equipment. The experimental aquarium was operated for 3 months without changing water.

After the experimental operation, the bio-carrier samples in the inbuilt MBBR were taken to a 1 L container filled with distilled water with a fraction of 28%. Ammonium chloride was added to the beaker until the NH₄⁺-N concentration of the water was 15 mg/L. Aeration was provided in the breaker to enable complete fluidization of the suspended bio-carriers and the dissolved oxygen (DO) concentration was above 6.0 mg/L. The water samples were taken at regular intervals, and NH₄⁺-N, NO₂⁻-N and NO₃⁻-N concentrations were also tested.

After the experimental operation, some bio-carriers that were still wet were scanned by the type microscope (OLYMPUS SZX10 microscope) to observe their physical features. The new carriers were monitored as blank groups.

At the same time, the biofilms attached to the carriers, which are called suspended biofilms hereafter, were observed by the scanning electron microscope (SEM; FEI Quanta FEG 250). The bio-carrier samples were cut to the appropriate size (about 5 mm × 5 mm), glued to the sample table with conductive tape, and tested after spraying gold. The acceleration voltage was adjusted to 8 kV for shooting, and the magnification was 2,000. The sediment in the aquarium was also scanned.

The cut bio-carrier samples were smoothed out, and the three-dimensional morphology was measured by laser scanning confocal microscopy (LSCM; Leica DCM8). And then, the shooting image was analyzed by the Leica Fotos App.

Four carrier samples of white, red, green, and blue were randomly taken after the experimental operation. Then, the biofilms attached to the carriers were scratched off, and the biofilms in the same color carriers were mixed evenly in one biofilm sample. For comparison, sediment samples were taken from three random places at the bottom of the aquarium. The seven samples were commissioned by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China) for genomic DNA extraction and sequencing according to the standard steps (Zhang et al., 2020). The hypervariable region V3–V4 of the bacterial 16S rRNA gene was amplified with the primer pairs presented in 338F and 806R (Zhou et al., 2022a). The raw data were submitted to NCBI and the accession number is PRJNA897105.

The steps of processing of sequencing data were also performed by the standard steps of Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China; Zhang et al., 2020). The taxonomy of each operational taxonomic unit (OTU) representative sequence was analyzed against the Silva138 database¹ using confidence threshold of 0.7. Most of the statistical analysis was performed using the online platform of Majorbio Cloud Platform² (Ren et al., 2022). The heatmap for the relative abundance of the top 30 genera was analyzed by the “pheatmap” package (1.0.8) of R Programming Language (Version 3.3.1). The beta diversity of bacterial communities (OTU level) was analyzed by using principal component analysis (PCA) after a centered log-ratio transformation by the R Programming Language (Version 4.3.2) based on the distance algorithm of Aitchison (Gloor et al., 2017). Simultaneously, the Analysis of Similarities (ANOSIM) was used to test the significance of the difference. The collinear network analysis was performed by Networkx (Version 1.11).

Phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt) was applied to analyze the potential functions (Zhao et al., 2022). Because of the renewal and elaboration of reference genome databases, the accuracy of PICRUSt2 has grown (Douglas et al., 2020); PICRUSt2 was therefore used in this study. The literatures introduced the detailed operating and analysis procedures (Baquiran et al., 2020; Li et al., 2022). The information on the functional genes for nitrogen removal, involving nitrification and denitrification, was predicted based on the map00910 of KEGG.³

As illustrated in Figure 2A, the COD and NH₄⁺-N concentrations of the blank contrast aquarium showed an upward trend over time due to the excretions of fishes and degradation of unconsumed feed (Hu et al., 2015; Zamora-Garcia et al., 2021). In addition, NO₂⁻-N and NO₃⁻-N concentrations were always below 0.6 mg/L, indicating less NH₄⁺-N inverting to NO₂⁻-N and NO₃⁻-N. Especially on the 16th day, the NH₄⁺-N concentration reached up to 1.985 mg/L with the NH₄⁺-N production rate of 0.12 g/m³/d, and the fish did not tolerate it and began to die. At this point, the blank contrast aquarium experiment stopped.

