**Changed Caecal Microbiota and Fermentation Contribute to the Beneficial E**ff**ects of Early Weaning with Alfalfa Hay, Starter Feed, and Milk Replacer on the Growth and Organ Development of Yak Calves**

### **Shengru Wu 1,\*,**†**, Zhanhong Cui 1,2,**†**, Xiaodong Chen 1, Peiyue Wang <sup>1</sup> and Junhu Yao 1,\***


Received: 27 September 2019; Accepted: 1 November 2019; Published: 5 November 2019

**Simple Summary:** Yak calves during the pre-weaning period are mainly fed by maternal grazing and nursing, which is beneficial to the oestrus and mating of female yaks or the survival and growth of calves. Barn feeding and early weaning with mixed rations of available roughage and grains was presented as an alternative to maternal grazing and was supposed to be beneficial to the tremendous ruminal and intestinal development and growth of yak calves. The caecum is also the primary site of microbial fermentation, but the limited research has focused on the role of caecal microbiota in regulating the growth of yaks. The findings of the current study indicated that early weaning by supplying calves with milk replacer, alfalfa hay, and starter feed improves yak calf growth performance compared with maternal grazing and nursing, in part through alterations of caecal microbiota and caecal volatile fatty acid (VFA) production induced by supplementation with alfalfa hay and starter feed.

**Abstract:** This study aimed to investigate the effect of early weaning by supplying calves with alfalfa hay, starter feed, and milk replacer on caecal bacterial communities and on the growth of pre-weaned yak calves. Ten 30-day-old male yak calves were randomly assigned to 2 groups. The maternal grazing (MG) group was maternally nursed and grazed, and the early weaning (EW) group was supplied milk replacer, starter feed, and alfalfa hay twice per day. Compared with the yak calves in the MG group, the yak calves in the EW group showed significantly increased body weight, body height, body length, and chest girth. When suffering to the potential mechanism of improved growth of yak calves, except for the enhanced ruminal fermentation, the significantly increased total volatile fatty acids, propionate, butyrate, isobutyrate, and valerate in the caecum in the EW group could also serve to promote the growth of calves. By using 16S rDNA sequencing, some significantly increased caecal phylum and genera, which were all related to the enhanced caecal fermentation by utilizing both the fibrous and non-fibrous carbohydrates, were identified in the EW group. In conclusion, early weaning of yak calves by supplying them with alfalfa hay, starter feed, and milk replacer is more beneficial to the growth of yak calves when compared with maternal grazing and nursing, in part due to alterations in caecal microbiota and fermentation.

**Keywords:** yak calf; early weaning; caecal microbiota; 16S rRNA gene sequencing; growth performance

#### **1. Introduction**

Yak calves during the pre-weaning period are mainly fed by maternal grazing and nursing, which are not beneficial to the oestrus and mating of female yaks or the survival and growth of calves [1]. However, the pre-weaning period is a critical period for the developmental plasticity and, subsequently, biological function changes of young ruminants [2,3]. Adequate nutrition during early life is beneficial to gastrointestinal microbiota establishment, development, and the subsequent functional transition from metabolizing the glucose from milk to the volatile fatty acids (VFAs) from a solid diet [4,5]. Barn feeding and early weaning with mixed rations of available roughage and grains was presented as an alternative to maternal grazing and was supposed to be beneficial to the tremendous gastrointestinal ramifications and growth of yak calves and other juvenile ruminants [5,6]. In previous studies, the significantly enhanced rumen fermentation and changed rumen microbiota condition were implicated as the main reasons for the observed improved growth performance of cattle and lamb by supplying them with alfalfa hay, starter feed, and milk replacer in barn feeding and early weaning groups [4–6], which were rarely studied in the yak calves.

In addition to rumen fermentation, hindgut fermentation, which includes caecal fermentation, is also an important factor that affects growth performance and healthy conditions [4,7]. The caecum is also the major site of fermentation and absorption in the large intestine of ruminants, and approximately 17% of digested cellulose is broken down there [8]. The VFAs produced in the caecum account for 12% of total VFA production in sheep [9]. However, compared with the extensive studies focusing on the rumen microbiota and fermentation, caecal microbiota and fermentation is also an important factor that affects growth performance, which was comparatively limited in studies but worth further studying of the roles of microbiota from different segments in utilizing the nutrients and promote the growth of yak calves. In the present study, the effect of early weaning with alfalfa hay, starter feed, and milk replacer versus maternal grazing and nursing on the caecal microbiota and fermentation of yak calves was evaluated and compared, with the aim of further adding knowledge of changed caecal microbiota in regulating the growth of yak calves. Moreover, we further compared the differences between ruminal and caecal microbiota and fermentation to justify the contribution of caecal microbiota and fermentation on the growth of yak calves.

#### **2. Materials and Methods**

#### *2.1. Ethics Approval Statement*

This study was carried out in accordance with the recommendations of the Administration of Affairs Concerning Experimental Animals (Ministry of Science and Technology, China, revised 2004). The protocol was approved by the Institutional Animal Care and Use Committee of the Northwest A&F University (protocol number NWAFAC1118).

#### *2.2. Animals, Experimental Design, and Sample Collection*

Before the commencement of the trial, all yak calves were only fed with the milk by maternal nursing in Datong Yak Breeding Farm of Qinghai Province. A total of ten 30-day-old male yak calves (34.86 ± 2.06 kg) with similar body conditions were randomly assigned to 2 groups with 5 calves per group. The maternal grazing (MG) group was maternally nursed and grazed, and the early weaning (EW) group was supplied with milk replacer, starter feed, and alfalfa hay. The yak calves in the maternal nursing group had access to fresh grass and yak milk. Briefly, the MG yak calves were allowed to graze a rangeland for a period of 8 h. Water was offered ad libitum twice a day at 08:00 and 16:00 h. Specifically, the experiment was performed from July to October and lasted for 90 d, allowing for the sufficient grazing of fresh grass. Moreover, at the last day of the feeding experiment, the fresh grass and the yak milk were collected and provided to the their yak calves, and the dairy intake were recorded and used to calculate the dry matter intake (DMI). The yak calves in the early weaning group were housed in a barn and kept in individual pens (7 × 4 m). The pens included a sawdust-bedded pack area and a feed lane equipped with an automatic cable scraping system. In addition to free access to starter feed and alfalfa hay, all yak calves in the early weaning group were supplied with milk replacer reconstituted from 100–350 g milk replacer powder (the supplementation of milk replacer were increased along with the increasing body weight) dissolved in 1 L 60 ◦C water twice per day at 08:00 and 16:30. Water was supplied ad libitum to the yak calves during the experimental period. Feed (include the alfalfa and starter feed) offered was adjusted daily to ensure at least 10% orts. Feed offered and refused by each calf was weighed and recorded on the last day of the feeding experiment. Meanwhile, the daily intake was calculated for further analysis of DMI (overall consideration of the dry matter intake of milk replacer, alfalfa, and starter feed). After the feeding experiment, the yak calves were weighed, and their body size indexes, including the body height, body length, and chest girth, were measured and recorded. Then all animals were euthanized by exsanguination after anaesthesia and immediately dissected, and the liver, thymus, spleen, and pancreas were collected and weighed immediately. At last, the ruminal fluid and caecal contents were collected and stored in −80 ◦C for further analyses. Specifically, rumen fluid was strained through 4 layers of sterile cheesecloth and collected for VFA and NH3-N analyses and 16S rRNA gene sequencing.

Composites of the fresh grass, starter feed, alfalfa hay, and milk replacer were measured (AOAC International, 2000) for DM (oven method 930.15), ash (oven method 942.05), CP (Kjeldahl method 988.05), fat (alkaline treatment with Röse–Gottlieb method 932.06 for MR; diethyl ether extraction method 2003.05 for starters and hay), Ca and P (dry ashing, acid digestion, and analysis by inductively coupled plasma, method 985.01), NDF with ash without sodium sulfite or α-amylase, ADF with ash, starch (α-amylase method), and sugar (colorimetric method), and the details of the nutrient composition are given in Tables S1 and S2.

#### *2.3. Determination Of VFA and NH3-N in Ruminal Fluid and Caecal Contents*

For the VFA and NH3-N measurements, the rumen fluid and caecal contents dissolved in the buffer were centrifuged at 13,000× *g* for 10 min. The VFAs were analysed on an Agilent 6850 gas chromatograph (Agilent Technologies Inc., Santa Clara, CA, USA) equipped with a polar capillary column (HP-FFAP, 30 m × 0.25 mm × 0.25 μm) and a flame ionization detector (FID), as previously described [10]. The NH3-N in the supernatant was quantified using a continuous-flow analyser (SKALAR San, Skalar Co., Breda, The Netherlands).

#### *2.4. Microbial DNA Extraction and 16S rRNA Gene Sequencing*

The ruminal fluid and caecal content samples from yak calves were subjected to DNA extraction using the QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany). The quantity and quality of those DNA samples were further assessed by a Nanodrop ND-1000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). The 16S rRNA gene amplicons of 8 DNA samples (4 samples from EW group and 4 samples from MG group) with high quality were used to determine the diversity and compare the community structures of the bacterial species in each of these samples using Illumina HiSeq sequencing at Novogene Bioinformatics Technology Co., Ltd., Beijing, China. The preparation of the amplicon library was performed by polymerase chain reaction amplification of the V3–V4 region of the 16S rRNA gene using the primer set 341F 5- -CCTAYGGGRBGCASCAG-3 and 806R 5- -GGACTACNNGGGTATCTAAT-3 with barcode. The identified sequences were deposited in the NCBI sequence archive (SRA) under the accession no. PRJNA552771.

Sequencing data splicing and quality filtering of the raw tags were performed using Trimmomatic (V0.36) and Usearch (V9.2.64) [11,12]. All sequences shorter than 200 bp and those with quality scores lower than 15 in the raw reads were removed, and high-quality clean tags were obtained. These sequences were classified into operational taxonomic units (OTUs) at an identity threshold of 97% similarity using UPARSE software [12]. For each OTU, by, a representative sequence was screened and used to assign taxonomic composition by comparison with the RDP 16S Training set (v16) and the core set using the SINTA (Usearch V9.2.64) and PyNAST (QIIME) programmed algorithms [13,14]. Subsequent analysis of alpha and beta diversity was performed based on the output of this normalized data. The taxon abundance for each sample was determined according to phylum, class, order, family, and genus. The *t*-test was performed to estimate the differential microbiota between the treatments. The threshold was set at *p* value < 0.05.

#### *2.5. Statistical Analysis*

Analysis was performed using Student's *t*test with SPSS 21.0 software with replicates as experiment units, and differences were considered statistically significant at *p* < 0.05.

#### **3. Results and Discussion**

#### *3.1. Early Weaning of Yak Calves with Alfalfa Hay, Starter Feed, and Milk Replacer Significantly Promoted Growth and Organ Development*

Compared with the yaks in the maternal grazing group, the yaks in the early weaning group showed significantly increased body weight, withers height, body length, and chest girth (Table 1). Additionally, the significantly increased weight of the liver, spleen, and thymus, as well as the significantly increased indexes of spleen and thymus (g/kg body weight) were also identified in the early weaning group (Table 1). Meanwhile, the ruminal fermentation characteristics of yak calves under the grazing and barn feeding conditions are presented in Table 2. The PH and NH3-N showed no differences between the different feeding groups. The total VFA concentration was significantly higher in the early weaning group than in the grazing group; of these, the propionate, butyrate, isobutyrate, and valerate were also significantly increased in the early weaning group (Table 2). Furthermore, the ratio of acetate/propionate and acetate/total VFA were significantly decreased in the early weaning group, while the ratio of butyrate/total VFA, isobutyrate/total VFA, and valerate/total VFA were all significantly increased in the early weaning group. Moreover, those significantly increased growth performance and ruminal fermentation were mostly resulted from the significant differences between the treatments in the daily DMI of yak calves, where the increased intake was found for calves on early weaning group, especially the increased intakes of concentrate supplement (Table 1).


**Table 1.** Effect of early-weaning feeding and maternal grazing feeding on body weight, body size indexes, and organ weight of yak calves.

a,b within a row with different superscripts means significantly difference.


**Table 2.** Effect of early-weaning feeding and maternal grazing feeding on caecal fermentation of yak calves.

a,b within a row with different superscripts means significantly difference.

Our results indicated that the early weaning yak calves provided with milk replacer, starter feed, and alfalfa hay during early life showed improved growth and development, in accordance with the results of previous studies on lambs during early life [5,6]. Supplementation of the diets of ruminants with carbohydrates such as alfalfa hay and starter feed during the pre-weaning period has a crucial long-term impact on ruminal fermentation in other ruminants that has been shown to be beneficial to their growth performance [15]. In accordance with the previous studies, DMI and ruminal VFA production were both significantly increased, which contributed to the significantly promoted growth performance of yak calves in early weaning group [5,6,10,15]. However, except for the VFAs from ruminal fermentation, the caecum VFAs produced accounted for 12% of total VFA production in sheep [9,16], while limited research focused on caecal fermentation in response to the early weaning with starter feed and alfalfa hay in yaks [7].

#### *3.2. Significantly Enhanced Caecel Fermentation Was Identified in the Yak Calves in the Early Weaning Group*

The caecal fermentation characteristics of yak calves under the grazing and early weaning conditions were further measured (Table 2). The pH and NH3-N also showed no differences between the two groups. The total VFA concentration was significantly higher in the early weaning group than in the grazing group (*p* = 0.026). Significantly higher concentrations of propionate, butyrate, and other VFAs were also identified in the early weaning group (*p* < 0.01), whereas concentration of acetate was not significantly altered by treatment in the present study. Meanwhile, the ratio of acetate/propionate and acetate/total VFA in caecum were also significantly decreased in the early weaning group, while the ratio of butyrate/total VFA and propionate/total VFA were both significantly increased in the early weaning group. Moreover, according to our results, we found that the concentration and production

of VFAs in the caecum could not be ignored when compared with the ruminal VFA concentration. In ruminants including the yak calves, the VFAs are absorbed by the rumen and caecal epithelium, and then metabolized into glucose, triglycerides, and amino acids and further provide energy and nutrients resources for the growth of yak calves. Our results indicated that changed caecal VFAs in early weaning, especially the increased acetate and propionate induced by supplementation of the diet with alfalfa hay and starter feed, can also be absorbed by the caecal epithelium [5,17], and then metabolized into glucose, triglycerides, and amino acids and further provide energy and nutrients resources for the growth of yak calves [7]. Moreover, the increasing ratio of butyrate/total VFA and propionate/total VFA further represented the improved energy utilization efficiency when compared with the acetate type fermentation and the promoted caecal development process.

