A total of 600 1-day-old male small-sized meat ducks (H strain) were procured from Ecolovo Group, China. All the ducks were raised on the floor (15 birds/m²) during the first 3 weeks. Ducks with the highest and lowest BWs and those that died or had leg problems were excluded at 21 days of age (n = 100). The remaining 500 birds, with similar initial body weights (645 ± 15 g), were then transferred to individual cages (73 × 55 × 80 cm), with feed and water provided ad libitum. All the birds were raised contemporaneously in the same experimental facility until slaughter at 56 days of age. The composition and chemical ingredients of the diet are shown in Table 1.

The final BW and feed consumption were weighed for each bird that received a 12-h-long fasting treatment. Growth performance parameters including metabolic body weight (MBW^0.75), BWG, FI, average daily weight gain (ADG), average daily feed intake (ADFI), and feed conversion ratio (FCR), were then calculated based on the results recorded from 22 to 56 days of age. The RFI was calculated following the method described by Aggrey et al. (2010). The equation used is provided as follows:

where a represents the intercept, b₁ represents the regression coefficients of FI on MBW^0.75, and b₂ represents the regression coefficients of FI on ADG. The RFI values were calculated following the REG procedure in SAS (version 9.4, SAS Institute Inc., Cary, NC, USA).

A total of 18 birds, including nine low-RFI (LR) and nine high-RFI (HR) birds, based on the calculation results, were selected for cecal content sample collection. Samples from each bird were collected in a sterile tube, followed by a rapid freezing treatment with liquid nitrogen, and stored at −80°C for further analysis. DNA from each sample was extracted using the cetyltrimethylammonium bromide and sodium dodecyl sulfate (CTAB/SDS) method described by Wang et al. (2012). The concentration and purity were then measured to ensure the availability of each DNA sample. Following that, the V4 region of the 16S rRNA gene was amplified using the universal primers 520F (F: GTGCCAGCMGCCGCGGTAA) and 802R (R: GGACTACHVGGGTWTCTAAT). Electrophoresis samples, with a bright main strip between 400 and 450 bp, were selected for further sequencing analysis. The PCR products were purified using a Qiagen Gel Extraction Kit (Qiagen, Hilden, Germany), followed by the generation of sequencing libraries using a TruSeq^® DNA PCR-Free Sample Preparation Kit (Illumina, USA). The library was sequenced on an Illumina Novaseq 6000 platform (Illumina Inc., San Diego, USA) after assessment of the library quality using a Qubit@ 2.0 Fluorometer (Thermo Scientific) and an Agilent Bioanalyzer 2100 system. Quality filtering of raw tags was performed under specific filtering conditions (length < 50 bp, or with a quality value of < 20, or having N bases) to obtain the high-quality clean tags in accordance with the Quantitative Insights Into Microbial Ecology (QIIME, V1.7.0) quality-control process. Sequences with a similarity of > 97% were assigned to the same operational taxonomic unit (OTU). For each representative sequence, the Greengenes database was used, based on the SILVA classifier algorithm, to annotate taxonomic information. All taxonomic results were then further processed by differential analysis of the bacterial community of the low- and high-RFI small-sized meat ducks.

Relative OTU abundances of each sample were calculated based on the normal distribution test, and then the independent Student's t-test of SAS 9.4 (version 9.4, SAS Institute Inc., Cary, NC, USA) was applied for the differential analysis. Alpha diversity and beta diversity of our samples were calculated by QIIME (version 1.7.0) and displayed using R software (version 4.1.3, R Core Team, Vienna, Austria). Principal component analysis (PCA) was performed using the ggplot2 package in R software. Pearson's correlations between bacterial communities and production performance were assessed using the PROC CORR procedure in SAS 9.4, and then the correlation matrix was created and visualized in a heatmap format by using R software.

The RFI values were calculated using the REG procedure in SAS (version 9.4, SAS Institute Inc., Cary, NC, USA). Differential analysis of growth performances between the low and high RFI was first verified by using the normal distribution test using the SAS procedure “proc univariate data=test normal,” followed by the independent Student's t-test to investigate the differences between the low- and high-RFI birds. The results were presented as mean ± SEM, and a p < 0.05 was considered significant, and 0.05 ≤ P < 0.10 was considered a tendency.

The effects of RFI divergence on the growth performance of small-sized meat ducks are shown in Table 2. The FI and ADFI of the birds in the LR group (5797.50 and 165.60 g, respectively) were found to be approximately 26.4% lower than those in the HR (7878.20 and 225.10 g, respectively) group (P < 0.01). Significant differences (P < 0.01) were observed in the RFI and FCR between the LR (−22.22 and 4.28 g) and HR (32.96 and 5.67 g) groups, respectively. Importantly, no differences were detected in the initial BW, final BW, MBW^0.75, BWG, and ADG between the two groups (P > 0.05).