The COD concentration of the aquarium with the inbuilt MBBR showed an upward trend over time during the first 5 days. Then it did not change with time, ranging from 0–25 mg/L (Figure 2B). At the same time, most carriers initially floating on the upper inside of the shell began to be suspended, indicating that microorganisms began accumulating after 5 days in the aquarium. Like COD concentration, the NH₄⁺-N concentration also showed an upward trend over time during the first 5 days, began to decrease on the 6th day, and then reached 0.440 mg/L on the 9th day because of the fish food dosing. This concentration was also the maximum NH₄⁺-N concentration during the experiment, less than a quarter of that in the blank contrast aquarium. And after that day, the NH₄⁺-N concentration gradually decreased over time and went down to not be detected (<0.05 mg/L) on the 16th day. Then until the end of the experiment on the 90th day, the NH₄⁺-N concentration was always below 0.05 mg/L, which could not threat the fish (Yildiz et al., 2017). It maybe because that NH₄⁺-N was removed by ammonia oxidizing bacteria in the biofilms (Yildiz et al., 2017). NO₂⁻-N concentration in this aquarium was always below 0.05 mg/L, indicating that NO₂⁻-N maybe converted to NO₃⁻-N by nitrate oxidizing bacteria (Yildiz et al., 2017). However, the NO₃⁻-N concentration did not increase gradually, but fluctuated in the 0–4.5 mg/L range. There may be simultaneous nitrification and denitrification (SND) in the biofilms (Pan et al., 2022). Numerous studies have shown that MBBR was a well-established technology for SND, because the biofilms of a certain thickness, which provide a suitable micro environment (Liu et al., 2020; Bhattacharya and Mazumder, 2021). In conclusion, the inbuilt MBBR with suspended biofilms could be used for the nitrogen removal of household aquarium and performed well.

To explore the biofilm performance, NH₄⁺-N, NO₂⁻-N, and NO₃⁻-N concentration variations were tested, and the results are shown in Figure 3. It was linearly fitted and calculated that the specific NH₄⁺-N removal rate reached 45.43 g/m³/d based on the bio-carrier’s volume. According to the NH₄⁺-N production rate of 0.12 g/m³/d, only 0.27% of suspended biofilms were needed in the aquarium. Given this, the inbuilt MBBR could be very small. Based on the bio-carrier’s specific surface area, the specific removal rate was 0.09 g/m²/d, which was lower than that applied in municipal wastewater (Zhou et al., 2022b)—the reason may be the limitation of the influent substrate (Han et al., 2022). During the test period, there was no accumulation of NO₂⁻-N in the reactor, indicating that the nitrification was carried out thoroughly. At the same time, the NO₃⁻-N concentration increased correspondingly due to the transformation of NH₄⁺-N to NO₃⁻-N by nitrification. Unlike the experimental operation in the aquarium, there was almost no NO₃⁻-N removal in this reactor. The reason may be denitrification was restrained by the high DO concentration (Ren et al., 2022) or that the retention time of this test was shorter. In conclusion, the suspended biofilms performed a strong nitrification capacity, which could protect the fish from the NH₄⁺-N.

The biofilm morphology in the aquarium was shown in Figure 4. Compared to the new carriers (Figure 4A), the biofilms attached to the carriers were clearly visible (Figure 4B). The biofilms presented a certain thickness, which was benefited for SND in MBBR (Bhattacharya and Mazumder, 2021). It was found that various forms of microorganisms and extracellular polymers were distributed in the suspended biofilms and the sediment by SEM magnification of 2,000 times (Figures 4C,D). The suspended biofilms and the sediment in the aquarium exhibited a considerable diversity of microorganisms, including coccus, bacillus, and hyphomycetes. Under the wash of oxygen, the suspended biofilms appeared sparse with many filamentous structures (Figure 4C), while the sediment in the aquarium was relatively dense (Figure 4D). The structure of the suspended biofilms is more beneficial to the transfer of substrate in microorganisms. There was little difference in the biofilm morphology among different colors of bio-carriers (Figure 4C). The construction of the inbuilt MBBR in the aquarium could help the suspended biofilm obtain oxygen and achieve a structure conducive to the transfer of substrate in microorganisms under the scouring of oxygen. According to the three-dimensional morphology (Figure 4E), the suspended biofilms attached to white, red, green, and blue carriers all showed a certain degree of roughness with the Ra values of 0.490, 0.809, 0.800, and 0.461 μm and the Rz values of 2.25, 5.59, 4.08, and 3.03 μm, indicating the formation of mature biofilms and efficient mass transfer of nutrients to biofilm (Wu et al., 2023). It could be seen that the suspended biofilms successfully cultured microorganisms in the household aquarium.

The bacterial communities of the suspended biofilms and sediment at the genus level are shown in Figure 5. As shown in the heatmap of relative abundance for the top 30 genera (Figure 5A), the seven samples were clustered into two groups, which were suspended biofilm samples and sediment samples. The PCA showed the same result, when PC1 and PC2 explained 82.9% and 6.3% variation, respectively (Figure 5B). The ANOSIM further confirmed that the grouping was statistically significant (R = 1.000, p < 0.05). In addition, there were 25 and 17 genera abundant (sequence number > 100) only in suspended biofilms and only in the sediment, respectively (Figure 5C). However, there were only 13 genera abundant (sequence number > 100) in both suspended biofilms and sediment. The above confirmed that the bacterial communities of suspended biofilms and sediment in the aquarium were quite differed.