#### *3.3. Di*ff*erent Responses of Caecal Microbiota to Early Weaning or Maternal Grazing Feeding Contribute to Enhanced Caecal Fermentation of Yak Calves*

Considering the significantly increased ruminal and caecal VFAs, the ruminal and caecal microbiota were both further analysed. The beta diversity analyses revealed that the compositions of the gastrointestinal prokaryotic community of the yak calves in two different feeding groups were significantly different (Figure 1A,B, and Figure 2A,B). Moreover, Chao1 indexes indicated that early weaning with the starter feed and alfalfa hay was beneficial to the diversity of ruminal and caecal microbiota (Table S3). Recently, several studies have focused on the effect of different feeding paradigms on the gastrointestinal microbiota of several animals, including lizards, cheetahs, yaks, lambs, deer mice, and seals [18–21]. These studies all identified that the diversity and abundance of the gastrointestinal microbiota were increased in animals from the wild environment than in captive animals. However, our study identified that early weaning and barn feeding significantly increased the diversity of ruminal and caecel microbiota. Different from the previous studies which focused on adult animals which obtained more varied nutrients in wild feeding paradigms, the early weaning in the barn feeding paradigm in the present study, which supplied the calves with milk replacer, starter feed, and alfalfa hay, provided enough carbohydrate, protein, and lipid for the growth and proliferation of microbes, which indicated that more microbial species could survive in the gastrointestinal tract due to the abundant sources of carbon and nitrogen [7,17]. In contrast, the yak calves from the maternal grazing group obtained limited nutrients from maternal milk and fresh grass, which resulted in less diversity and abundance of the ruminal and caecal microbiota. Moreover, the similar results of the increased diversity of ruminal and caecal microbiota also indicated that the dietary changes can simultaneously altered the ruminal and caecal microbiota, which also indicated that caecal microbiota could also have potential response to dietary supplementation and further influence the growth of yak calves. Overall, our results indicated that the nutrition supplementation is beneficial to the richness of caecal microbiota such as the alteration of the diversity of ruminal microbiota, which could even changeover the beneficial effect of a wild environment on the diversity and richness of gut microbiota. Our results also indicated that nutrition was the main effect among environmental indices which influence the gastrointestinal microbiota [22].

**Figure 1.** Ruminal microbial community difference between the different feeding paradigm groups (n = 4). (**A**) PCoA analysis. (**B**) Anosium analysis. (**C**) Differential ruminal microbes at phylum level based on *t*-test analysis. (**D**) Differential ruminal genera based on *t*-test analysis.

**Figure 2.** Caecal microbial community difference between the different feeding paradigm groups (n = 4). (**A**) PCoA analysis. (**B**) Anosium analysis. (**C**) Differential caecal microbes at phylum level based on *t*-test analysis. (**D**) Differential caecal genera based on *t*-test analysis.

The differential microbiota were further identified based on the counts of different microbes by using *t*-test analyses. In rumen, the significantly increased phylum of Proteobacteria, Fibrobacteres, Bacteroidetes, Actinobacteria, Spirochaetes, and SR1 (Figure 1C), as well as the genus of Succiniclasticum, Clostridium\_sensu\_stricto, Treponema, Escherichia/Shigella, Prevotella, and Fibrobacter (Figure 1D) were identified in the early weaning group. In caecum, the significantly increased Proteobacteria, Euryarchaeota, and Actinobacteria in the phylum level (Figure 2C), as well as significantly increased genera of Turicibacter, Methanobrevibacter, Clostridium\_sensu\_stricto, Bacteroides, Clostridium\_XlVb, Clostridium\_XI, Escherichia/Shigella, Oscillibacter, Ruminococcus, Blautia, Clostridium\_IV, and Prevotella (Figure 2D) were identified in the early weaning group. Accordingly, the phylum of Proteobacteria, Euryarchaeota, and Actinobacteria and the genera of Clostridium\_sensu\_stricto, Escherichia/Shigella, and Prevotella were co-influenced by the early weaning with alfalfa hay, starter feed, and milk replacer, which were all involved in the utilization of fibrous and non-fibrous carbohydrates and the production of propionate and butyrate. Meanwhile, these results again proved that dietary alteration could have a similar effect on the ruminal and caecal microbiota. Moreover, the main finding of our study lies in the fact that supplemental feeding with alfalfa hay and starter feed exceeded maternal grazing and nursing in shaping hindgut functional achievement. The significantly increased caecal genera of yak calves identified in the early weaning groups, including the *Prevotella*, *Clostridium\_XIVb*, *Turicibacter*, *Clostridium\_IV*, *Clostridium\_XI*, *Clostridium\_sensu\_stricto*, *Bacteroides*, *Oscillibacter* and, *Ruminococcus* were mainly involved in the utilization of fibrous and non-fibrous carbohydrates and the production of acetate, propionate, and butyrate [17,23–27]. The primary determinant for this could be that the early-weaning calves consumed a greater amount of concentrate and alfalfa hay, and dietary fiber and starch were the suitable fermentation substrate when they reached the hindgut in significant quantities. Considering the identified significantly increased VFAs in the present study, the effect of early weaning with alfalfa hay and starter feeds on the identified variational microbiota and the roles of these changed microbiota were again proved. According to previous studies, in goats during the early life, caecal propionate, butyrate, and isobutyrate concentrations also significantly increased in response to a grain-rich diet [7,17,28]. In accordance with these previous studies, significantly higher concentrations of propionate, butyrate and total VFAs were also identified in the early weaning group of the present study, which were produced by our identified differential microbes by using starch or fibrous carbohydrates. Moreover, the effect of differential supplementing of carbohydrates during early life, induced by early weaning with alfalfa hay and starter feed, on the subsequent gastrointestinal microbiota and the related caecal fermentation, could further increase the absorbed VFAs from the caecal epithelium, and further provide more energy for the growth of yak calves [29,30]. Overall, except for ruminal fermentation, caecal fermentation could also be enhanced by providing enough fermentable carbohydrates in the EW group, which was induced by the increased abundance of microbes involved in the utilization of fibrous and non-fibrous carbohydrates and subsequently increased; and then the increased caecal VFAs could contribute to promoting the growth of yak calves.

#### **4. Conclusions**

Early weaning and barn feeding with milk replacer, alfalfa hay, and starter feed is recommended during pre-weaning to improve yak calf growth performance. Except for their beneficial roles in ruminal microbiota construction and ruminal VFAs production, the facilitating caecal starch-using and fibre-using microbial colonization and the subsequently improved caecal fermentation can also contribute to the growth of yak calves, which may play similar roles to the changed ruminal microbiota.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-2615/9/11/921/s1, Table S1: Nutrient composition of the alfalfa, starter feed, and milk replacement used in the present study; Table S2. Nutrient content of fresh grass and yak milk for yak calves from maternal grazing group; Table S3 Effect of early-weaning feeding and maternal grazing feeding on caecal microbial alpha diversity index of yak calves.

**Author Contributions:** S.W., Z.C., and J.Y. conceived and designed the experiments; S.W., Z.C., and X.C. mainly performed the experiments; S.W. analyzed the data; J.Y., S.W., and Z.C. contributed reagents/materials/analysis tools; S.W. wrote the manuscript. J.Y. and S.W. had primary responsibility for final content. All authors (including P.W.) read and approved the final manuscript.

**Funding:** This work was funded by the National Key R&D Program of China (2016YFC0501805), the China Postdoctoral Science Foundation (2019M653774), the Thousand-person Plan of Qinghai High-end Innovative Talents (top-notch talents Train), and the Program for Qinghai Science & Technology (3-4), and the National natural science foundation of China (31902184).

**Conflicts of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **E**ff**ects of Body Condition Score Changes During Peripartum on the Postpartum Health and Production Performance of Primiparous Dairy Cows**

#### **Yujie Wang, Pengju Huo, Yukun Sun and Yonggen Zhang \***

College of Animal Science and Technology, Northeast Agricultural University, Harbin 150030, China; wangyjanet@163.com (Y.W.); huopengju@163.com (P.H.); sun\_yukun@126.comg (Y.S.) **\*** Correspondence: zhangyonggen@sina.com

Received: 8 November 2019; Accepted: 9 December 2019; Published: 17 December 2019

**Simple Summary:** This study systematically describes the effects of body condition score (BCS) changes in primiparous cows during the peripartum period on hormone indexes, health, and production. The BCS and its changes indirectly measure the degree of fat mobilization and is a good predictor of the risk of postpartum disease. In production practice, confounding the management of primiparous and multiparous cow risks neglecting the postpartum characteristics of primiparous cows. A prospective observational study observed that primiparous cows that have a lower BCS have higher non-esterified fatty acids (NEFA) and β-hydroxybutyrate (BHBA) concentrations and more dramatic hormonal changes. Prepartum BCS changes were inconsistent and small, while after calving, there was a drastic decline in the BCS, suggesting that even a slight drop in the prepartum BCS may be a warning of a postpartum risk for primiparous cows. It is suggested that operators attach importance to the primiparous cow prepartum BCS and keep it stable through prepartum management adjustments, since an ideal BCS at calving reduces the incidence of postpartum disease.

**Abstract:** This is a prospective observational study that evaluates the effects of body condition score (BCS) changes in primiparous Holstein cows during peripartum on their NEFA and BHBA concentrations, hormone levels, postpartum health, and production performance. The cows under study (*n* = 213) were assessed to determine their BCS (5-point scale; 0.25-point increment) once a week during the whole peripartum by the same researchers; backfat was used for corrections. Blood samples were collected 21 and 7 days before calving and 7, 21, and 35 days after calving, and were assayed for NEFA, BHBA, growth hormone (GH), insulin, leptin, and adiponectin concentrations. The incidence of disease and milk yield were recorded until 84 days after calving. Cows were classified according to their BCS changes during peripartum as follows: Those that gained BCS (G; ΔBCS ≥ 0.25), maintained BCS (M; ΔBCS = 0–0.25), or lost BCS (L; ΔBCS ≥ 0.5). The BCS at −21 days and at 7, 14, and 21 days were different *(p* < 0.01), but trended toward uniformity in all groups at calving. The L group had higher NEFA and BHBA concentrations and hormone levels (*p* < 0.01) than the M and G groups at 21 and 35 days after calving, and had a higher incidence of uterine and metabolic diseases; however, there were no differences in production performance between the various groups. In conclusion, a lower BCS in primiparous cows during peripartum influences the NEFA and BHBA concentrations, hormone levels, and occurrence of health problems postpartum. The postpartum effects of BCS changes appear prior to calving.

**Keywords:** body condition score; peripartum; fat mobilization; primiparous dairy cow

#### **1. Introduction**

For a cow approaching calving, the periparturient period, from three weeks prepartum to three weeks postpartum, is an important stage that determines whether milk yield and dry matter intake (DMI) will rapidly increase postpartum; recovery of postpartum DMI is an important measure for avoiding a negative energy balance (NEB) [1,2]. Numerous metabolic and hormonal changes, together with a series of stress reactions, such as calving, lactating, and ration changes, involving feeding management during this period, have a direct effect on the health, reproduction, and lactation performance of cows [3,4]. Parity is a well-known risk factor for disease: Multiparous cows are more likely to develop ketosis and hypocalcemia [5,6]; the evolution of metabolic profiles in healthy and sick cows during the periparturient period varies according to parity [7]. Studies have found that primiparous cows have higher concentrations of insulin-like growth factor-I, lower concentrations of BHBA throughout periparturient, and higher concentrations of leptin; these differences are associated with significantly lower milk production and body condition scores [8,9]. These results suggest that the management of primiparous cows during the periparturient period should be different.

Body condition score (BCS) is strongly correlated with energy reserves, directly reflecting the fat reserves of individual dairy cows. Changes in the BCS rather than a single BCS measurement, which is frequently used to monitor energy balance as a practical tool for dairy farm management, are widely used and easy to determine [10]. The peripartum BCS and a series of changes, including the BCS at calving and the rate and degree of BCS reduction after calving, may indicate the increase of non-esterified fatty acids, possibility of postpartum diseases and differences in production performance. Studies have shown that the BCS of multiparous cows can be regarded as a prediction tool due to the strong association between the BCS and metabolic diseases, including hepatic lipidosis, ketosis, and abomasum displacement [11–13]. The main reason for this strong association is that weight loss over 50 kg due to improper prepartum feeding significantly inhibits DMI and milk production; meanwhile, high milk production and the consequent synthesis of milk fat result in a high degree of fat mobilization, causing cows to go through NEB [14]. Health and performance at the primiparous stage have a profound impact on later stage incidence of disease and production potential; thus, reasonable management of cows' body condition is particularly important. Little is known about the characteristics and the reference range of the BCS in primiparous cows on disease prediction, hormonal levels, and lactation performance. Considering that primiparous cows represent a high proportion of cows in production, it is particularly necessary to further study the prepartum BCS of primiparous dairy cows and to use a hormonal index to specifically investigate the influence of BCS changes on fat metabolic, health, and lactation.

Adipose tissue reserves are predominantly controlled by the energy balance and abundance of insulin, with the expression and tissue responsiveness of key hormones being altered to maintain physiological equilibrium at the beginning of chronic energy deficiency [15]. Growth hormone (GH) directly regulates ruminant adipose stores. Insulin is an antagonist of the lipolytic actions of GH and lowers mobilization of the tissue reserves. Adiponectin is recognized to play an important role in metabolic syndrome. Leptin serves as an intake satiety signal by predominantly acting on the brain [10]. The somatotropic axis, primarily consisting of growth hormone (GH; somatotropin) and insulin-like growth factor-I (IGF-I; somatomedin), is essential for the regulation of intrahepatic lipid metabolism [16]. Association between peripartum BCS, fat mobilization, and postpartum serum insulin concentration has been demonstrated in several studies [17–20]; besides, an increase in serum concentrations of NEFA and BHBA, as well as decreased serum concentrations of insulin and glucose, are indicators of NEB [21]. Studies convincingly demonstrate that exogenous bovine somatotropin (BST) results in an increase in milk yield in treated animals and results in a series of coordinated adaptations in their body tissues to support the increased use of nutrients for milk synthesis [22]. A recent study exploring the association between postpartum plasma insulin and NEFA and BHBA concentrations demonstrated that cows with low plasma insulin had significantly higher concentrations of circulating NEFA; moreover, cows with low plasma insulin during early postpartum produced more milk and

had higher FCM (fat corrected milk) or ECM (energy corrected milk) compared with cows with high plasma insulin [23]. The serum adiponectin concentration was positively associated with the insulin responsiveness of glucose and NEFA metabolism [24]. Most of the studies that investigated hormone concentration in the context of fat and glucose metabolism used multiparous cows in their experiments; however, differences and changes in fat metabolic hormone levels among primiparous cows, especially the direct effects mechanism, are not fully understood and deserve further investigation.

This study highlights the need for the preferential treatment of primiparous cows to ensure that their BCS trajectory is sufficient for calving, and the need to adopt a management strategy for adjusting the prepartum BCS to maximize the prevention of postpartum disease and production potential during later stages. Furthermore, it was hypothesized that testing the levels of the key regulatory hormones related to fat mobilization would result in a better understanding of the factors that influence BCS mobilization and replenishment. The objectives of the current study were to evaluate the association between BCS changes and hormone levels during peripartum and postpartum with the health and performance characteristics of primiparous lactating Holstein cows, and to examine the herd- and cow-level factors that influence the BCS profile, thus promoting animal management aimed at improving farm productivity, profit, and animal welfare.

