All procedures involving animal care and management used in this experiment were authorized by the Institutional Animal Care and Use Committee of Guangdong Ocean University (Zhanjiang, Guangdong, China).

The alfalfa hay was imported from the State of California, USA. The whole-plant amaranth at heading stage was harvested with a reaping hook to a 5-cm stubble height and cut into fragments of 4-cm in length by a forage chopper (New Shengtai Machinery Manufacturing Co. Ltd., Jining, China). Next, fresh amaranth forages were naturally dried (approximately 3 d) to make air-dried samples. Chemical composition of alfalfa and amaranth hay are shown in Table 1.

The animal experiment was performed at a commercial dairy farm that has approximately 8,000 milking cows. Forty-five multiparous (parity = 2) Holstein dairy cows with similar body weight (586±18 kg), milk yield (34.68±1.4 kg) and days in milk (72±5 d) were randomly assigned to 3 groups (15 cows in each group): a control group without amaranth hay (CON) and either 50% (AH1) or 100% (AH2) of the dietary alfalfa hay was replaced by an equal level of amaranth hay.

All the cows were provided with total mixed rations which was formulated based on the NRC [10]. To ensure that the three diets were isoenergetic and isonitrogenous, the proportion of oat hay in the diet was increased with the elevation of amaranth hay while the quantity of corn silage in the diet was decreased. Feed ingredients and nutritional levels of experimental rations are described in Table 2. Cows were fed three times daily at 06:30, 14:30, and 22:30 respectively, allowing 5% to 10% orts and had free access to clean water. All experimental animals were mechanically milked three times each day at 05:30, 13:30, and 21:30 in rotary milking parlor. The feeding trial was conducted with a 10-d adaptive phase followed by 60 d experimental period.

The feed intake was recorded according to the difference of feed offered and refused and converted into dry matter intake (DMI). DMI of each treatment was used to calculate the average daily DMI per animal. The milk yield was collected daily. On d 60, approximately 50 mL of milk samples were obtained three times all day and blended by volume in the ratio of 4:3:3 (morning, afternoon, and evening) for milk composition analysis. The 4% fat-corrected milk (FCM) yield was calculated from the following equation: 4% FCM = 0.4× milk yield+1.5×milk fat production. Feed efficiency was obtained via dividing 4% FCM by DMI.

On d 60, ruminal fluid samples of all animals were collected through a flexible esophageal tube (Anscitech Animal Husbandry Technology Co. Ltd., Wuhan, China) at 4 h after morning feeding. Immediately, the ruminal fluid pH was measured with a pH meter (Jingcheng Instrument Co. Ltd., Qingdao, China). Next, fluid samples were filtered by 4 layers of cheesecloth and preserved in 10 mL centrifuge tubes and stored at 20°C for analysis of ruminal fermentation parameters.

Fecal samples of all cows were collected from d 58 to 60 according to the procedures described by a previous study [11]. Briefly, fecal samples were collected at 6 h intervals from rectums of cows and the sampling time were as follows: d 58, 20:00, 02:00, 08:00, and 14:00; d 59, 18:00, 00:00, 06:00, and 12:00; and d 60, 16:00, 22:00, 04:00, and 10:00. In the meantime, the fresh feed and orts were sampled daily. The fecal samples were mixed per cow and subsampled. All feed, orts, and fecal samples (the 100 g feces were blended with 10 mL of 10% sulphuric acid) were dried at 65°C in a forced-air oven for 48 h to a constant weight. Subsequently, dried samples were ground to pass through a 1-mm sieve (Aizela Electric Appliance Co. Ltd., Ningbo, China) for measurement of nutrient digestibility.

A total of 10 mL blood were sampled from the caudal vein of each cow using evacuated tubes containing no anticoagulant before morning feeding on d 0 and 60. All blood samples were centrifuged at 3,200 rpm and 4°C for 15 min to collect serum, then preserved at −20°C for further analysis.

The milk components, including protein, fat, lactose, urea nitrogen (MUN) and solids, were determined through an automated multifunctional milk analyzer (Hengmei Electronic Technology Co. Ltd., Weifang, China). Moreover, somatic cell count (SCC) was analyzed with a fossomatic apparatus (Haiyi Technology Co. Ltd., Beijing, China). The ruminal fluid samples were firstly thawed, then the supernatant was obtained by centrifugation at 15,000 rpm for 12 min at 4°C to measure the contents of volatile fatty acid (VFA), ammonia nitrogen (NH₃-N) and microbial crude protein (MCP) following previous procedures [11].