As shown in Figure 5A, the most abundant genus in the aquarium was norank_f__Caldilineaceae, which belongs to the family Caldilineaceae. Caldilineaceae was reported to be highly related to domestic wastewater treatment system (Zhang et al., 2017) and was the critical active denitrifier in nitrogen removal (Cao et al., 2020). Correspondingly, denitrification occurred in the aquarium during the experimental operation. However, the average sequence number (relative abundance) of norank_f__Caldilineaceae in the biofilms and sediment was 5,706 (24.2%) and 434 (1.8%), respectively. The inbuilt MBBR enriched the genus norank_f__Caldilineaceae, which would benefit to denitrification in the aquarium. In addition to that, among the top 30 genus in the aquarium, the certain genera Pseudofulvimonas (Lu et al., 2023), Gemmobacter (Tan et al., 2020), Runella (Zhao et al., 2017), Rhodobacter (James et al., 2023), Flavobacterium (Fan et al., 2021), Fusibacter (Han et al., 2021), Reyranella, Mycobacterium (Jiao et al., 2022), Hyphomicrobium (Fan et al., 2021), Pseudoxanthomonas (Jian et al., 2023), Novosphingobium (Sun et al., 2022), Gordonia (Tran et al., 2022), and Methyloversatilis (Zhai et al., 2020) were all reported to be denitrifiers. Pseudofulvimonas, Novosphingobium, and Methyloversatilis were abundant (sequence number > 100) only in the suspended biofilms, while Flavobacterium, Fusibacter, and Pseudoxanthomonas were abundant only in the sediment (sequence number > 100; Figure 5C). In conclusion, there were a variety of denitrifiers in the aquarium, and the communities of the suspended biofilms and sediment were different. It explained the phenomenon of SND in the aquarium from a microscopic perspective.

The second most abundant genus was Azospira, which fixed nitrogen (Tan et al., 2003). This genus was abundant in both biofilms and sediment (sequence number > 100). However, the average sequence number (relative abundance) of Azospira in the biofilms and sediment was 108 (0.5%) and 4,111 (17.4%), respectively. Therefore, nitrogen fixation may occur in the aquarium. The third most abundant genus was norank_o__PeM15, which may come from the animal gut (Zhu et al., 2022). It is worth noting that nitrifying bacteria Nitrospira appeared in the aquarium, which was reported to be capable of complete ammonia oxidation to nitrate (Daims et al., 2015). Nitrospira was abundant (sequence number > 100) only in suspended biofilms, and the average sequence number (relative abundance) was 335 (1.4%). In sediment, the average sequence number of Nitrospira was only 1. The inbuilt MBBR with suspended biofilms enriched Nitrospira, which brought the function of nitrification to the aquarium. Correspondingly, nitrification occurred in the aquarium with the inbuilt MBBR but not in the blank contrast aquarium.

The potential functions for nitrogen removal, including nitrification and denitrification, of the suspended biofilms and sediment in the aquarium are shown in Figure 6.

For the enzyme involved in denitrification, the nitrite oxidoreductase [EC:1.7.99.4] converts nitrate and nitrite into each other, and ferredoxin--nitrate reductase [EC:1.7.7.2] converts nitrate to nitrite (Zhou et al., 2022a). The average predicted sequence numbers of these two enzymes in suspended biofilms were 12,857 and 1,504, while the numbers in the sediment were 18,767 and 599. Nitrite reductase (NO-forming) [EC:1.7.2.1], nitric-oxide reductase (cytochrome c) [EC:1.7.2.5], and nitrous-oxide reductase [EC:1.7.2.4] converted nitrate to nitric oxide, then to nitrous oxide, then finally to gas nitrogen, respectively (Zhou et al., 2022a). The average predicted sequence numbers of these three enzymes in suspended biofilms were 1,675, 954, and 6,536, while the values in the sediment were 8,509, 6,421, and 5,152. So, there were certain sequences for these enzymes involved in denitrification in both suspended biofilm and sediment. It indicated that the suspended biofilms and sediment both had potentials in functions for denitrification. These conversion processes involved in denitrification are not restrictive steps for nitrogen removal in the aquarium.

After the fish ate, their excretion produced ammonia nitrogen, which harmed the fish. For the enzyme involved in nitrification, the ammonia nitrogen was converted to hydroxylamine by ammonia monooxygenase [1.14.99.39] and then to nitrate by hydroxylamine dehydrogenase [EC:1.7.2.6] in the aquarium (Zhou et al., 2022a). The average predicted sequence numbers of ammonia monooxygenase [1.14.99.39] in the suspended biofilms was 220, while it was only 5 in the sediment. The values of hydroxylamine dehydrogenase [EC:1.7.2.6] were 221 in the suspended biofilms, but it was only 1 in the sediment. It indicated that the suspended biofilms have the potential for the nitrification function. Therefore, the inbuilt MBBR played an essential role in ammonia nitrogen removal in the aquarium. Building an inbuilt MBBR in the household aquarium is feasible to protect the fish from nitrogen.