#### **2. Materials and Methods**

#### *2.1. Animals and Management*

This prospective observational experiment was conducted on a commercial farm in Harbin City, Heilongjiang Province, China from August 2018 to January 2019. Two hundred and thirteen primiparous cows from a total of 692 lactation cows met the enrollment criteria and were used in the current study. All cows were synchronized using a Double-Ovsynch protocol for first TAI (Timed Artificial Insemination) with a progesterone implant during the Ovsynch. Protocol of synchronization started when cows had 60–65 days in milk. Primiparous cows entered the peripartum period during August and September. The calving date was synchronized to be in fall, during September and October; thus, there was no calving month effect. Milk yield data were collected from 5 to 84 days after calving. Disease records were kept to 84 days postpartum. The collection of postpartum data was completed in January. The living environment of the experimental primiparous Holstein cows (*n* = 213) during peripartum was consistent: Cows were housed in free-stall barns that included mattresses composed of rice hulls, and were equipped with self-locking head gates at the feed line and with a cross-ventilation system with fans and spray devices. One to 14 days after calving, cows were relocated to the fresh-cows cowshed for postpartum care. Cows were milked three times a day at 7:00 a.m., 2:30 p.m., and 10:30 p.m. The milk yield of individual cows was recorded and stored in the software of the automatic milking system. Cows were fed total mixed ration (TMR) formulated according to NRC (2001) to meet the nutritional requirements of each period (pre- and postpartum). Prepartum cows were fed beginning at 4 p.m., fresh cows were fed at 7:35 a.m., with ad libitum access to food and water. All cows were synchronized using a Double-Ovsynch protocol for the first TAI with a progesterone implant during Ovsynch. The synchronization protocol was started when cows were in milk for 60–65 days.

#### *2.2. Collection, Treatment, and Assessment of Blood Samples*

Blood samples were collected throughout the experimental period relative to parturition at −21, −7, 7, 21, and 35 days from coccygeal vessels using a sterile syringe prior to the morning feeding. The samples were collected in a 5 mL evacuated centrifuge tube containing heparin anticoagulant and centrifuged at 2000× *g* for 15 min. Supernatant was collected, aliquoted into 1.5 mL centrifuge tubes, and stored at −20 ◦C for the measurement of blood biochemical and hormonal indexes, including NEFA, BHBA, GH, insulin, adiponectin, and leptin. The analyses were performed using commercial

kits (Jiang Lai biotechnology co. LTD, Shanghai, China) by the enzymatic colorimetric endpoint method. The values were recorded using a Switzerland Tecan multifunctional microplate reader.

#### *2.3. Assessment of the BCS and Classification*

The BCS were recorded at the beginning of peripartum (21 d before calving) and scored weekly throughout peripartum at 21, 14, and 7 days before calving; the day of calving; and 7, 14, and 21 days after calving. The BCS assessment was completed by a doctor and a master trained by the College of Animal Science and Technology, Northeast Agricultural University. These personnel had also completed four-level course training, had practical operation experience in a commercial farm, and used the visual and tactile technique to determine the BCS based on the US 5-point system with a 0.25 increment, with 1 being too thin and 5 being too obese. Each cow was evaluated on each occasion by both researchers. Before each morning feeding, cows were kept in a normal standing posture [25]. To guarantee the accuracy and objectivity of the final BCS, a portable ultrasound backfat instrument was used to measure the fat thickness of the rump at each BCS assessment, and developed a linear regression model of BCS according to US 5-point system rules to determine BCS. The maximum penetration depth of the backfat instrument probe was 10 cm. The probe was placed vertically at 1/4 to 1/5 of the line connecting the ischial tuberosity and the hip tuberosity, and all values were measured on the right side. The final BCS of each cow was calculated using the average of three data points. Cows were classified according to BCS changes during peripartum (BCS at 21 days minus BCS at −21 days). The classifications included: Gained BCS (Gained; ΔBCS > 0), maintained BCS (Maintained; −0.25 ≤ ΔBCS ≤ 0), and lost BCS (Lost; ΔBCS < −0.25).

#### *2.4. Disease Definition*

Data on health status (mastitis, metabolic and digestive disorders, and metritis) were collected from the day of calving to 84 days in milk. Early warning of disease was based on a system that was fitted to the cows: A neck-mounted electronic rumination and activity monitoring tag (HR Tags, SCR Dairy, Netanya, Israel). At all times, cows that the tag identified were examined to establish a preliminary diagnosis by the same personnel. Information on the treatment and collected data were stored on the on-farm software. Fresh cows were observed daily from calving to 14 days postpartum until relocation to a cowshed for healthy cows. Cows with health disorders were subjected to required milk withdrawal and placed in a separate pen, and their milk was discarded until it became saleable. The clinical examination included a direct observation (general appearance and attitude, muscle strength, presence of fetal membranes outside the vulva, evaluation of vaginal discharge, foot health, udder health, and manure consistency), rectal temperature, urinary ketones, and rumen auscultation.

Cows were evaluated for metritis postpartum by palpation. Metritis was characterized by an enlarged uterus with a fetid watery red–brown discharge within 21 days postpartum. The rectal temperature was measured for cows with metritis, and those with a temperature of 39.5 ◦C were diagnosed with puerperal metritis. Abortion was defined as failure to deliver a normal calf. Retained fetal membrane was defined as failure to detach fetal membranes within 24 h postpartum. At every milking, all cows were examined for signs of clinical mastitis by the herd personnel immediately before milking: Clinical mastitis was characterized by the presence of abnormal milk or by signs of inflammation in one or more quarters. The herd personnel milked three handfuls and checked whether characterized by the presence of abnormal milk. A case of milk fever was defined as a prostrated cow with minimal rumen contractions that responded to an intravenous calcium treatment within 30 min. Cows with a decreased appetite and altered patterns of milk production had their urine tested for ketone bodies (Keto-Stix, Bayer Diagnostics, Tarrytown, NY, USA), and those that tested at or above moderate were diagnosed with ketosis. Cows with a metallic (ping) sound at percussion auscultation of the left or right abdomen (between the 4th and 13th ribs) were diagnosed with a displacement of the abomasum. Cows with scant manure, lack of appetite, and rumen stasis were diagnosed with indigestion. Respiratory disease was characterized by panting, rectal temperatures >39.5 ◦C, crackling,

rales, or percussion dullness when auscultating the lungs. Cows with traumatic events (cesarean section, udder/teat cuts, and broken limbs) were excluded from the study. Occurrences of retained fetal membranes, abortion, and metritis were grouped into one variable: Uterine disease. Milk fever, ketosis, and displacement of the abomasum were grouped into one variable: Metabolic disease. Cows with a diagnosis of respiratory disease and cows with undefined sickness were grouped into the category: Other diseases.

#### *2.5. Statistical Analysis*

This study was a prospective observational study. The BCS at various times and the NEFA, BHBA, GH, insulin, leptin, and adiponectin concentrations were analyzed by GLM using MIXED PROC using SPSS software (version 22.0, IBM SPSS Statistics, Chicago, state of Illinois, USA). The model included the fixed effects of the experimental group (G, M, or L), the fixed effect of week in lactation, and interaction of the groups by week in lactation. A BCS of −21 days was used as covariance. To verify significant differences between the groups, data were analyzed using ANOVA in SPSS. Milk yield data were processed using ANOVA in SPSS. Significance was declared at *p* < 0.05 unless otherwise indicated. The distribution of the BCS in cows between various groups at day −21 was processed using Microsoft Office Excel. Health event statistics were recorded using Microsoft Office Excel (version MSO 16.0, Microsoft, Redmond, state of Washington, USA).

#### **3. Results**

Primiparous cows were divided into three groups based on changes in the prepartum BCS: Gained BCS (Gained, G), maintained BCS (Maintained, M), and lost BCS (Lost, L); the proportions of each group were 15.96% (34/213), 30.99% (66/213), and 53.05% (113/213), respectively. The BCS at −21 days were different; the mean BCS (± SEM) at −21 d were 3.09 ± 0.06, 3.39 ± 0.03, and 3.45 ± 0.02 for the G, M, and L groups, respectively. The L group had the highest BCS (3.45), followed by the M group (3.39). The G group (3.09) had a mean BCS lower than the other groups (*p* < 0.01; Table 1). Cows had similar BCS on days −14, −7, and 0. However, the postpartum BCS were different from the prepartum BCS: The L group had the lowest BCS versus the BCS of the other two groups (*p* < 0.01) at 7 days. The BCS at 14 and 21 days had very significant differences between groups.


**Table 1.** Comparison of body condition score (BCS; least squares means ± SEM) on days 21, 14, 7 before calving, in relation to calving, and 7, 14, and 21 after calving for primiparous cows in different groups.

a–c Values within a row with different superscript letters differ at *p* < 0.05. Cows had their BCS evaluated during the transition period (−21 to 21) using a 5-point scale with 0.25 increments. G, gained BCS; M, maintained BSC; L, lost BCS.

The G group experienced a slow rise in the BCS throughout the prepartum period, increasing significantly from −21 days to calving and remaining essentially constant postpartum. Cows that gained BCS had a lower BCS at −21 days. In contrast, the L group had a declining trend over the entire prepartum period: Cows that lost BCS had the highest BCS at the beginning, followed by a slow downward trend and a more dramatic decline after calving. High BCS cows entering the prepartum

period had a higher likelihood of losing BCS. This slight decrease was apparent even before calving (Figure 1a). Moreover, when entering the prepartum period, the average BCS of the G group cows was below 3.25. The BCS of the other groups were higher than 3.25. All the BCS were higher than 3.25 at −21 days, except that of the G group. The L group had a higher percentage of cows with BCS greater than or equal to 3.25 (*p* < 0.01; Figure 1b) than the other groups, suggesting that cows with a lower prepartum BCS at −21 days are more likely to have an increased BCS during the prepartum period.

**Figure 1.** (**a**) Comparison of variation trend of body condition score (BCS) on days 21, 14, 7 before calving, calving day, and 7, 14 and 21 after calving for primiparous cows in different groups. (**b**) Distribution of cows that G (*n* = 34), M (*n* = 66), and L (*n* = 113) groups on −21 days BCS.

The NEFA and BHBA concentrations in the three experimental groups are presented in Table 2. The NEFA and BHBA concentrations differed (*p* < 0.01) between groups at postpartum 21 and 35 days. The L group had higher concentrations compared with those in cows that gained or maintained BCS. NEFA and BHBA did not change in a time-dependent manner. There were group-time interaction effects on NEFA (*p* < 0.01) and BHBA (*p* = 0.02). The changes during the prepartum period are shown in Figure 2. The variation trends of the three indicators in the L group had a common feature: The prepartum concentration was slightly higher than that in the other two groups and reached its lowest value on 7 days; then, the concentration sharply increased to significantly higher levels than those of the other two groups, reaching a maximum on 21 days. In the G group, the fluctuations in the

prepartum changes were higher than those during postpartum. Cows that maintained BCS showed a smaller change with only slight fluctuations.


**Table 2.** Comparison of NEFA and BHBA contents in different groups of primiparous cattle before and after delivery.

a,b Values within a row with different superscript letters differ at *p* < 0.05.

**Figure 2.** (**a**) Serum NEFA (upper panel) and (**b**) BHBA (lower panel) concentrations (least squares means ± SEM) in different groups during prepartum period.

The changes in hormone indexes during peripartum are shown in Table 3. The hormone levels of the groups were highly significantly different (*p* < 0.01), but time had no effect. The GH and adiponectin concentrations exhibited significant difference in group-time interaction (*p* = 0.04). At 21 and 35 days, the L group values were significantly higher than those of the other two groups with regard to GH, leptin, and adiponectin, while the insulin concentrations differed only at 35 days. Each hormone index in the L group showed the most dramatic changes during the prepartum period, with the lowest concentrations observed at 7 days and the highest concentrations at 21 days. The insulin concentrations continued to show an upward trend after 21 days. The GH and insulin indexes of the G group dramatically changed, while the hormone indexes in the M group remained essentially unchanged (Figure 3). The concentrations in all three groups tended to be consistent at 7 days.


**Table 3.** Comparison of growth hormone (GH), insulin, leptin, and adiponectin contents in different groups of primiparous cattle before and after delivery.

a,b Values within a row with different superscript letters differ at *p* < 0.05.

**Figure 3.** *Cont.*

**Figure 3.** Serum GH (**a**), insulin (**b**), leptin (**c**), and adiponectin (**d**) concentration (least squares means ± SEM) in different groups during prepartum period.

The postpartum incidence of disease in cows that gained, maintained, and lost BCS is presented in Table 4. Cows that gained and maintained BCS had fewer health problems than cows that lost BCS. Moreover, when we evaluated cows with one or more health problems, cows that gained and maintained BCS had fewer health events than cows that lost BCS. Milk yield was similar among the experimental groups, and there was no group–time interaction. As reported in Table 5, the cows averaged 24.89 kg/d; however, the time of lactation influenced milk production (*p* < 0.01).


**Table 4.** Postpartum incidence (%) of health problems for primiparous cow in different groups.

**Table 5.** Milk yield (least squares means ± SEM) for primiparous cow in different groups.


#### **4. Discussion**

The BCS and its changes are used as indirect indicators to measure fat mobilization and energy balance in individual cows and as a good predictor of the risk of disease. To our knowledge, most studies focus on high-yield multiparous cows, and little is known about the health status, milk yield, hormonal levels, and interrelationships with BCS changes in primiparous cows. A novel contribution of this study is its summary of BCS changes during peripartum in primiparous cows and its emphasis on the effects of varying management of BCS changes on fat mobilization, milk yield, and health between primiparous and multiparous cow. Blood biochemical and hormone indices were analyzed, focusing on primiparous cow management before calving. It is commonly accepted that milk production gradually increases after calving and reaches a peak at approximately 4 weeks postpartum. By contrast, changes in the BCS are inversely related to the lactation curve; however, the use of BCS values as a management tool can be enhanced when the prepartum period is added to the analysis [10,26,27]. Interestingly, it was found that the trend of changes in the prepartum BCS was different from that in the early postpartum lactation period, showing irregular variability. In actual production, it is generally recommended that modern high productivity dairy cows have a moderate BCS (≥3.25 and <3.5) at the beginning of the peripartum period, consequently resulting in lower mobilization of body reserves [28]. Similarly, our results show that primiparous cows with a higher BCS (3.40) at the outset had higher fat mobilization that was specifically reflected in higher NEFA and BHBA concentrations and hormone levels after calving. Surprisingly, the BCS of these cows tended to be consistent at calving, despite the initial variability, implying that the BCS is likely to reach the ideal body condition for calving through prepartum management adjustments, ensuring a suitable BCS prior to calving to minimize fat mobilization.