The serum biochemical parameters, including total protein (TP), albumin (ALB), globulin (GLB), alanine transaminase (ALT), aspartate transaminase (AST), alkaline phosphatase (ALP), glucose (GLU), triglyceride (TG), and urea nitrogen (SUN) were determined by an automated biochemical analyzer (BS350; Baden Medical Co. Ltd., Nanjing, China). Besides, the concentrations of glutathione peroxidase (GSH-Px), superoxide dismutase (SOD), catalase (CAT), malondialdehyde (MDA), and total antioxidant capacity (T-AOC) were analyzed utilizing commercially available kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) according to the instructions.

The feed and fecal samples were analyzed for DM (method 934.01), organic acid (OM, method 942.05), CP (method 984.13), and ether extract (EE, method 954.02) based on the AOAC procedures [12]. Also, the neutral detergent fiber (NDF) and acid detergent fiber (ADF) contents were measured with an ANKOM fiber analyzer (ANKOM A2000i, USA). The acid-insoluble ash content in feed and feces was determined according to the procedure of Van Keulen and Young [13] and used as an indicator for determination of nutrient digestibility [11].

All data were analyzed with one-way analysis of variance procedure of the SPSS statistical software (version 20.0 for Windows; SPSS, Chicago, IL, USA). Each cow was used as the experimental unit. The model for statistical analysis is as follows: Y = μ+T+e, where Y = dependent variable, μ = general mean, T = treatment effect and e = residual error. Orthogonal polynomial contrasts were completed to detect the linear and quadratic effects according to increase in the dietary amaranth hay levels (three groups) of dairy cows. Tukey test was utilized to determine the differences among three treatments. Results were presented as means and standard error of the mean. Statistical differences were considered significant for p<0.05 and as a tendency for 0.05≤p<0.10.

As shown in Table 3, the DMI of CON and AH1 groups was higher (p<0.005) than that of AH2 group. Similarly, the milk yield of AH2 group was reduced (p<0.05) when compared to AH1 group. In addition, the 4% FCM yield in AH1 group tended to be higher than that in AH2 group. Compared with CON group, the ratio of 4% FCM yield to DMI in AH2 group tended to increase by 3.62%.

Notably, the milk protein, lactose, total solids, MUN and SCC were similar among three groups (Table 4). However, dietary substitution of alfalfa hay by amaranth hay increased (p<0.05) milk fat.

Ruminal pH was similar and averaged 6.27, 6.22, and 6.18 in CON, AH1, and AH2 groups respectively (Figure 1A). Likewise, there was no significant difference of MCP (Figure 1C), propionate (Figure 1E), butyrate (Figure 1F), total VFA (Figure 1H) concentrations and acetate-to-propionate ratio (Figure 1G) among three groups. Nevertheless, the NH₃-N (Figure 1B) and acetate (Figure 1D) concentrations of AH1 and AH2 groups were higher (p<0.05) than those of CON group. Moreover, the NH₃-N concentration in AH2 group was elevated (p<0.05) as compared with AH1 group.

The OM (Figure 2B), ADF (Figure 2E), and EE (Figure 2F) digestibility did not show significant difference among three groups. However, the CP digestibility (Figure 2C) of AH2 group was lower (p<0.05) than that of CON group. Similarly, the DM digestibility (Figure 2A) of CON group tended to be higher than AH2 group. Compared with AH2 group, the NDF digestibility (Figure 2D) of AH1 group was increased (p<0.05).

On d 0, no obvious difference of all serum biochemical parameters was observed among three treatments (Table 5). Likewise, the concentrations of TP, ALB, GLB, GLU, TG, ALT, AST, and ALP were similar among three treatments on d 60. However, the SUN content of AH2 group was increased (p<0.05) when compared to CON group.