16S rRNA gene amplicon sequencing analysis of the discrepancies between LR and HR ducks was conducted to identify the critical gastrointestinal microbiota. Taxonomy results of all bacteria are shown in Supplementary Table 1. In total, 19 kinds of phyla and more than 250 kinds of genera were identified in the present study, after data filtering. All the results were subsequently used for further differential and functional analyses. An integral insight into the differential OTUs between LR and HR groups using Venn diagrams was first conducted, and the results are shown in Figure 1. Based on the results, 422 of the identified OTUs were clustered in both LR and HR groups. HR ducks processed 121 unique OTUs, while LR processed 84 unique OTUs. All these OTUs were selected for the following α-diversity and β-diversity analyses.

For functional prediction analysis, the differentially identified gut microbiota of LR and HR ducks were identified using Tax4Fun (Aßhauer et al., 2015), and the result is shown in Figure 3. Metabolism processes, which included carbohydrate metabolism, amino acid metabolism, energy metabolism, cofactor and vitamin metabolism, and lipid metabolism, were the most abundant functions enriched from differential bacteria. Specifically, the biosynthesis of other secondary metabolites played an important role in the nutrient digestion process and also enriched the result. Other functions including cell viability and immune diseases were slightly correlated with the RFI.

An interactive analysis of gut bacteria and growth performance was conducted, and the results are shown in Figure 4. Integrally, all bacterial phyla could be separated into two large groups: One group mainly consisted of Bacteroidetes, Planctomycetes, Euryarchaeota, Nitrospirae, and Actinobacteria, which showed a positive correlation with FI, ADFI, and FCR and a negative correlation with BWG, ADG, and MBW⁷⁵. The other group mainly consisted of Firmicutes, Proteobacteria, Verrucomicrobia, Tenericutes, Epsilonbacteraeota, and Acidobacteria, which showed a completely converse correlation with the growth performance compared with the former group. In particular, Bacteroidetes showed a significant positive correlation with FI, ADFI, and FCR (P < 0.05), while Firmicutes showed a significant negative correlation with the aforementioned growth performance (P < 0.05). No other significant correlations were found between the gut microbiota and phenotypic parameters (P > 0.05).

Finally, the relationships between the gut bacteria and RFI, and the interactive effects among the bacterial communities at both phylum and genus levels were determined. The results are shown in Figures 5, 6. As shown in Figure 5, the RFI showed a significant correlation with Bacteroidetes and Firmicutes (P < 0.05). No other significant correlations were observed between the bacterial phyla and RFI (P > 0.05). Specifically, Bacteroidetes negatively correlated with Firmicutes and synergistically worked with Actinobacteria in regulating the RFI. In addition, Proteobacteria showed significantly positive correlations with Epsilonbacteraeota and Acidobacteria (P < 0.05), while Tenericutes antagonistically worked with Chloroflexi in regulating the RFI (P < 0.05). No other significant correlations were observed among the bacterial phyla focused on RFI regulation (P > 0.05). Further analysis of the bacterial genera and RFI was performed, and the results are shown in Figure 6. Integrally, the RFI was significantly correlated with Bacteroides, Alistipes, Phascolarctobacterium, Erysipelatoclostridium, Parasutterella, Fournierella, Subdoligranulum, Bifidobacterium, Ruminiclostridium_9, Sellimonas, Oscillospira, Blautia, Oscillibacter, Escherichia-Shigella, Lactococcus, and Streptococcus (P < 0.05). Specifically, the most abundant genus Bacteroides showed a significant negative correlation with Alistipes, Erysipelatoclostridium, Parasutterella, Subdoligranulum, Ruminiclostridium_9, Sellimonas, Oscillospira, Blautia, Oscillibacter, Escherichia-Shigella, and Streptococcus (P < 0.05), while it showed a significantly positive correlation with Phascolarctobacterium, Bifidobacterium, and Lactococcus (P < 0.05). Moreover, Faecalibacterium synergistically regulated the RFI with Bifidobacterium and Lactococcus and antagonistically worked with Parasutterella, Subdoligranulum, Ruminiclostridium_9, and Sellimonas (P < 0.05). Escherichia-Shigella mainly showed a negative correlation with the RFI and a negative correlation with Bacteroides, Megamonas, Phascolarctobacterium, Desulfovibrio, and Bifidobacterium (P < 0.05). Probiotics such as Bifidobacterium and Lactococcus showed a synergistic correlation with Bacteroides, Faecalibacterium, and Phascolarctobacterium (P < 0.05), while they antagonistically worked with Alistipes, Erysipelatoclostridium, Parasutterella, Fournierella, and Subdoligranulum (P < 0.05).