Ricardo C. et al. (2017) [29] found that BCS changes are strongly related to the BCS at dry off. Dechow et al. (2002) [30] evaluated correlations between the BCS and loss of BCS and found that a

higher BCS at calving was phenotypically associated with a higher loss of BCS during early lactation. Our results show that BCS changes during early lactation are largely dependent on the BCS at −21 days, that is, the same as previous finding stated; but have no association to calving BCS because the BCS values at calving were similar between groups. The most important is that the similar BCS for calving comes from changes during prepartum, indicating that change in prepartum BCS plays an important role in the change of early lactation BCS, which determines the change of postpartum. Cows that lost BCS had the highest BCS at −21 days; in contrast, the BCS of cows that gained BCS as lower. This effect is similar to multiparous cows and it is more likely to happen in primiparous cows: Roche et al. (2009) [10] suggested that it is easier for primiparous cows to reach the optimal BCS and have better results in their metabolic and hormonal profiles than multiparous cows. Due to different nutrient partitioning, these cows have a high requirement to maintain good appetite and DMI for continued body growth and development [9,31], leading to a uniform BCS during calving. Moreover, these cows do not go through the drying milk stage and avoid stress caused by changes in feed structure.

Previous studies have confirmed that avoiding BCS loss and maintaining energy balance have positive effects on the fertility, health, and performance of lactating dairy cows. R.V. Barletta reported that a BCS loss was initiated even prior to calving. Our results are clearly consistent with this observation and indicate that BCS changes have an identifiable trend during the prepartum period [32]. The general conclusion is that cows with a relatively consistent BCS exhibit approximately similar changes. The postpartum changes were generally consistent with prepartum changes. The prepartum changes exhibited only a slight change, while the postpartum changes were especially evident in cows that lost BCS. However, there were no significant differences in NEFA, BHBA, or hormone indexes during the prepartum period, indicating that a loss of the prepartum BCS did not involve substantial fat mobilization. Higher circulating concentrations of NEFA, BHBA, and hormones postpartum and a BCS loss happened at the same time, suggesting that primiparous cows are in a parturition stress environment that causes a sharp decrease in DMI, as suggested by Proudfoot et al. (2009) [33]. The majority of studies have reported that DMI is a major driver of variations in the BCS during the early postpartum period. In contrast, insulin and GH are associated with growth [34] and a higher BCS [10], which may explain, in part, the higher concentrations of these metabolites in cows that lost BCS in our study. Additionally, insulin plays a role in energy metabolism, and its concentration is positively correlated with energy intake; however, this relationship has been observed in multiparous cows but not in primiparous cows [8], as confirmed by our experiments.

Usually, circulating NEFA concentrations and DMI have an inverse relationship and are associated with negative effects on health and production. Hayirli et al. (2002) [35] demonstrated that cows classified as obese 21 days before the expected calving date had lower DMI (as a percent of BW, body weight) from 21 days before the expected calving date to calving compared with thinner cows. The reduction in DMI from 21 days before the expected calving date to calving was 40, 29, and 28% for obese, moderate, and thin cows, respectively. In our study, cows that lost BCS had higher levels of NEFA and BHBA than those in the other groups. The effects remained significant after calving, thus confirming that obese cows have low DMI, significant reductions in BCS, and underwent higher fat mobilization after calving. Parity is a well-known pivotal factor for differences in feed intake. Studies have shown that cows with different parity have different characteristics in terms of milk yield, incidence of metabolic disease postpartum, and reproductive performance, with significant differences in feeding habits and behavior during the prepartum period. Primiparous cows have a lower dry matter intake, eat more slowly, are replaced at the feeder more frequently and are typically smaller than multiparous cows. Proudfoot et al. (2009) [33] showed that primiparous cows ate less than multiparous cows during weeks 1–2 after calving. Similarly, H. W. Neave found that even after controlling for BW and milk production, primiparous cows ate less than multiparous cows, with the differences increasing over the postpartum period [36]. Primiparous cows have higher pregnancy rate, lower milk yield, and higher glucose receptor sensitivity than multiparous cows, as well as their physiological functions are vigorous, hormone levels are adequate, there is almost no disorders of

glycometabolism and metabolic disease, which is also proven in the study that no ketosis among all the experimental cows. Additionally, cows that lost BCS had higher adiponectin, implying fat mobilization. C. Urh (2018) [37] confirmed the involvement of adiponectin in the regulation of energy partitioning in primiparous cows, with adiponectin concentrations higher than those in multiparous cows. Therefore, nutritional management should be differentiated from multiparous cows that high nutrient and energy concentration to avoid excessive fat in body condition and low DMI postpartum. As for BCS management, it is better to keep a moderate BCS, about 3.25, to enter the peripartum period; higher BCS cows show poor health and performance postpartum, and even a slight drop in the prepartum BCS can be a warning of a postpartum risk of low DIM in primiparous cows. A slight decrease in the prepartum body condition is correlated with a BCS loss after calving and with the magnitude of postpartum NEB.

The increased energy requirements due to lactogenesis and reduced dry matter intake mean that all dairy cows undergo a state of negative energy balance (NEB) from late gestation to early lactation. Additionally, feeding peaks appear after lactation peaks, forcing cows to mobilize their fat reserves or proteins to meet their needs. These problems are increasingly understood to be rooted in DMI 2–3 weeks before calving, arguing for the importance of nutritional management in the prepartum period [38]. Although DMI was not included in this study's statistics, cows that lost a condition had an increase in leptin, demonstrating that a decrease in DMI leads to lower energy storage levels in the group that lost BCS. At the same time, an increase in the GH concentration meets the higher requirement for body growth and development and fat metabolism. The data indicate that the concentration of insulin increases in cows that lost BCS; however, it is possible that a negative energy balance causes the cows to become insulin resistant, an early warning sign of metabolic disease. One of the purposes of this study was to identify hormone indicators that can efficiently predict changes in the body condition score before calving and can be used as an effective tool for nutritional management; however, these results do not show unified and regular changes in the hormones as a clear means of guiding management.

Our study showed that cows with higher losses of BCS during the peripartum period had higher incidences of abortion, metritis, obstetric canal strain, retention of the fetal membrane, and mastitis during early lactation compared with cows that maintained or gained BCS. Meanwhile, cows that lost BCS had higher risks of lameness and milk fever and were more likely to have more than one health event. In general, cows that maintained BCS had better health than cows that gained or lost BCS. Health status is also associated with elevated concentrations of NEFA and BHBA. An increase in the circulating concentration of NEFA 7 to 10 days prepartum and of BHBA in the postpartum period can indicate metabolic problems, indirectly related to multiple common diseases of energy metabolism [39,40]. Excessive fat mobilization and a long period of NEB can attenuate the function of the immune system [41]. Ospina et al. (2010) [40] demonstrated that an increase in prepartum and postpartum NEFA levels is associated with an increased risk of retained fetal membranes, metritis, clinical ketosis, and displacement of the abomasum. It can be speculated that cows that lost BCS had a negative energy balance and, consequently, an impaired immune response during the periparturient period, possibly predisposing them to metabolic disorders. It should be noted that cows that lost BCS postpartum had higher insulin levels, which is a key adaptive mechanism that raises a concern for the development of insulin resistance and changes in insulin responsiveness [37,42]. Furthermore, primiparous cows experience a suite of stressful events that they have not experienced previously, including regrouping, diet changes, parturition, and the onset of lactation, resulting in reduced food intake and excessive NEB [36].

Intensive genetic selection to increase milk production increases the demands for dietary nutrients and body tissue reserves, resulting in poor health and infertility [43]. The changes in nutrient metabolism required to support lactation in high producing dairy cows are controlled by hormones that coordinate a variety of processes. If the nutritional environment is adequate, cows can meet their energy demands from DMI, and consequently, tissue mobilization will be minimized [10]. Insulin is an especially powerful mediator of a number of various physiological effects, most of which serve to acutely maintain metabolic equilibrium in the face of short-term variations in nutrient supply and demand. Insulin levels were higher after calving in cows that lost BCS: Fewer nutrients are directed to body fat reserves and other non-mammary tissues because of the altered response to insulin, and more nutrients are taken up by the mammary gland consistent with an increase in milk synthesis [44]. However, there were no significant differences in milk production between cows that gained, maintained, or lost BCS. Cows that lost BCS did not have a better production performance.

#### **5. Conclusions**

In conclusion, it may be easier for primiparous cows to attain an ideal BCS at calving through sufficient prepartum management adjustments, even though a slight drop in the prepartum BCS may constitute a warning of a postpartum risk of large changes in the BCS, greater health problems, and a poor productive performance. Meanwhile, simultaneous increases in hormone levels in cows that lost BCS allows the prevention of excessive lipolysis and metabolic disease, suggesting that their vigorous physiological function and adequate hormone secretion is enough to sustain postpartum demand. In production, it is recommended a moderate BCS of approximately 3.25 for primiparous cows entering peripartum, keeping BCS stable during the prepartum period. Attention should be paid to primiparous cows in the prepartum stage through efficient management, avoiding excessive nutrition, leading to a higher BCS, possibly ensuring higher postpartum DMI and enabling the maximization of potential productivity.

**Author Contributions:** Conceptualization, Y.W. and Y.S.; methodology, Y.W.; investigation, P.H.; writing—original draft preparation, Y.W.; writing—review and editing, Y.S. and Y.Z.; supervision, Y.Z.; project administration, P.H.; funding acquisition, Y.Z.

**Funding:** This research was funded by Yonggen Zhang, grant number China Agriculture Research System (CARS-36).

**Acknowledgments:** This study was financially supported by the China Agriculture Research System (CARS-36).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **The Limiting Sequence and Appropriate Amino Acid Ratio of Lysine, Methionine, and Threonine for Seven- to Nine-Month-Old Holstein Heifers Fed Corn–Soybean M-Based Diet**

## **Yuan Li 1, Yanliang Bi 1, Qiyu Diao 1, Minyu Piao 1, Bing Wang 2, Fanlin Kong 1, Fengming Hu 1, Mengqi Tang 3, Yu Sun <sup>3</sup> and Yan Tu 1,\***


Received: 19 April 2019; Accepted: 29 September 2019; Published: 30 September 2019

**Simple Summary:** Research on the amino acid nutrition of cattle is limited, particularly research on the amino acid patterns of growing heifer. This lack of research has made it difficult to minimize the costs and reduce nitrogen emission of dairy heifers. Lysine might be the first limiting amino acid for seven- to nine-month-old Holstein heifers that are fed a corn–soybean meal-based diet, followed by methionine and threonine. The appropriate ratio of lysine, methionine, and threonine—calculated based on the nitrogen retention of seven- to nine-month-old Holstein heifers—were 100:32:57. We expect to reduce the input of protein feed and nitrogen emissions for dairy farms by using this ratio.

**Abstract:** An "Amino acid (AA) partial deletion method" was used in this experiment to study the limiting sequences and appropriate ratio of lysine (Lys), methionine (Met), and threonine (Thr) in the diets of 7- to 9-month-old Holstein heifers. The experiment was conducted for three months with 72 Holstein heifers (age = 22 ± 0.5 weeks old; BW = 200 ± 9.0 kg; mean ± standard deviation). Following an initial two weeks adaptation period, heifers were allocated to one of four treatments: a theoretically balanced amino acid diet (positive control [PC]; 1.00% Lys, 0.33% Met, and 0.72% Thr), a 30% Lys deleted diet (partially deleted Lys [PD–Lys]; 0.66% Lys, 0.33% Met, and 0.72% Thr), a 30% Met deleted diet (partially deleted Met, [PD–Met]; 1.00% Lys, 0.22% Met, and 0.72% Thr), and a 30% Thr deleted diet (partially deleted Thr [PD–Thr]; 1.00% Lys, 0.33% Met, and 0.45% Thr). Experimental animals were fed a corn–soybean meal-based concentrate and alfalfa hay. In addition, the animals were provided with supplemental Lys, Met, and Thr (ruminal bypass). The results found no differences in the growth performance and nitrogen retention between PD–Thr treatment and PC treatment (*p* > 0.05). The average daily gain (*p* = 0.0013) and feed conversion efficiency (*p* = 0.0057) of eight- to ninr-month-old heifers were lower in both PD–Lys and PD–Met treatment than those in PC treatment. According to growth performance, Lys was the first limiting AA, followed by Met and Thr. Moreover, nine-month-old Holstein heifers in PD–Lys treatment and PD–Met treatment had higher levels of serum urea nitrogen (*p* = 0.0021), urea nitrogen (*p* = 0.0011) and total excreted N (*p* = 0.0324) than those in PC treatment, which showed that nitrogen retention significantly decreased (*p* = 0.0048) as dietary Lys and Met levels decreased. The limiting sequence based on nitrogen retention was the same as that based on growth performance. The appropriate ratio of Lys, Met, and Thr in the diet based on nitrogen retention was 100:32:57. In summary, the limiting sequence and appropriate amino acid ratio of Lys, Met, and Thr for seven- to nine-month-old Holstein heifers fed a corn–soybean meal-based diet were Lys > Met > Thr and 100:32:57, respectively.

**Keywords:** amino acid pattern; Holstein heifers; lysine; methionine; threonine

#### **1. Introduction**

Nitrogen (N) loss is a major source of environmental pollution and causes significant economic losses for dairy farms. Given the high amount of N excretion that occurs in dairy cattle relative to their N intake, it is likely that these heifers were fed unbalanced amino acids and that their amino acid requirements were ignored [1]. A key factor for improving dietary amino acid (AA) utilization is the formulation of diets with appropriate amino acid patterns that meet but do not exceed the requirements. Many attempts have been made to decrease the environmental effects of cattle N excretion by manipulating the metabolizable amino acid levels of rations to increase the capture of dietary N by cattle [2,3]. However, deleterious effects may occur in cattle not only due to over-doses amino acid but also due to amino acid imbalances, where there is a lack of an appropriate amino acid pattern.

The amino acid partial deletion method is the most common method used to develop balanced AA models in animals [4]. This method can be used to determine the sequences needed to limit AA and calculate the optimal ratios. Dorigam et al. [4] estimated the essential AA profile of poultry and determined the ideal pattern for maintenance using this method. Wang et al. [5] also used the deletion method to determine the AA patterns in calf diets. Lysine (Lys), methionine (Met), and threonine (Thr) were found to be the most limiting amino acids, and their concentrations were related to the growth, physiology, and reproductive performance of calves. Ragland et al. [6] also reported that the limiting amino acids for beef cattle were ranked as Lys > Met > Thr, leading us to the conclusion that Lys, Met, and Thr may be the first three limiting amino acids for dairy heifers.

