A total of 480 Jiaji ducks (42 days old, male:female ratio = 1:1) with similar body weight were randomly assigned to one of four treatment. There were six replicates per treatment, with 20 ducks per replicate. The leaves of E. scaber were obtained from a local farm, and dried, and then ground into power. The ducks in the control group (Con) were fed a basal diet; the three experimental groups were fed a basal diet supplementation with 30 (T1), 80 (T2), and 130 mg/kg (T3) of E. scaber. The experiment was conducted from 42 to 90 days of age. The basal diet (Table 1) was formulated according to the recommendation of the Nutrient Requirements of Poultry (National Research Council, 1994). All ducks were reared in raised plastic wire-floor pens with a pen size of 20 m² (4 m × 5 m) and had free access to water and feed. The duck house (10 m × 80 m) was equipped with a window. Each pen was provided with a 20-W incandescent white fluorescent lamp, and 24-h continuous incandescent lighting was provided. The temperature was set initially at 30°C and reduced gradually to 20°C at the end of the experiment. The relative humidity in duck house was 75–65%. The feed intake was recorded daily, and the final body weight was recorded at 90 days age. Average daily gain (ADG), average daily feed intake (ADFI), and the ratio of feed intake to body weight gain (F:G) of ducks for each replicate were measured. This study was performed in accordance with the recommendations of the Guide for the Care and Use of Chinese Academy of Tropical Agricultural Sciences.

After 12 h of feed deprivation, one duck with a similar average body weight was selected from each replicate for sample collection. The animals were stunned electrically, killed immediately by neck cutting, and eviscerated manually. Blood samples were collected from the wing vein using 10-ml centrifuge tubes and centrifuged at 3,000 × g and 4°C for 15 min to recover the serum. The skin with subcutaneous fat, liver, heat, gizzard, abdominal fat, breast meat (pectoralis muscle), and leg meat (all thigh muscle) was removed from the right side of carcasses and weighed; and the percentages relative to live body weight, eviscerated, and semi-eviscerated were calculated. Breast muscle samples were collected from the breast and then stored at −80°C for further analysis. The anterior jejunum and posterior ileum segments (~2 cm in length) were fixed in 4% paraformaldehyde for analysis of intestinal morphology. The colon was separated, and the fecal sample was collected into a 10-ml sterile tube and stored at −80°C for further analysis.

The concentrations of albumin (ALB), alanine aminotransferase (ALT), alkaline phosphatase (ALP), aspartate aminotransferase (AST), creatinine (CRE), glucose (GLU), high-density lipoprotein-cholesterol (HDL-C), low-density lipoprotein-cholesterol (LDL-C), total cholesterol (T-CHO), triglyceride (TG), total protein (TP), and uric acid (UA) in the serum were measured using the biochemical analytical instrument PUZS-600B (Medical Equipment Co., Ltd., Beijing, China) and respective commercial assay kits.

The concentration of inosinic acid (IMP) in the breast muscle was determined according to the procedure described previously (11). In brief, breast muscle samples were dried in a vacuum freeze dryer and ground into powder. Muscle powder (about 0.1 g) was mixed with 10 ml of perchloric acid (5%), and then the mixture was homogenized. The homogenate was centrifuged at 3,000 rpm for 10 min at 4°C to obtain the solution. The solution was filtered through a 0.45-μm membrane filter; then the samples were analyzed using high-performance liquid chromatography system (Agilent 1200 series, Waldbronn, Germany).

The concentrations of Fe, Se, and Zn in breast muscle sample were determined according to the method described previously (12). Muscle sample (0.5 g) was mixed with 5 ml of acid solution (HNO₃:HClO₄ = 4:1) in 50-ml conical flask for 24 h, and then the lysate was heated to 180°C on an electric heating plate. Ten milliliter of 5 M hydrochloric acid solution was added to the conical flask after the lysate returned to room temperature. The lysate was then heated until clarification. Then, the solution was diluted to a volume of 25 ml with 2 M of hydrochloric acid and 10% potassium ferricyanide solution. Fe, Se, and Zn standard solutions were used to construct standard curves. The concentrations of Fe, Se, and Zn were detected using an atomic fluorescence spectrometer (AFS-9330; Beijing Titan Instruments Co., Ltd., Beijing, China).