Nutrient digestibility is significantly modulated by both the gastrointestinal microbiota communities and the interactive effects between bacteria and the host intestinal epithelial proteins (Rowland et al., 2018; Zhao et al., 2019). In general, more than 90% of the gut microbial profiles belong to Firmicutes and Bacteroidetes (Tremaroli and Backhed, 2012), which was also found in the present study. The ratio of Firmicutes to Bacteroidetes is positively correlated with energy metabolism, which indicates that a higher Firmicutes abundance boosts starch digestibility, and thus, more energy is provided (Tremaroli and Backhed, 2012). In this study, Firmicutes significantly increased in LR ducks compared with HR ducks, and the proliferated Firmicutes may help to promote the degradation of feed concentrates and to improve the FE. Furthermore, increased amounts of probiotics may also help to maintain intestinal homeostasis and thereafter promote nutrient utilization. Specifically, Bifidobacterium, which plays an important role in maintaining intestinal homeostasis (Luo et al., 2018; Engevik et al., 2019), was significantly proliferated in LR ducks. This alteration helped to re-establish the gut microbial communities and modulated the interactions among bacterial communities, which further improved the nutritional digestibility and FE. Moreover, the increase in Bifidobacterium closely correlated with adiposity and microbe-derived inflammatory molecule reduction (Ramirez-Farias et al., 2009). Such an increase after the increment of Bifidobacterium protected the intestinal structure, enhanced intestinal barrier functions, and further promoted the nutrient intestinal absorptivity.

In broad terms, there are likely to be at least four major processes that impacted FE, namely, FI, feed digestion, anabolism, and catabolism (Bezerra et al., 2013; Soleimani and Gilbert, 2020; Martin et al., 2021). Previous studies highlighted that animals with a lower RFI generally showed enhanced bacterial degradability, higher carbohydrate metabolism, and activated enzymatic catalytic capacity (Liu et al., 2021, 2022). In the present study, the RFI was significantly correlated with Bacteroides, Bifidobacterium, Ruminiclostridium_9, Sellimonas, Escherichia-Shigella, Lactococcus, and Streptococcus, which actively participated in physiological carbohydrate metabolism and may have provided further energy for body growth. Furthermore, genes that participated in the volatile fatty acid absorption process, such as SLC16A3, SLC26A3, and HIF1A, and those that clustered into ketogenesis including LDHA, LDHB, and FFAR2 were proved to be highly positively correlated with a lower RFI (Salleh et al., 2017). These findings indicate that bacteria that participated in volatile fatty acid and ketosis metabolism processes may interact with the host genes, which ultimately proves the close correlation between the gut microbiota and RFI.

Nowadays, research focusing on the functions of gut microbiota is growing vigorously. The regulatory effects of gut microbiota on enteritis, cardiovascular disease, fecal microbiota transplants, and the subsequent interactions with host genes are well-studied. However, it is a challenge to discern general principles explaining the diversity and dynamics of complex multi-species communities, since detailed biological parameters are not typically available in large ecological networks (Krumbeck et al., 2021). As the body is a complicated system, the regulatory mechanism of the gut microbiota on the RFI also presents a systematic coexistence and a tunable interactive effect, as shown in Figures 3, 4. Bacterial communities can synergistically or antagonistically interact with one another and can ultimately determine nutrient digestibility and absorption. Traditionally, bacteria with similar degradative orientations synergistically interact with one another to further promote nutrient digestion, for example, Bifidobacterium and Lactococcus in carbohydrate metabolism (Fernández et al., 2020). Besides interactiona within bacterial communities, the morphogenesis effects of gut microbiota on body intestinal epithelial development through secondary metabolites such as butyrate also work to regulate the FE (He et al., 2021). To date, host and bacterial interactions are a great causal factor that can shape the gastrointestinal tract, impact growth rates, and carry capacity by providing more energy and scavenging inflammatory molecules (Jackrel et al., 2020; Houwenhuyse et al., 2021). Gut microbiota can regulate intestinal development by acquiring indispensable materials from degrading carbohydrates and proteins. Nutrient absorption occurs in the gastrointestinal tract and is regulated by high levels of acids, abundant digestive enzymes, and antimicrobials (O'Callaghan and Corr, 2019). The epithelium also contains plentiful functional proteins, which conduct substance interchange or achieve information exchange with other target metabolites. Intriguingly, microbial communities that positively correlate with the RFI are mainly clustered into carbohydrate-degraded organisms; therefore, more effective energy utilization is provided for body growth.

In summary, the RFI was significantly correlated with gut microbial communities and was regulated by the interactive effects between the gut microbiota and crosstalk with the host genes. These microbiota communities related to carbohydrate metabolism have positive regulatory effects on the low-RFI ducks. In future, we aim to improve the relative abundance and utilization of these communities to promote the FE of small-sized meat ducks. The present study provides valuable insights into the dynamics of gut microbiota underlying the variations in the FE of small-sized meat ducks.