Research on the amino acid patterns of cattle is limited, particularly regarding the amino acid patterns for each growth stage. Several studies [7] have reported on the amino acid patterns in calf diets [8]. However, one "ideal amino acid pattern" cannot, alone, reliably meet the AA requirements at all growth stages. Balanced AA models should account for changes in growth, body protein composition, and physiological requirements throughout life. The costs of growing heifers are the second largest part in the annual operating expenses of a dairy farm. The lack of optimal amino acid patterns made it difficult to minimize the costs and reduce the nitrogen emissions of heifers. The objective of this study was to determine the amino acid limiting sequence and establish an amino acid ratio in corn–soybean meal and alfalfa hay-based diets for Holstein heifers, aged seven to nine months, using the amino acid partial deletion method.

#### **2. Materials and Methods**

#### *2.1. Animals, Diets, and Experimental Design*

The experimental procedures were approved by the Animal Ethics Committee of the CAAS. Human animal care and handing procedures were followed throughout the experiment (AEC-CAAS-2017-01).

In this experiment, an AA partial deletion method developed by Wang et al. (1989) [9] was used to prepare the different patterns of the Lys, Met, and Thr diets. The AA levels of the total mixed ration (TMR) in the theoretically balanced AA ration were calculated according to the formula proposed by Zinn et al. (1998) [10]: METR = 1.956 + 0.0292 <sup>×</sup> ADG <sup>×</sup> [268 – (29.4 <sup>×</sup> 0.0557 <sup>×</sup> BW0.75 <sup>×</sup> ADG1.097)/ADG] + 0.112 <sup>×</sup> BW0.75 (METR = methionine requirement; ADG = average daily gain; BW = body weight; BW0.75 = metabolic weight). Because of the absence of amino acid patterns in cattle at this stage, Lys and Thr were added according to the AA patterns of the growing swine [11] using a Lys: Met: Thr ratio of 100:30:65. Seventy-two Holstein heifers (age = 5.5 ± 0.5 months old; BW = 200 ± 9.0 kg; mean ± standard deviation) were reared at the Third Dairy Farm of Yinxiang Group Company in Shandong Province, China. The basal diet nutrient level is shown in Table 1.


**Table 1.** Composition and nutrient levels of basal total mixed ration (TMR) (dry matter basis).

<sup>1</sup> The premix provided the following minerals and vitamins for TMR: Cu, 12.5mg/kg; Fe, 90 mg/kg; Zn, 90 mg/kg; Mn, 30 mg/kg; I, 1.0 mg/kg; Se, 0.3 mg/kg; Co, 0.5 mg/kg; vitamin A, 15,000 IU/kg; vitamin D35,000 IU/kg; vitamin E, 50 mg/kg; <sup>2</sup> nutrient levels were measured values, except for metabolizable energy, which was measured and calculated through digestibility and metabolism trials. The energy of CH4 was calculated by equation 10.21 (IPCC, 2006), Ym = 5.5%.

A completely randomized design was used for this study. Heifers were randomly allocated to four treatments with 18 heifers each, based on body weight and age, and fed one of the four total mixed rations (TMRs): (1) theoretically balanced AA TMR (Positive control, PC); (2) 30% Lys deleted TMR (partially deleted Lys; PD–Lys); (3) 30% Met deleted TMR (partially deleted Met; PD–Met); and (4) 30% Thr deleted TMR (partially deleted Thr; PD–Thr). Ruminal bypass Lys (Yahe Nutrition Co., Beijing, China, 36% content, 80% bypass rate), Ruminal bypass Met (Adisseo Co., Hebei, China, 44.4% content, 50% bypass rate), and Ruminal bypass Thr (King Technology Co., Hangzhou, China, 40% content, 90% bypass rate) were added to the basal TMR diet. The AA levels in the four treatments are shown in Table 2. The amounts of AA added were adjusted monthly according to BW and dry matter intake (DMI). After an adaptation period of two weeks, each animal was weighed and began the study with an average initial BW of 226 ± 10 kg and age of 6 ± 0.5 months old.


**Table 2.** Amino acid (AA) levels of TMRs (dry matter basis).

<sup>1</sup> Treatments: PC = theoretical amino acid balance TMR; PD–Lys = 30% Lys deleted TMR; PD–Met = 30% Met deleted TMR; PD–Thr = 30% Thr deleted TMR.

A digestibility and metabolism trial was conducted by selecting four heifers from each treatment during the week before the end of the trial, with a 4-day adaptation period and a 3-day feces and urine collection period. Feces (weight) and urine (volume) outputs were recorded and sampled daily at 07:00, and nitrogen was immediately fixed with 10 mL 10% dilute hydrochloric acid per 100 g feces to determine the N retention (NR). Heifers were housed in individual iron cages (3 × 2.2 m, 6.6 m2/ head) bedded with rice husks and fermented cow dung. Fresh water was added ad libitum and replaced daily. The animals were fed TMR twice daily at 08:00 and 17:00. Amino acids were supplemented into the TMR during morning feeding. Individual intakes of TMR were recorded daily and collected weekly during the entire experiment to calculate the dry matter intake (DMI). Environmental conditions (including air temperature) were continuously recorded. The mean air temperature was 11.87 ± 7.54 ◦C.

The experimental feeding periods were 90 days in duration (September to November 2017). All heifers were immunized according to the standard immunization procedure of the farm, with the brucellosis vaccination administered at 7 months of age.

#### *2.2. Sampling and Analyses*

BW and body size were measured before the morning feeding period every 30 days. Diet samples were collected weekly before the morning feeding and stored at −20 ◦C for further analysis. TMR, feces, and urine samples were sent to the Lab of Ruminant Physiology and Nutrition, Feed Research Institute, Chinese Academy of Agricultural Sciences (Beijing, China) for nutrient analysis. The TMR and feces samples were dried in a forced-air oven at 65 ◦C for 48 h. Then, the DM (105 ◦C for 5 h), crude protein (CP), ash, and ether extract (EE) contents were analyzed (method 968.08; AOAC, 1990) [12]. Calcium (Ca) content was analyzed using an atomic absorption spectrophotometer (M9W–700; Perkin–Elmer Corp., Norwalk, CT, U.S.A.) (method 968.08; AOAC, 1990) [12]. Total phosphorus (P) content was analyzed by the molybdovanadate colorimetric method (method 965.17; AOAC, 1990) [12] using a spectrophotometer (UV–6100; Mapada Instruments Co., Ltd., Shanghai, China). The neutral detergent fiber (NDF) and acid detergent fiber (ADF) contents were determined with an Ankom A200 apparatus (Ankom Technology, Macedon, NY, USA) with heat-stable amylase (Ankom Technology) and sodium sulfite (Fisher Scientific, Waltham, MA, USA) and an expressed inclusive of residual ash [13].

A blood sample was collected from six heifers in each treatment (at 24 and 36 weeks of age) before morning feeding, by a jugular venipuncture, and transferred into vacuum tubes without anticoagulants. Serum was immediately separated from the blood by centrifugation at 3000× g at 4 ◦C for 10 min and stored at −20 ◦C until analysis. Serum urea nitrogen (SUN) was analyzed using blood colorimetric commercial kits (DiaSys Diagnostics Systems GmbH, Frankfurt, Germany).

#### *2.3. Amino Acid Partial Partial Deletion Method*

The principle of the amino acid partial partial deletion method is that there is a linear relationship between the first limiting amino acid and the NR. In other words, the NR will decrease greatest after deleting the first limiting animo acid and will result in the largest slope. The model diagram is as follows (Figure 1):

**Figure 1.** In this model, amino acid intake (AAI) should be presented as the percentage or ratio to control treatment for better distinguishing differences of NR among three AA deleting treatments. To keep the linear relationship between the NR and the first limiting amino acid, the NR should also be converted to the percentage or ratio of the control treatment. The model assumes that deleting the first limiting amino acid (as A) reduces NR to the greatest extent (largest slope); deleting C does not reduce the NR at all (slope = 0), as it remains in excess (over 20%) relative to the first limiting amino acid. Deleting B results in a reduction in the NR intermediate between A and C (0 < Slope B (dashed) < Slope A), and part of B is in excess relative to the first limiting amino acid. In other words, A is the first limiting amino acid while B is second limiting amino acid. According to the principle of the "wooden barrel", all essential AAs can be controlled by the same limitation by adjusting the amount of AAs in the diet. In this model, B is 10% more than A, which means that we should reduce 10% of B from the control treatment to achieve the minimum addition and ensure it is co-limiting with A [9]. Then, we can calculate the ratio of A and B.

#### *2.4. Statistical Analyses*

Data on SUN and N retention were analyzed with a one–way ANOVA procedure using the SAS software (SAS version 9.4; SAS Institute Inc., Cary, NC, USA). Least square means were calculated and separated using the PDIFF option, and differences between diets were detected by Duncan's multiple comparison in SAS. A MIXED procedure was used to analyze the growth performance data. Month, treatment, and treatment by month of age interactions were fixed effects, and the heifers within each treatment were random effects. The effect of the month was included as a repeated measure. For the repeated measures analysis, the covariance structure with the lowest Akaike information criterion was used. The results were reported as the least squares. A significance level was declared at *p* < 0.05.

#### **3. Results**

#### *3.1. Growth Performance*

The results of the growth performance are presented in Table 3. No significant differences (P > 0.05) were observed in the BW and DMI of heifers among the four treatments during the experiment. However, the ADG (*p* = 0.0013) and G/F (*p* = 0.0057) of heifers in the PD–Lys and PD–Met treatment were decreased significantly compared to PC treatment at eight to nine months old.


**Table 3.** Effects of deleting Lysine (Lys), Methionine (Met), and Threonine (Thr) levels in corn–soybean based TMR on the growth performance of heifers aged seven to nine months old (n = 72).

<sup>1</sup> BW=body weight, ADG=average daily gain, DMI=dry matter intake; <sup>2</sup> Treatments: PC = theoretical amino acid balance TMR; PD–Lys = 30% Lys deleted TMR; PD–Met = 30% Met deleted TMR; PD–Thr h = 30% Thr deleted TMR; <sup>3</sup> <sup>T</sup> <sup>=</sup> Treatment, M <sup>=</sup> Month of age, T × <sup>M</sup> <sup>=</sup> The interaction between treatment and month of age; a,b,c values within the same row with different superscripts are different (*p* < 0.05).

#### *3.2. Serum Urea Nitrogen Levels*

The SUN levels of heifers aged eight months old (*p* = 0.0013) and nine months old (*p* = 0.0021) in the PD–Lys and PD–Met treatments were higher than those in the PC treatment (Figure 2). No significant differences of SUN were observed between the PD–Thr treatment and PC treatments (*p* > 0.05).

**Figure 2.** Comparison of serum urea nitrogen levels of seven- to nine-month-old heifers fed corn–soybean based TMRs among the four treatments (n = 24); PD–Lys = 30% Lys deleted TMR diet (diagonal stripes bar), PD–Met = 30% Met deleted TMR diet (vertical stripes bar), PD–Thr = 30% Thr deleted TMR diet (horizonal stripes bar), PC = theoretically balanced amino acid TMR diet (gray bar); The y-axis represents the serum urea nitrogen levels of four treatments; the x-axis was the age of heifers. Error bars indicate SEM. The a,b above the bars indicate the significant differences among treatments (*p* < 0.05).

#### *3.3. Nitrogen Metabolism*

There was no difference in N intake among treatments (Table 4). Total excreted N significantly increased (*p* = 0.0208) when dietary Lys and Met were reduced, as there were significant increases in urine N (*p* = 0.0011). However, fecal N and Digestible N did not differ among four treatments (*p* > 0.05). Moreover, the amount of urine N and NR of heifers in the PD–Thr treatment were not significantly different from those in the PC treatment (*p* > 0.05).

**Table 4.** Effects of deleting Lysine, Methionine, and Threonine levels in corn–soybean based TMRs on nitrogen metabolism of heifers aged seven to nine months old (n = 16).


<sup>1</sup> N = nitrogen; Total excrete N=Fecal N + Urine N, Absorbed N =Intake – Total excrete N, NR (N retention) = N intake – fecal N – urinary N, N utilization = (N intake – fecal N excretion)/N intake × 100%, N digestibility = (N intake – fecal N excretion)/N intake × 100%; <sup>2</sup> treatments: PC <sup>=</sup> theoretical amino acid balance TMR; PD–Lys <sup>=</sup> 30% Lys deleted TMR; PD–Met = 30% Met deleted TMR; PD–Thr = 30% Thr deleted TMR; a, b, c values within the same row with different superscripts differ (*p* < 0.05).

#### *3.4. Appropriate Amino Acid Model*

#### 3.4.1. N Retention and Amino Acid Intake

N retention (NR) and amino acid intake (AAI) based on metabolic weight (Table 5) were converted in proportion to the PC treatment based on the requirements of the "Amino acid partial deletion method model". Then, the proportions of the intakes of Lys, Met, and Thr in the PD–Lys, PD–Met, and PD–Thr treatment to those in the PC treatment were calculated (e.g., the AAI of Lys in the PD–Lys treatment is 0.60, the AAI of Lys in the PC treatment is 0.90, and the ratio of Lys in the PD–Lys treatment to Lys in the PC treatment is 0.60/0.90 = 0.67 ). After conversion, the ratio of Lys, Met, Thr in amino acid to PC treatment were 0.67, 0.69, and 0.62, respectively. This result differs slightly from 0.7 due to the deferences of the metabolic body weights of the heifers in the four treatments.


**Table 5.** The proportions of amino acid intake and nitrogen retention in PD–Lys, PD–Met, and PD–Thr to those in the PC treatment.

<sup>1</sup> PD–Lys = 30% Lys deleted treatment; PD–Met = 30% Met deleted treatment; PD–Thr = 30% Thr deleted treatment; PC = theoretical amino acid balanced treatment; <sup>2</sup> NR=N retention; AAI=amino acid intake.

3.4.2. Calculation and Model Diagram of the Effect on Nitrogen Retention

The proportions of the three essential AAs were calculated using the simple linear model based on the amino acid partial deletion method [9]. The model diagram of the effect on NR after deleting 30% of Lys, 30% of Met, and 30% of Thr in corn–soybean based diets is shown in Figure 2.

Figure 3a shows the rate of NR in relation to the daily AA intake. The values of AAI (x-axis) and NR (y-axis) are provided in Table 6. Point "PC" represents the corresponding AAIs and NRs of the three AAs in the PC treatment (all values = 1 only one point). "Lys" is the point of the Lys intake and NR in the PD–Lys treatment (0.67, 0.80). "Met" is the point of the Met intake and NR in the PD–Met treatment (0.69, 0.83). "Thr" is the point of Thr intake and NR in the PD–Thr treatment (0.62, 0.85).

**Figure 3.** The pattern diagram (**b**) when Met and Thr are converted to an equivalent slope (**a**) with Lys. The y-axis represents the ratio of NR after deleting Lys, Met, Thr to that of the PC treatment; the x-axis is the ratio of the amino acid intake (AAI) in the amino acid deleting treatments to that in the PC treatment. (Lys, -) = Lys intake and NR level in PD–Lys treatment, (Met, ) = Met intake and NR level in PD–Met treatment, (Thr, ) = Met intake and NR level in PD–Thr treatment, and (PC, -) = Lys, Met, Thr intake and NR level in PC treatment; all values =1.