Muscle powder (about 0.1 g) was hydrolyzed in 5 ml of 6 M hydrochloric acid solution at 110°C for 24 h. The suspension was diluted with water, 1 ml of the supernatant was filtered through a 0.45-μm membrane, and then the samples were analyzed using an amino acid (AA) analyzer (L8800, Hitachi, Tokyo, Japan).

Fatty acid contents in the breast muscle were analyzed by gas chromatography (GC). In detail, total lipids were extracted from the breast muscles using the chloroform–methanol method. The lipids were converted to fatty acid methyl esters using KOH/methanol and analyzed using GC. The GC conditions were conducted according to our previous publication (13). Each fatty acid peak was identified by comparing their retention times with standards. The fatty acid composition was calculated as a percentage of the total fatty acids.

Hematoxylin and eosin (H&E) staining was used to analyze intestinal morphology. Intestine tissues fixed in 4% paraformaldehyde were paraffin-embedded, sectioned at 5-μm thickness, and stained with H&E. The images were captured using a light microscope (Olympus Bx51, Tokyo, Japan). The villus height and crypt depth were measured using ImageJ software.

DNA samples were extracted from feces through the HiPure Stool DNA kit B (Magen, Shanghai, China). The quality and concentration of DNA were measured using a NanoDrop® ND-1000 instrument (NanoDrop Technologies Inc., Wilmington, DE, USA). The V3–V4 region of the 16S rRNA gene was amplified using primers 515F (5′-GTGCCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) (14). The amplified PCR products were identified, purified, and quantified. Equal amounts of purified products were subject to paired-end sequencing on the Illumina MiSeq platform (Illumina, Inc., San Diego, CA, USA).

QIIME and FLASH (v.1.2.11) software package were used to analyze the Raw Illumina fastq files; then the raw tags were analyzed using Trimmomatic (v.0.30) and FLASH (v.1.2.11) software packages to obtain high-quality clean reads. The criteria for high-quality sequence selection was described in our previous study (14). The high-quality sequences were clustered into operational taxonomic units (OTUs) with sequence similarity of 97% using USEARCH (v7.0.1090). OTU sequences were taxonomically classified according to the Ribosomal Database Project (RDP) Classifier based on the Greengene (V201305) reference database.

Data were statistically analyzed by one-way ANOVA using SPSS 20.0 software (SPPS Inc., Chicago, IL, USA). Homogeneity of variances was tested using Levene's test. The differences among the groups were analyzed using Duncan's multiple range tests or Tamhane's T2 tests. The relative species abundances (at phylum and genus levels) of gut microbial communities were analyzed using the Kruskal–Wallis test. Pearson's correlation coefficient was conducted to assess the relationships among environmental factors. Differences were considered statistically at p-value of <0.05. Data were expressed as the mean ± SEM.

As shown in Table 2, no differences (p > 0.05) were observed in the final body weight, ADG, or ADFI among the four groups. Compared with the Con group, the ratio of F:G in the T1 group was decreased (p < 0.05) by 9.3%; however, no difference (p > 0.05) was observed in the F:G among the Con, T2, and T3 groups.

As shown in Table 3, the T3 group was higher (p < 0.05) than the Con group in the semi-eviscerated. The rates of carcass yield, eviscerated, breast muscle, leg muscle, abdominal fat, skin and subcutaneous fat, heart, liver, and gizzard were not affected (p > 0.05) by dietary supplementation with E. scaber.