There was no significant difference of GSH-Px, CAT, SOD, MDA, and T-AOC serum concentrations among CON, AH1, and AH2 groups on d 0 (Table 6). On d 60, the CAT and SOD concentrations were similar among three groups. However, dietary substitution of alfalfa hay by amaranth hay increased (p<0.05) the concentration of GSH-Px in serum of cows. Compared with CON group, the MDA activity of AH1 group was decreased (p<0.05). In addition, the T-AOC activity in amaranth hay groups tended to be higher than CON group.

As shown in Table 7, the AH1 group displayed the greatest economic benefit. Compared with CON and AH2 groups, the farming benefit of cow per day in AH1 group was increased by 5.01% and 5.12% respectively.

Feed intake is critical for dairy cows to maintain the milking performance. Previous study in fattening lambs found that dietary supplementation with amaranth silage increased the DMI [7]. In the current study, 50% dietary alfalfa hay replacement by amaranth hay had no obvious effect on DMI of dairy cows, whereas 100% replacement significantly reduced the DMI. The chemical composition, especially NDF content, of ration affect DMI of ruminants [1]. Thus, the reduced DMI of AH2 group may be related to high NDF content of amaranth hay. Correspondingly, the cows in AH2 group showed lowest milk yield, illustrating that the 100% substitution ratio might be too much when replacing alfalfa hay with amaranth hay in dairy cows. Different physical characteristics of the experimental feeds such as fiber texture have direct influence on DMI [14]. The increased fiber content in amaranth hay may induce satiety and decrease DMI of dairy cows, then result in reduced milk yield. According to the results of current study, 50% substitution of alfalfa hay by amaranth hay did not affect lactation performance of dairy cows but improved the economic benefit of dairy farming.

Milk composition is the main index to evaluate the quality of milk. A previous study has reported that the substitution of alfalfa hay by different levels of roughage did not affect the milk composition [15]. Our study found that dietary replacement of alfalfa hay with amaranth hay did not affect the milk protein, lactose, total solids, MUN and SCC of dairy cows, which was consistent with previous study [15], suggesting that amaranth hay did not have negative influence on milk composition. However, the amaranth hay treatments significantly increased the milk fat. Milk fat is an important parameter to assess the production performance of dairy cows and can provide energy, fatty acid, and essential nutrients for humans. The milk fat can be synthesized by acetate and butyrate, and high acetate and butyrate contents commonly lead to increased milk fat production [16]. In this experiment, although the butyrate content was not changed among different treatments, the content of ruminal acetate was elevated significantly in amaranth hay substitution treatments, which was matched to the milk fat result.

In general, the ruminal fermentation characteristics are largely affected by chemical composition of diet. Ruminal pH affects the growth and proliferation of microorganisms and regulates the VFA production within a normal range 6 to 7 [17]. In our research, the ruminal pH of three treatments were within the normal range of 6.18 to 6.27, indicating that amaranth hay had no negative effects on ruminal fermentation of dairy cows. Ruminal NH₃-N is an intermediate product of dietary protein and non-protein nitrogen degradation and microbial protein synthesis, and its concentration reflects the balance state of protein degradation and synthesis, which is mainly affected by dietary protein degradation, ruminal epithelium absorption, microbial utilization and digesta outflow rate [18]. With the elevation of substitution level, the NH₃-N concentration was significantly increased. The result could be attributed to the fact that the fiber content of amaranth hay is higher than that of alfalfa hay. Thus, the feed requires multiple rumination to digest, then the retention time of feed in the rumen is increased, resulting in a relatively high NH₃-N content. In addition, the reduction of dietary fiber content increases the activity of protein-utilizing bacteria, then decreases the NH₃-N content in the rumen [19]. The fiber structure of amaranth hay and effects of amaranth hay on ruminal microbial community need further investigation.

The acetate in the rumen is mainly produced by microbial fermentation with NDF as substrate [20]. Therefore, in the present experiment, increased ruminal acetate content in amaranth hay treatments could be attributed to the high NDF content. However, although the ruminal acetate concentration was increased in AH1 and AH2 groups, the total VFA concentration was similar among three groups, which may be associated with low nutrient digestibility. Previously, a study has found that the NDF content in the ration had no obvious effect on ruminal total VFA concentration [21], which was in accordance with our study. Moreover, a recent study reported that the acetate-to-propionate ratio had positive correlation with milk fat yield [22]. We did not observe significant difference of ratio of acetate to propionate. However, the ratio of acetate to propionate of three groups was more than 3, thus the milk fat percentage was above 3.9% in all dairy cows. An in vitro study demonstrated that the VFA production of amaranth diet was low [23], which was inconsistent with our study. In our experiment, the ruminal fluid samples were collected after morning feeding and might only reflect the ruminal fermentation characteristics at that time. The dynamic patterns of rumen fermentation caused by dietary substitution of alfalfa hay with amaranth hay require an in-depth study.