**Table 6.** The appropiate amino acid ratio of Lysine (Lys), Methionine (Met), and Threonine (Thr) based on the nitrogen retention (NR) of heifers aged seven to nine months old, fed corn–soybean meal-based TMRs.


<sup>1</sup> S(Slop) <sup>=</sup> (1-NR)/(1-AAI); P(proportion) <sup>=</sup> [(1-NR) <sup>+</sup> <sup>S</sup> × AAI]/S; C (concentration) <sup>=</sup> AAI × P; R (ratio) <sup>=</sup> AA/Lys.

The slope (Table 5) describes the effect of deleting an AA from the PC on the NR (e.g., for Lys, (1–0.80)/(1–0.67) = 0.61). Among the three AAs, a higher slope for the Lys deletion treatment (PD–Lys) means that Lys is the first limiting AA in the PC treatment. The limiting sequence of the three amino acids is ranked as: Lys > Met > Thr.

Figure 3b is the pattern diagram for when Met and Thr are converted to an equivalent slope with Lys. To calculate the proportion of each AA that could be removed from the PC amino acid pattern to make it equally limiting to Lys, it was assumed that when all AAs are equally limiting, they should all have the same slope. Therefore, the required amount of Met can be caculated as: S (Lys) = (1 − 083)/(x − 0.69)—that is, 0.61 = (1 − 0.83)/(x − 0.69), x = 0.98, thence 1.00 − 0.98 = 0.02. In other words, 0.02 of Met should be removed from the PC to make the Met co-limiting with the Lys (the actual requirement of Met is 0.98 × Met in the PC treatment). In the same way, we calculate that the 0.21 of Thr should be removed from the PC treatment when it is equally limited with Lys (the actual requirement of Thr is 0.79 × Thr in the PC treatment).

#### 3.4.3. Appropriate Amino Acid Ratio

An appropriate amino acid model of seven- to nine-month-old Holstein heifers is shown in Table 6 (calculated from Figure 2). S (slope) represents the effect of deleting 30% amino acid on N retention (S = (1 − NR)/(1 − AAI), calculated from Figure 2a). The S value of the Lys deleted treatment was the highest, indicating that Lys was the first limiting amino acid. P (proportion) is the proportion of amino acid (except for Lys) in the PC treatment when it was equally limited to that of Lys (P = [(1−NR) + S × AAI]/S, calculated from Figure 2b). The P value was calculated based on the principle of "equal limitation means equal slope." C (concentration) is the actual concentration of amino acid when it was equally limited with Lys (C = AAI ( in PC treatment) × P). R (ratio) is the ratio of the actual amino acid concentration to the Lys concentration (R = AA/Lys). The optimal ratios based on the NR of the three amino acids for seven- to nine-month-old Holstein heifers was 100:32:57.

#### **4. Discussion**

#### *4.1. Growth Performance and Body Size*

The function of dietary protein is determined by amino acid composition, the nutrient digestive abilities of animals, and how well the composition of absorbed amino acid matches the balance required by the animals. The deficiency and overdose of certain amino acids in the diet will cause an imbalance between amino acids and thus affect the growth and development of the animals. For calves, the addition of Lys and Met in a milk replacer significantly increased the feed conversion efficiency of calves, but the addition of Thr had no significant effect on the growth performance of calves [7]. Ludden et al. [14] observed that supplementation with Lys improved the ADG in the growing cattle. Awawdeh et al. [15] showed that when Met was limiting amino acid, the dietary supplementation of other amino acids increased the utilization efficiency of Met and increase the growth of bulls. In this experiment, deleting 30% of Lys and Met led to a decrease of ADG and G/F. Such a reduction of growth performance might be due to the unbalanced amino acids. Another important observation is that DMI seems not to be affected by treatment. Wang et al. [16] found no significant differences in the DM and N intake of dairy cows after adding Lys and Met to the diet. That is to say, growth performance is affected by limiting amino acid deficiencies rather than feed intake [17]. Lys and Met might be the first two limiting amino acids for growing cattle. Unlike Lys and Met, the growth responses to Thr deletion were not significantly decreased. It remains possible that the theoretical Thr addition in this experiment was relatively higher than the requirement of heifers due to the absence of accurate data on Thr requirements in this trial.

#### *4.2. Serum Urea Nitrogen and Nitrogen Retention*

SUN, as an end metabolite of the liver's N metabolism [18], is negatively correlated with the utilization rate of protein [19]. The balance of amino acids is the basic condition needed to improve protein utilization and reduce SUN concentration [19]. Jiang et al. [20] estimated sharp decreases in the content of SUN and the emission of urine nitrogen, as well as an increase of the N retention of cows after adding Met and Lys in their diets. In our study, SUN concentration increased after deleting 30% of dietary Lys and Met, which might indicate an imbalance of amino acids and decrease N utilization. Urine N is the main excretion pathway of SUN, accounting for a large part of the N excretion of heifers. An amino acid balanced diet can improve the N utilization rate and reduce the excretion of fecal and urine nitrogen (about 46%) of dairy cows, especially urine N excretion [21]. Adding Lys and Met to diets can promote a balance of amino acids, reduce urinary nitrogen concentration, and improve the protein utilization rate of dairy cows [22]. Recent research by Lee et al. [2] concluded that the efficiency of feed N absorbed by the small intestine increased when dietary amino acid was balanced. We also

observed that urine N was significantly increased as the Lys and Met levels decreased. Therefore, it was further confirmed in this study that deleting Lys and Met led to an imbalance of amino acids, which resulted in an increase of urine nitrogen.

N retention reflects the efficiency of protein deposition and amino acid utilization [23], which is also closely related to the production performance of animals [24]. A balance of amino acids in the diet can enhance the digestion and absorption of N in animals [25]. In particular, the metabolic amount of the first limiting amino acid has a linear relationship with N retention [26]. Balancing a complete amino acid profile increased the efficiency of dietary N utilization in both a low and a high small intestine protein supply [27]. The efficiencies of nitrogen estimated in the current study confirmed that adding Lys to the Lys deficient diet of calves reduced the rate of N excretion and increased the rate of N deposition [28]. Conversely, heifers fed with Lys and Met deficient diets caused an increase of nitrogen retention, indicating inefficiencies in their use of absorbed amino acid for protein accretion [25]. Importantly, the effects of dietary Lys, Met, and Thr levels on the N retention of heifers is not consistent and largely depends on the balance and limiting sequences of these three amino acids. In this study, the decrease of N retention, in combination with the deficiency of Lys and Met, indicated that Lys and Met are the first and second limiting amino acids for heifers , respectively.

In addition, an increase in N retention was commonly reported for cows fed diets with a supplementation of rumen-protected amino acids [29,30], similar to the present study, which indicated that the added ruminal protected amino acids were effectively protected from ruminal degradation and guaranteed amino acids to be released and absorbed in the small intestine for better utilization. Dietary supplementation of rumen-protected Met and Lys could ensure a balance of amino acids, promote the increase of nitrogen deposition, and improve the utilization rate of proteins [31]. In this case, supplementation with rumen-protected amino acids may be a successful strategy for establishing the amino acid pattern of heifers based on dietary amino acids.

#### *4.3. Limiting Sequence and Appropriate Ratio of Amino Acid*

The limiting sequence of amino acids in ruminants was affected by the composition of their diets. Maize silage/maize gain based diets can supply adequate protein but do not provide enough Lys to growing cattle, which indicates that Lys is the first limiting amino acid [32]. Klemesrud et al. [33] also reported that Lys is the first limiting amino acid in steers fed with diets containing large amount of maize products. Wang [8] found that Lys was the first limiting amino acid (the second and third were Met and Thr, respectively) due to the large decrease of ADG in calves fed with a milk replacer, starters, and Leymus chinensis after reducing Lys. We found that the limiting sequence of seven- to nine-month-old Holstein heifers was Lys > Met > Thr, based on corn–soybean meal–alfalfa TMRs. Therefore, Lys plays the most important role in the growth of heifers that are fed rations made from corn–soybean meal.

An optimal amino acid pattern is needed as a standard for evaluating the diets of animals. The requirements of amino acids are not well defined for heifers with corn–soybean meal-based diets, and modifying amino acid patterns can increase the bypass protein efficiency [34]. When the Lys ratio is expressed, variation in the estimated requirement of the specific AA is greatly reduced compared to the amino acid ratio of the diets [25]. NRC (2012) [35] pointed out that the ideal amino acid pattern should be expressed as the ratio of amino acid to Lys. In this study, Lys happened to be the first limiting amino acid, so the calculated model is appropriate. The results from this study show that the appropriate pattern of amino acids in the diet based on maximum N retention in seven- to nine-month-old Holstein heifers (fed with corn–soybean meal) was 100:32:57. However, we did not precisely determine the limiting amino acid pattern because of our inability to accurately calculate metabolic proteins, so this pattern cannot be applied to all type of diets. This may offer an explanation for the differences between this pattern and the patterns for calves and cows (Table 7). Of course, the amino acid requirement for growing heifers might change according to age. It is possible that differences in diet type and digestion among heifers, calves, and cows directly affect the profile of delivered amino acids to the intestine. However, the ruminal bypass amino acid products and microbial metabolism made it difficult to determine whether there was a large difference between intake N and metabolic N. Therefore, further research is needed to determine the precise amino acid patterns based on metabolic protein.

Amino acid is mainly used for the bodily growth and development of heifers. Amino acid requirements are mainly determined by body protein retention and N emission, similar to beef cattle, so the amino acid requirement of heifers may be determined by the composition of the body's amino acid [36]. However, the amino acid pattern in this research was different from that of beef cattle (Table 7). Studies have shown dietary amino acid patterns are not equivalent to carcass amino acid composition. Decomposition and conversion by intestinal bacteria produced a significant difference between amino acid in the diet and amino acid absorbed into the blood. Moreover, different tissues have different uses and metabolic efficiencies for amino acid, which may cause a deviation between a carcass's amino acid composition and dietary amino acid patterns. Considering animal welfare and economic benefits, the calculated amino acid model was not verified by a carcass's amino acid components. Whether a carcass's amino acid components can be used as an appropriate amino acid model for growing heifers needs to be further verified.



<sup>1</sup> NR = nitrogen retention; ADG = average daily gain.

#### **5. Conclusions**

In this study, there were negative effects on the average daily gain, feed conversion rate, and nitrogen retention of seven- to nine-month-old heifers after deleting 30% dietary Lys and Met. However, deleting Thr content did not affect the growth performance and N metabolism of heifers. The sequence of the three amino acids for seven- to nine-month-old Holstein heifers that were fed a TMR of corn–soybean meal concentrate and alfalfa hay was Lys > Met > Thr. Additionally, the appropriate amino acid ratio calculated from nitrogen retention of this ratio of diet was 100:32:57.

**Author Contributions:** Conceptualization, Y.T. and Q.D.; methodology, Y.T and Y.L.; software, Y.L.; validation, Y.T; formal analysis, Y.L.; investigation, F.H., M.T., Y.S.; resources, Q.D.; data curation, F.K.; writing—original draft preparation, Y.L.; writing—review and editing, M.P. and B.W.; visualization, Y.T.; supervision, Y.B.; project administration, Y.B.; funding acquisition, Y.T. and Q.D. All authors participated in writing the final draft of the manuscript and agreed on the final format.

**Funding:** This work was supported by the Beijing Innovation Team for Technology Systems in the Dairy Industry (BAIC06), the Chinese Academy of Agricultural Science and Technology Innovation Project (CAAS-XTCX2016011-01), Fundamental Research Funds for Central Non-profit Scientific Institution of CAAS (Y2019CG08), and Science and Technology Open Cooperation Project of Henan Province (182106000035)—Study on the Amino Acid Pattern of Diet for Different Physiological Stages of Heifers.

**Conflicts of Interest:** We certify that there are no conflicts of interest with any financial organizations regarding the material discussed in this manuscript.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

**Dairy Heifers**

## *Article* **An Economic Analysis of the Costs Associated with Pre-Weaning Management Strategies for**

## **Anna Hawkins 1, Kenneth Burdine 2, Donna Amaral-Phillips <sup>1</sup> and Joao H.C. Costa 1,\***


#### Received: 1 June 2019; Accepted: 17 July 2019; Published: 23 July 2019

**Simple Summary:** Rearing of replacement female calves on a dairy farm is of critical importance to maintain herd sizes, improve the genetic quality of the herd, and remain economically sustainable. A 2-year investment period is needed for replacement female heifers to grow before entering the milking herd. The management of replacements over this 2-year period can vary greatly among operations, making it difficult to compare producers' cost to benchmark. The objective of this project was to develop a model to calculate the cost of rearing a replacement heifer from birth to weaning under different housing, milk source, allotments, and labor and health management decisions to be used as a dairy farm decision support tool. We calculated the cost for management options with general cost values. We found that the average feed cost represented 46% of the total cost while labor, and fixed and variable costs represented 33%, 9%, and 12%, respectively. The total cost increased as milk allotment increased, but cost per Kg of gain decreased. The ranges in total cost within each management scenario often exceed the difference in cost from one scenario to the next. In conclusion, variable costs have the potential to vary among operations, playing a major role in the total cost of rearing replacements from birth to weaning.

**Abstract:** Dairy calves are raised in various housing and feeding environments on dairy farms around North America. The objective of this study was to develop a simulation model to calculate the cost of raising replacement dairy heifers using different inputs that reflect different management decisions and evaluate their influence on the total cost. In this simulation, 84 calves were modeled between 0–2 months of age to reflect a 1000 heifer herd. The decisions associated with housing, liquid diet source and allowance, labor utilization, and health were calculated. Costs and biological responses were reflective of published surveys, literature, and market conditions. A 10,000-iteration economic simulation was used for each management scenario using @Risk and PrecisionTree add-ons (Palisade Corporation, Ithaca, NY, USA) to account for variation in pre-weaning mortality rate, weaning age, and disease prevalence. As milk allotment increased, total feed cost increased. Feeding calves a higher allowance of milk resulted in a lower cost per kg of gain. Average feed cost percentage of the total cost was 46% (min, max: 33%, 59%) while labor, and fixed and variable cost represented 33% (20%, 45%), 9% (2%, 12%), and 12% (10%, 14%), respectively. Total pre-weaning costs ranged from \$258.56 to \$582.98 per calf across all management scenarios and milk allotments.

**Keywords:** calf economics; replacement; ADG; cost per kg

#### **1. Introduction**

Heifer availability is critical for the dairy operation to maintain a consistent herd size and remain economically sustainable in most cases [1]. Improved fertility and increased use of sexed semen have made replacement heifers more available for dairy operations [2]. Some producers keep all newborn replacement heifers in case more replacements are needed than anticipated, which can create a heavy financial burden for producers when raising excess heifers. Heifer raising expenses are often lumped into broad farm-wide expenses such as feed, labor, and health costs, making it difficult to accurately calculate heifer raising costs [3]. In addition, failing to identify the on-farm cost to raise a replacement heifer can allow for inefficiencies in feed, labor, housing, or health costs to go unnoticed, which accumulate unanticipated replacement female costs.