As shown in Figure 1, no differences (p > 0.05) were observed in the concentrations of ALB, ALP, GLU, LDL-C, T-CHO, TG, TP, UA, or urea among the four groups (Figures 1A,C,F,H–M). Compared with the Con group, the T2 group was shown to decrease (p < 0.05) the concentration of ALT in the serum (Figure 1B). In addition, the T3 group was lower (p < 0.05) than the Con group in the concentration of AST (Figure 1D) and was higher (p < 0.05) than the Con group in the concentration of HDL-C (Figure 1G). The highest concentration of CRE was observed in the T1 group (p < 0.05; Figure 1E).

As shown in Figure 2, the concentration of IMP in the T1, T2, and T3 groups was higher (p < 0.05) than that in the Con group; however, there was no difference in the concentration of IMP in the breast muscle among the T1, T2, and T3 groups (p > 0.05).

As shown in Table 4, the Con group was higher than the T2 or T3 group in the Zn content. No difference (p > 0.05) was observed in the Fe or Se contents among the four groups.

As shown in Table 5, the contents of Phe, Ala, Gly, Glu, Arg, Lys, Tyr, Leu, Ser, Thr, Asp, and total AAs in the breast muscle were the highest (p < 0.05) in the T2 group. Compared with those in the Con group, the contents of Phe, Gly, Glu, Tyr, Leu, Ser, Thr, and Asp were increased (p < 0.05) in the T1 and T3 groups. The T3 group was higher than the Con group in the content of total AAs (p < 0.05).

As shown in Table 6, dietary supplementation with E. scaber had no effects (p > 0.05) on the contents of C14:0, C16:0, C18:1n−9, C18:3n−6, C20:3n−6, C20:4, C24:0, C22:6ns, or monounsaturated fatty acid (MUFA) in the breast muscle. The content of C16:1 was highest in the Con group (p < 0.05). The T1 group was higher than the Con and T2 groups in the content of C18:0. The content of saturated fatty acid (SFA) was lower in the T2 group than that in the Con, T1, or T3 group, and that of polyunsaturated fatty acid (PUFA) was higher in the T2 group than in the Con group.

As shown in Figure 3, the crypt depth of jejunum was significantly decreased in the T1, T2, and T3 groups than in the Con group (p < 0.05); however, no difference (p < 0.05) was observed in the jejunal villus height and the ratio of villus height to crypt depth among the four groups. In the ileum, the T3 group was higher (p < 0.05) than the Con group in the villus height and the ratio of villus height to crypt depth.

As shown in Table 7, dietary supplementation with E. scaber showed no effect on observed OTUs, Shannon, Simpson, or Chao1 index of microbiota. At the phylum level (Table 8), the most dominant phyla among microbiota communities (>1%) were Firmicutes, Bacteroidetes, Proteobacteria, Fusobacteria, and Actinobacteria; these microbiotas accounted for more than 93% of the total microbiota found in fecal samples. The abundances of Firmicutes and Bacteroidetes were the highest in the T2 group. The T3 group had a trend to increase the abundance of Fusobacteria. At the genus level (Table 9), the most dominant genus among microbiota communities were Bacteroides, Fusobacterium, Ruminiclostridium_9, Desulfovibrio, Lachnospiraceae, Clostridiales_vadinBB60_group, Intestinimonas, Gorbachella, Oscillospira, Alistipes, and Erysipelatoclostridium. Compared with the Con group, the T1 and T2 groups displayed a higher abundance of Subdoligranulum (p < 0.05). The abundances of Fusobacterium and Ruminiclostridium_9 were the highest in T3, and the abundance of Desulfovibrio was the highest in the T1 group.

As shown in Figure 4, the abundance of Alistipes was positively correlated with the F:G, the contents of C18:2n−6, and PUFA in the breast muscle (p < 0.05), while it was negatively correlated with the villus height (p < 0.05). The abundance of Olsenella was negatively correlated with the content of Asp in the muscle (p < 0.05). The abundance of Ruminococcus_2 was positively correlated with the level of PUFA in the muscle (p < 0.05), and that of Fusobacterium was negatively correlated with the F:G and the content of Leu in muscle. In addition, the abundance of Bacteroidales was positively correlated with the content of C18:2n−6 in the breast muscle.