In addition to rumen fermentation, nutrient digestibility directly affects the production performance of cows. A previous study investigated the influence of different roughage on nutrient digestibility of milking cows and found that the nutrient digestibility was reduced after dietary replacement of alfalfa hay with straw (rice or maize) [24], which was basically consistent with our experiment. The nutrient digestibility of dairy cows is commonly affected by feed type, nutritional level, and fiber content [25]. In this study, the 100% substitution ratio treatment reduced the CP and NDF digestibility, which might be related to high NDF content of amaranth hay. These results might explain the lowest milk yield in AH2 group. Furthermore, we found that 50% substitution ratio of amaranth hay displayed a positive effect on NDF digestibility. The positive effect might be associated with a synergistic effect between amaranth and alfalfa, which would then increase the relative abundance of bacteria associated with cellulose degradation, but this mechanism of action needs to be fully elucidated. A recent study found that partial replacement of millet for alfalfa hay did not affect the nutrient digestibility of dairy cows; however, a high ratio of replacement decreased the nutrient digestibility [26], which was in line with our study.

The changes of serum biochemical parameters which are closely related to the health of animals can be used to evaluate the physiological metabolism of the body and changes in various organ function [27]. The TP, ALB, GLB, and SUN concentrations in serum reflect protein metabolism, and the serum GLU and TG contents are associated with lipid and energy metabolism. Additionally, the ALP, ALT, and AST concentrations in serum are indicators of hepatic function [27]. In this experiment, the serum concentrations of TP, ALB, GLB, GLU, TG, ALP, ALT, and AST were similar among three groups, indicating that amaranth hay replacement did not have negative effects on health of dairy cows. A previous study found that dietary substitution of corn silage by amaranth silage did not change the serum biochemistry parameters of fattening lambs [7], which was consistent with our research. However, our result showed that 100% substitution treatment significantly increased the serum SUN concentration, suggesting that the nitrogen conversion was low in AH2 cows. The variation of nitrogen conversion was mostly in accordance with CP digestibility, which indicated that the low nitrogen conversion could be partly explained by the decreased CP digestibility in AH2 group. Nousiainen et al [28] reported that increased content of MUN, SUN, and ruminal NH₃-N suggested a reduced nitrogen efficiency of dairy cows. In dairy cows, an increase of MUN and ruminal NH₃-N contents and a decrease of nitrogen conversion were observed in cows when replacing cereal straw for alfalfa hay [24]. In the current study, although the MUN was similar among three treatments, considering the high content of ruminal NH₃-N and SUN in AH2 cows, a decreased nitrogen conversion efficiency should be expected. A previous study in dairy cows also found that high level of substitution of corn silage by amaranth silage increased the SUN content [6]. According to our findings, we speculated that higher NH₃-N absorption from the rumen and urea production in the liver were mainly responsible for elevation of SUN in AH2 cows. In the present experiment, the DMI, milk yield, feed efficiency and nitrogen conversion were similar between CON and AH1 treatments, which only reduced in AH2 treatment, indicating that the 50% substitution ratio was appropriate when replacing alfalfa hay with amaranth hay in dairy cows.

The serum GSH-Px, CAT, SOD, and T-AOC are the important biomarkers related to antioxidative ability of animals. As a lipid peroxidative product, serum MAD can be used to reflect the oxidative tissue impairment [29]. In our experiment, serum GSH-Px activity was enhanced, and T-AOC activity tended to be increased by amaranth hay treatment, which may be associated with the antioxidant properties of polyphenols and carotene that naturally occur in amaranth [8]. Our finding was in accordance with the previous study which found that laying hens treated with amaranth grain displayed higher antioxidant enzyme activity and lower MDA concentration in serum [30]. The lower serum MDA caused by amaranth hay indicated decreased lipid peroxidation, which had positive effects on metabolism of dairy cows.