Previously reported replacement heifer rearing costs are variable and can be explained in part by differences in rearing management systems. For example, the average total cost to raise replacement heifers to wean was found to only increase by \$82.88 per heifer from 2000 to 2015 but ranges within each study can exceed \$350 per heifer [4–6]. Heinrichs [6] found a range in feed cost on 44 farms of \$29.06 to \$259.17 per calf; total cost per calf ranged \$89.00–\$442.78 during the pre-weaning period. In a 2014 survey of 2545 heifer calves in the United States, individual housing was the dominant form of housing pre-weaned heifers at 86.6% and 13.4% were managed in group housing, yet 8 different housing types were reported [7]. Little research has examined the cost between housing types, although the University of Wisconsin has conducted surveys of producers in an automatic and conventional housing scenario. The average cost (min, max) of producers utilizing individual housing was \$363.69 (\$195.06, \$530.76) and those with group housing was \$401.58 (\$138.39, \$585.52), a difference in average cost of \$37.89 per calf, but with a difference range of over \$300 for individual and \$400 for group housing [8].

Housing is the first of many decisions a producer makes on how pre-weaned calves will be managed. Utilization of labor and milk source requires additional decisions based on resources and availability. While gaining in popularity, only 1.9% of the calves were fed through an automatic feeder while almost half of the surveyed calves were fed using a bottle or a bucket [9]. On average, one calf requires 7–12 labor hours during the pre-weaning period, or 7–10 mins per day [8]. Unpasteurized whole milk was the most common milk source utilized by producers but close to 50% of those also utilized milk replacer [7]. More recent surveys show a similar trend, with 40.1% of calves fed whole or waste milk, 34.8% fed milk replacer, and 25.1% fed a combination of the two. Calf starter was provided, starting on average at 5 days old, to all calves surveyed [9].

Thus, it is important to understand the costs associated with the myriad of rearing systems for dairy calves in the United States. The objective of this paper was to evaluate the economic impact of different calf raising management decisions, especially housing, liquid diet and allowance, and health expenses on the total pre-weaning cost of rearing heifer replacements.

#### **2. Materials and Methods**

A cost simulation model was developed at the University of Kentucky Dairy Science program during 2018. This economic model was developed in Excel 2013 (Microsoft, Redmond, WA, USA) utilizing @RISK and PrecisionTree add-ons (Palisade Corporation, Ithaca, NY, USA). The base herd used included 1500 milking cows, 1000 replacement heifers in total and 84 heifer calves in the pre-weaning period, assuming a 30% replacement rate and an average age at first calving of 25 months. Costs were calculated on a per head basis for housing, feed, labor, mortality, and health. All remaining variables are static. Interest was accounted for on infrastructure and mortality as well as the depreciation of assets related to replacement females. Costs associated with herd-wide parameters, such as disease prevalence and mortality rates, were distributed across all remaining calves in the pre-weaning phase. The model required a management decision at 3 points: housing type, milk source, and labor shown in Figure 1. Three main housing types were modeled: individual housing outside (IHO), individual housing inside (IHI) [10], and group housing (GH). Three milk sources were built into the model: whole milk (WM), pasteurized whole milk (PWM), or milk replacer (MR). Four possible liquid feeding plans were modeled: 6, 8, 10, and 12 L of milk per calf per day. Labor was modeled for two categories: conventional, where a person was assigned to feeding and caring for the calves; or automatic, where

an automatic calf feeder was utilized in addition to human labor. Totals costs were reported per calf for each management decision, the entire pre-weaning period per calf, and per day per calf. Per day cost was calculated by dividing days of age at weaning by the total cost per calf during the pre-weaning period.

**Figure 1.** Decision tree of possible management decisions for housing, milk source, and labor for pre-weaned calves.

#### *2.1. Housing*

Housing systems that required a barn (IHI and GH) used values found from Table 1 to determine barn value and monthly payment. Barn cost was derived from the Dairy Calf and Heifer Association Gold Standard recommendation of 3.3 M<sup>2</sup> per calf, with an additional 15% of space to account for walkways and feed areas. Thus, replacement heifers were assumed to require 3.7 M<sup>2</sup> per calf. Construction cost [11] varied based on the infrastructure required for each situation, ranging from \$10.00 to \$15.50 per M2. Estimated barn value (BV) was then calculated with Equation (1).

$$\text{BV} = \text{CC} \times 3.7 \,\text{M}^2 \times \text{number of pre-weared calves} \tag{1}$$


**Table 1.** Model inputs were adapted from the published literature, the latest USDA reports, and heifer raising surveys.

Barn cost per heifer (BC) was calculated using the payment function in excel with 7% interest, 20 years useful life and BV. BC was included in IHI and GH situations. Calves housed in individual housing outside followed the same payment function. Housing calves year-round in individual housing with an average occupancy time of 2 months ± rest period would allow 5 calves per hutch per year. Days of age at weaning was used as the length of time a heifer was incurring cost during the pre-weaning period. Housing costs also included utility costs, such as water and electricity. Electricity was only factored for housing systems that included a barn (IHI and GH). The bedding was included at a flat evaluation of \$11.00 per calf. For pasture scenarios, a cash value price per acre was used as the value of land to try to account for the opportunity cost of a specific acre being used for other purposes. An additional annual maintenance cost of \$31.50 per acre was assumed.

#### *2.2. Feed*

Milk replacer was mixed at a concentration of 0.11 kg per liter of water. A pasteurizer was depreciated over all calves over the 15-year useful life. The model accounted for four possible feeding milk allotments: 6, 8, 10, and 12 L per calf per day. A 2016 survey of producers in the United States showed over half of the farms were feeding calves between 4–6 L per day [7]. Recent studies have shown that increasing milk allotment can increase average daily gain (ADG) pre-weaning, result in larger skeletal measurements at weaning, and decrease vocalizations caused by milk deprivation [12,13]. Milk allotments and starter intakes per calf for this model were reflective of experimental data [14]. In this study, calves were randomly assigned to 6, 8, 10, or 12 L feeding treatments of pasteurized whole milk with *ad libitum* access to calf starter. A step-down weaning program was performed: milk was fed at maximum allotment until weaning at 42 days. Milk allotment was decreased by 50% until day 50, where allotment was decreased daily by 20% until weaned. Calves were assumed to be consuming at least 2.25 kg of calf starter at weaning. An additional 20% was assumed to be fed to account for waste and loss. This additional expense was added to daily calf starter cost. Milk and calf starter costs were calculated on a daily basis for the entire pre-weaning period, from day 0 to 65. ADG was determined using the dry matter intake requirements and resulting gain from NRC, 2001. Calf weight was modeled daily to determine appropriate weaning weights based on dry matter intake from milk replacer or whole milk.

The assumed birth weight was 40 kg for each calf. Assumed average daily gains on each feeding plan (6, 8, 10, and 12 L) are described in Table 2, following the equation presented in NRC, 2001. The weaning weight was calculated by multiplying ADG by 65 days and adding the weight gain to BW. Feed cost was reported for three variables: total cost during the pre-weaning period, daily feed cost, and feed cost per kg of gain. Total feed cost included milk replacer or whole milk expenses and feeding equipment. Daily feed cost was derived from dividing total feed cost by weaning age (65 days). Daily feed cost was then used to calculate feed cost per kg of gain. Daily cost under each feeding plan was divided by ADG to determine the cost per kg of gain.

**Table 2.** Birth weight and weaning weight were a result of milk allotted and calf starter intake per calf following experimental data from Rosenberger et al. 2017. Average daily gain (ADG) followed the equation presented in NRC 2001.


Birth weight (BW) was assumed at 40 kg, weaning weight (WW) was calculated based on ADG for 65 day weaning age.

#### *2.3. Labor*

Labor to care for calves and the number of employees working were adapted from published surveys of producers employing individual and group housing (Table 1). Because of the lack of data on group housing without automatic feeder labor time requirements, we assumed the median of an automatic feeder and individually housed heifers (5.5 mins/calf/day). Management labor was calculated separately to represent additional labor required from owners, managers, and/or family. Management followed the trend of 10% of the paid labor, creating the assumption of 0.55 mins/calf/day for group housing without an automatic feeder. Minutes per calf could be input directly or total time per all pre-weaned calves could be used to calculate total labor cost using Equation (2).

$$\begin{aligned} \text{((Daily Paid Labor Hours/Number of Calves)} \times \text{Hourly Paid Labor}) + \text{((Daily Calves)} \times \text{(Nuclear Hours/Number of Calves)} \times \text{(Nourly Monagnement Pay)} \end{aligned} \tag{2}$$

The expenses related to buying and using an automatic calf feeder were included in the labor section. Justified by the change in labor demands, the use of an automatic calf feeder can be viewed as an additional autonomous employee. The cost of the feeder was assumed at \$15,000 value, 10 years useful life and \$200.00 annual maintenance. These values were assumed based on market prices and a routine maintenance program. Equation (3) represents the calculation of daily feeder cost per calf using the payment (PMT) function in excel.

$$(-\text{PMT (interest rate, 120, initial value)}) \text{(number of pre-we2ned values} \tag{3}$$

#### *2.4. Mortality and Health*

The cost of each calf was calculated daily and, therefore, monthly cost to raise one calf in each management style was determined. All calf mortality events were assumed to occur at the end of the first month of life, accruing the additional monthly cost plus interest. This additional cost is divided over the remaining number of calves. Equation (4) explains how calf mortality was added as an additional cost to each remaining calf.

$$\begin{aligned} \text{(Value of Newton Call} + \text{(Cost up to death} \times (\text{(Interest Rate/365}) \times 60))}\\ \text{Rate}) \text{(Remaining Calves} \end{aligned} \tag{4}$$

Health costs are reflective of a standard vaccination protocol including fly control, respiratory vaccine, vitamin A, D, and E, selenium, and a vaccine for rotavirus and coronavirus scours, and *E. Coli*. Labor costs related to health tasks were compiled into a "working heifer" labor expense. The total health cost was figured at \$9.22 per calf. Fair market prices were assumed on all vaccine and health-related equipment through averaging online prices obtained in January 2019.

The prevalence of respiratory illness and diarrhea was determined by the 2014 Heifer Raiser Survey conducted by the USDA, 18% for respiratory illness and 25% for diarrhea on average. Because of the variation in this measure from farm to farm it was made stochastic to account for variation between farms. The minimum incidence was 16% for respiratory illness with a maximum of 19%; the minimum of diarrhea was 22% with a maximum of 28%. Based on the selected prevalence, there was a direct relationship to the additional treatment cost for each calf. We modeled a protocol that would include electrolytes and 3 days of antibiotics. We assumed an 85.6% improvement rate and culled the remaining heifers at the end of that week.

#### *2.5. Statistical Simulation*

A simulation model was developed in Excel 2013 (Microsoft, Redmond, WA, USA) utilizing @RISK and PrecisionTree add-ons (Palisade Corporation, Ithaca, NY, USA) to evaluate the cost of raising an individual heifer from birth to weaning under different management styles and systems. 10,000 simulations of the model were performed for each of the situations. Stochastic simulations allowed for variation of inputs values which are reflected in ranges of potential outcomes, unlike a static model which will always produce the same outcome. Modeling variables stochastically, such as weaning age, mortality rates, and disease prevalence, we can simulate different outcomes. All variables were modeled following a Pert distribution set with minimum, most likely, and maximum value. Assumptions were made based on published literature, surveys, and market assumptions were also used to calculate the total cost (Table 1). A month in the cost spreadsheet was considered 30 days. This model is available online at https://afs.ca.uky.edu/dairy/decision-tools, wherein all variables and assumptions can be modified to reflect different situations and individual farms.

#### **3. Results and Discussion**

#### *3.1. Housing*

The total cost to house calves in individual housing outside, individual housing inside and group housing were \$21.12, \$70.52, \$94.30, respectively. All of these costs were within 1 SD of the average found in published literature. For housing that included a barn, the barn payment per heifer was the largest contributor to cost, while bedding was the largest contributing cost per calf for individual housing outside (Table 3).

**Table 3.** Percentage breakdown of hutch/barn infrastructure, bedding and, water and electric on total housing cost per housing management decision.


\* includes interest and depreciation of infrastructure.

#### *3.2. Feed*

Feed cost was heavily dependent upon the amount of milk allotted per day. Table 4 shows the total cost of each milk source with 6, 8, 10, and 12 L allotments. As milk allotment per calf increased, the cost of milk increased.


**Table 4.** Cost of milk replacer, whole milk, and pasteurized whole milk as a milk source for calves with 6, 8, 10, and 12 L milk allowances.

The cost of pasteurizing whole milk ranged from 10 to 18% of the total cost of feeding calves in applicable scenarios. This model assumed the same nutritional value and gain from milk replacer and whole milk, creating a limitation in the model. However, calves fed pasteurized or unpasteurized whole milk have been shown to increase model-produced ADG by at least 0.03 kg/day with the potential to be over 0.25 kg/day of gain in comparison to milk replacer [15]. The additional cost to feed whole milk has the potential to be offset by an increase in weight gain.

The estimated cost per kg of gain decreased as milk allowance increased, and with increasing ADG, shown in Table 5. For example, group-housed calves on milk replacer with an automatic feeder fed 6 L will cost \$3.50 per kg of gain. When these same calves are increased to 12 L the cost decreases to \$2.67 per kg of gain. The minimum decrease in cost was from feeding 10 L of milk replacer to 12 L of milk replacer at \$0.01 difference per kg of gain, and the maximum savings per kg of gain was \$0.41 increasing from 10 to 12 L of pasteurized whole milk. If birth weights were 44 kg with a goal of weaning calves at 100 kg, we could assume a minimum of \$0.56 to \$22.96 in feed efficiency savings per calf alone. Modeling cost per kg of gain following experimental data presented in the (NRC, 2001) equations indicates that feeding calves a higher allowance of milk decreases the cost per kg of gain. The cost of milk and calf starter, with our current assumptions in inputs and ADG, decrease cost per kg of gain.


**Table 5.** Feed cost per kg of gain of pre-weaned calves fed milk replacer, pasteurized whole milk and whole milk.

#### *3.3. Labor*

The labor decisions depended on the housing system selected. Hourly wages for management are higher than those for paid employees as shown in Table 1. Employees contributed more to the total cost than management in conventional and automatic systems even though their hourly rate is lower. Labor costs associated with the automatic calf feeder were responsible for 23% of the total labor cost. Labor cost of individual housing and group housing contributed 33% and 26%, respectively. The minutes and total cost per hourly laborer were decreased from inside individual housing to group housing by 36% per calf for a value of 2.4 minutes or \$0.50 per calf per day. This shows a reduction in overall labor cost but an increased demand for fixed and variable expenses. These include the paying for the feeder, annual maintenance and a barn to house calves.

This breakdown of cost follows the same trends of Wisconsin surveys of conventional and automatic calf raisers. The paid labor cost alone was reduced by 39% for farms utilizing an automatic calf feeder, and paid management decreased by 14%. The total pre-weaning cost decreased 6% from conventional to automatic labor; the cost difference was recovered in an additional fixed variable cost of the automatic calf feeders.

#### *3.4. Health*

Mortality rate and prevalence of diarrhea or respiratory illness, which were included in variable costs, impacted the total cost. The average cost, including the risk of each calf being healthy or experiencing diarrhea, totaled (mean ± SD) \$5.39 ± 14.42 per calf. The average cost per BRD case was \$0.70 ± 7.33 per calf. Preventative health costs added an additional \$9.22 to each calf.

The change in total cost per calf, accounting for additional expenses with fewer calves as mortality rate increases (2%, 8%, 10%, and 15%), are reported in Table 6. As mortality rate increased, the cost of infrastructure and higher cost management systems showed a larger increase in the dollar amount added for each calf. Across management styles, decreasing the mortality rate from 15% to 2% reduced overall cost from \$39.47 to \$36.84 per calf. For a farm raising 500 pre-weaned calves annually, potential savings by decreasing mortality 10% alone could be over \$18,000.

**Table 6.** Total cost under each management pathway per calf when mortality rate is set at 2, 8, 10, and 15%.


It has been found that management practices specific to a housing type can change illness prevalence. For example, calves housed in groups of 12–18 had a higher incidence of respiratory illness and lower daily gains than calves housed in groups of 6–9 [9]. We assume a constant square footage per calf, therefore the barn square footage increases as the number of calves increase and this may not always be reflective of true management practices. A limitation to the model is the same probabilities in averages and ranges in mortality for all management pathways for calculated cost.

#### *3.5. Total Cost of Management Scenarios*

All possible combinations of management decisions (each combination of housing type, milk source, and labor type) and for each of the 4 milk allotments were analyzed for total cost, daily cost, and percentage of feed, labor, and fixed and variable costs (Table 7). Fixed costs included barn and housing infrastructure, depreciation of assets, and interest. Variable costs included health-related expenses, mortality, and utilities for electricity and water. Feed represented the largest factor in all management scenarios, followed by labor, then variable and fixed costs. This follows the same results found in previously published models where 57% of total cost were due to feed costs [6].



#### *Animals* **2019** , *9*, 471

Using the assumptions in Table 1, on average, the most expensive management style was the one utilizing group housing, feeding pasteurized whole milk with conventional labor. The least expensive management pathway was the one utilizing individual housing outside, feeding whole milk with conventional labor. The main difference in cost can be attributed to the larger infrastructure needs for group housing and the additional cost of a pasteurizer. Total and daily cost for all management scenarios with 6, 8, 10, and 12 L allotments are shown in Table 3.

The mean for total cost ranged between \$258.56 to \$582.98 per calf across all management pathways. As seen in previous literature, the mean cost in each milk allotment has less variation than when looking at the range of projected costs per management scenario. This can be attributed to variation in health and mortality rates. Increasing the mortality rate and disease prevalence increased the cost for the remaining calves by spreading infrastructure costs, the loss of the calf and incurred expenses, and additional illness treatments over fewer calves. Variation in costs is not always related to efficiency on-farm but instead related to trade-offs in management styles.

The least expensive pathways were the 3 combinations for individual housing outside. In these scenarios, housing cost contributed 7–8% of the total cost compared to other management pathways utilizing more infrastructure, where housing accounted for 21–30% of the total cost. The addition of barns with individual housing inside and in group housing was the contribution of the additional 14–23% of housing cost.

When costs were broken down by day, assuming a 65-day weaning age, the cost ranged from \$3.83 to \$6.19 per calf per day. The average daily charge for a contract raiser from birth to weaning was \$1.88/day [16]. Based on our calculated total cost for rearing pre-weaning calves, this would create a significant loss for the contract raiser. But in the Wisconsin heifer raising survey the cost per day of fixed and variable costs, which most closely matches our model, \$2.05–\$8.73 for minimum and maximum daily cost [8]. This simulation model can be compared to surveys to validate the results are reflective of on-farm total values. Finally, this model can be used to estimate other housing situations other than the ones presented in the current survey, and herd numbers could be used to estimate heifer costs for individual herds. In this survey, we chose the most common management practices to raise dairy heifers in North America, but there are many other options of housing and feeding dairy heifer calves that require further investigation.

#### **4. Conclusions**

Raising calves from birth to weaning contributes to a major portion of the total heifer raising cost. Milk and calf starter contributed over half the cost to raise a calf from birth to weaning. Costs calculated by this model are based on currently available data; it is likely some of our assumptions will under or overestimate total and specific costs of calf raising practices across the US. More data are needed to improve accurate assumptions for farms. However, no model will be able to accurately describe all situations of calf rearing in various locations. Calculating pre-weaning cost for each individual farm is critical in making management decisions and remaining sustainable.

**Author Contributions:** Conceptualization, A.H. and J.H.C.C.; methodology, A.H. and J.H.C.C.; formal analysis, A.H., K.B., and J.H.C.C.; investigation, A.H.; resources, J.H.C.C.; writing—original draft preparation, A.H. and J.H.C.C.; writing—review and editing, A.H., K.B., D.A.-P., and J.H.C.C.; visualization, A.H. and K.B.; supervision, K.B. and J.H.C.C.; project administration, J.H.C.C.

**Funding:** The authors would like to thank Dannon for their continued support of this project.

**Acknowledgments:** The authors would like to acknowledge the dairy producers in Kentucky, Indiana and Ohio, USA that provided feedback and insight into our model design. We are thankful for the time and effort they spent with us consulting on this project. The authors would like to thank Dr. Jeffrey Bewley and Dr. Michael Overton for the input and assistance during the development of this economic model. This open model can be found on https://afs.ca.uky.edu/dairy/decision-tools.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Mortality-Culling Rates of Dairy Calves and Replacement Heifers and Its Risk Factors in Holstein Cattle**

**Hailiang Zhang 1, Yachun Wang 1,\*, Yao Chang 1, Hanpeng Luo 1, Luiz F. Brito 2, Yixin Dong 1, Rui Shi 1, Yajing Wang 3, Ganghui Dong <sup>4</sup> and Lin Liu <sup>5</sup>**


Received: 23 July 2019; Accepted: 25 September 2019; Published: 26 September 2019

**Simple Summary:** High mortality and involuntary culling rates cause great economic losses to the dairy industry around the world, and the survival of dairy calves and replacement heifers is paramount in modern dairy breeding. However, little has been done to genetically improve mortality rates of dairy calves and replacement heifers in Chinese Holstein cattle. In this study, we investigated population parameters (descriptive statistics) of mortality rates of dairy calves and replacement heifers and risk factors affecting mortality and involuntary culling rates in Chinese Holstein cattle. The mortality rate of dairy calves and replacement heifers from day 3 to 60, 61 to 365, and 366 to first calving was 5.5%, 7.4%, and 8.7%, and an unfavorable increasing trend has been observed in the Chinese Holstein population. Health events associated with digestive and respiratory or circulatory systems were the main reasons for deaths. Herd-birth year, birth season, and dam parity had significant effects on survival. Our findings will help farmers to better manage dairy calves and replacement heifers and highlight the need to include these survival traits as part of the national genetic evaluation schemes.

**Abstract:** The rates of mortality and involuntary culling of dairy calves and replacement heifers have great economic implications on the dairy cattle industry around the world. The main objectives of this study were: (1) to obtain population parameters of mortality and involuntary culling rates of dairy calves and replacement heifers; and, (2) to investigate the factors affecting mortality and involuntary culling rates in Chinese Holstein cattle. Two datasets containing records of birth, calving, and culling events from 142,833 Holstein cattle born between 1991 and 2018 were used in this study. The population parameters were obtained using dataset 1, which consisted of dairy calves and replacement heifers that died or were involuntarily culled. Three survival traits were defined in dataset 2, which consisted of females born from 1999 to 2018. A binomial logistic regression was used to analyze the risk factors on the survival traits. The mortality rate of dairy calves and replacement heifers from day 3 to 60, 61 to 365, and 366 to first calving was 5.5%, 7.4%, and 8.7%, and an unfavorable increasing trend was observed. Health events associated with digestive and respiratory or circulatory systems were the main death reasons. Herd-birth year, birth season, and dam parity had significant effects on survival traits. The results from this study will help farmers to

better manage calves and replacement heifers and highlight the need to include survival traits in dairy calves and replacement heifers as part of national genetic evaluation schemes.

**Keywords:** dairy calf; involuntary culling; mortality; replacement heifer; survival rate

#### **1. Introduction**

The large majority of dairy cattle herds are divided into two groups: milking cows and replacement heifers, in which the latter does not generate any direct income to the producers until the first calving [1,2]. Estimates of expenses associated with rearing replacement heifers range from 15% to 20% of the total milk production costs [3]. However, many potential replacement heifers do not reach their first lactation due to premature death or involuntary culling [2]. Therefore, in addition to welfare issues, high mortality and culling rates cause great economic losses to the dairy industry around the world [4,5]. Various breeding programs include indicator traits of health and longevity measured in dairy cows [6]. However, less importance has been given to survivability of dairy calves and replacement heifers.

The mortality rates of calves and replacement heifers vary across countries, production systems, and populations. For instance, in the United States, annual calf and heifer mortality rates were estimated to be around 9.6%, with pre-weaning calves accounting for 7.8% [7]. In Danish Holstein, the frequency of pre-pubertal mortality was estimated to be 5% to 6% [8], and the most frequent diseases affecting calves were diarrhea and respiratory diseases. Other studies showed that the main causes of replacement heifer mortality or involuntary culling were different depending on the life stage of the animal. Scours, diarrhea, and other digestive problems were the most important death causes of pre-weaning calves, while respiratory diseases were the largest death cause of weaned calves [9]. Furthermore, many herd- or animal-level risk factors affecting dairy calf and heifer mortality have been identified in various populations. These factors include dystocia, sex, twinning rate, dam parity, herd size, and birth season [10–13].

The impacts of mortality rates of calf and replacement heifer on dairy cattle herds should not be neglected. However, there is a lack of literature reporting on this issue in calves and replacement heifers in Chinese Holstein population, especially studies based on individual records. In this context, the main objectives of this study were: (1) to estimate population parameters (descriptive statistics) of mortality and involuntary culling rates of dairy calves and replacement heifers; and, (2) to investigate the risk factors affecting mortality and involuntary culling of dairy calves and heifers in Chinese Holstein cattle, using individual records. The findings of this study will help farmers to design better management strategies for the dairy calves and replacement heifers, and provide a reference for further investigation on the genetic background of survival traits in dairy calves and replacement heifers.

#### **2. Materials and Methods**

The records of birth, calving and culling/death events from 1999 to 2018 in female Holstein from 31 herds located in Beijing, Tianjin, Yunnan, Hebei, Henan, Heilongjiang, Jilin, and Inner Mongolia were extracted from the farm management software (AfiFarm, http://www.afimilk.com.cn). The free-stall barn system was used in all herds, and the herds' sizes ranged from 1000 to 10,000 animals. The test-day milk yield in these herds ranged from 30 kg to 40 kg. The herd records before using management software (before 2005) were added into software from herdbook records, and thus, early records might be incomplete or less accurate. Two datasets were defined using event records: Dataset 1 and dataset 2. Dataset 1 included records of dairy calves and replacement heifers that left herds (before the first calving) between 2006 and 2018, which was used to obtain population parameters of involuntary culling/death age on dairy calves and replacement heifers in Chinese Holstein cattle, including average involuntary culling/death age and culling/death reason. The dataset 2 included records from all animals (that left herds either before or after first calving) born from 1999 to 2018, which was only

used to investigate risk factors affecting survivability of dairy calves and replacement heifers using logistic regression. In Chinese Holstein herds, most calves left the herd due to premature death, while heifers can also be culled for reproduction disorders, severe disease and other reasons. In this study, we are interested in both mortality and involuntary culling. The records of calves that died within 2 days after birth, replacement heifers that died after 1800 days of age (60 months) and censored records (alive dairy calves and replacement heifers, and sold and transferred individuals) were removed from both dataset 1 and dataset 2. The death records before first 48 h were considered as stillbirth, which is usually a separate dam trait and thus is not part of the current study. In dataset 1, the involuntary culling/death reasons were grouped in a total of 10 categories: digestive system diseases, diseases of respiratory or circulatory systems, reproduction disorders, death without clear reasons, infectious diseases, developmental disorders (e.g., abnormalities and dysplasia), feet and leg diseases, accidental injury, other diseases (e.g., septicemia and meningitis) and unknown reason. Only females were kept in the datasets. After editing, records for 18,077 culled dairy calves and replacement heifers remained in dataset 1 and 113,218 records of all animals in dataset 2. Death/culling age (days) was calculated for each animal in dataset 1 and referred to the interval from birth to death/culling on both dairy calves and replacement heifers. A total of 3 survival traits were defined for females in dataset 2, including survival from 3 to 60 days (Sur1), 61 to 365 days (Sur2), and from day 366 to first calving (Sur3). Survival traits were analyzed as binary traits, in which a value of "0" was assigned to animals that left the herds and "1" to those that survived up to next life stage.

A binomial logistic regression was used to evaluate the risk factors affecting survivability of dairy calves and replacement heifers using the LOGISTIC procedure of SAS software (version 9.1; SAS Institute, 2004 [14]). A total of 4 risk factors associated with survival traits were analyzed using the dataset 2. These factors were herd-birth year (402 levels), birth season (divided into Spring: March to May, Summer: June to August, Fall: September to November, and Winter: December to February), dam parity (defined as 0 = unknown, 1 = first parity, 2 = second parity, 3 = third and greater parities), calving ease score (defined as 0 = unknown, 1 = unassisted, 2 = easy pull, and 3 = hard pull or surgery). The factor of herd-birth year represented the combined effect of herd and birth year of calf, and 402 levels were combined into 5 levels using logit (*p*) of each level. The statistical model for Sur1, Sur2, and Sur3 can be described as follow:

$$\text{logit}(p) = \ln\left(\frac{p}{1-p}\right) = \beta\_0 + \beta\_1\mathbf{x}\_1 + \beta\_2\mathbf{x}\_2 + \beta\_3\mathbf{x}\_3 + \beta\_4\mathbf{x}\_4.$$

where *p* is the culling probability of dairy calves and replacement heifers in each life stage; β<sup>0</sup> is the overall mean (intercept); β<sup>1</sup> to β<sup>4</sup> are the regression coefficients of ranked factors; *x*1, *x*2, *x*3, and *x*<sup>4</sup> correspond to the herd-birth year, birth season, dam parity and calving ease score, respectively, associated with each observation.

#### **3. Results**

#### *3.1. Descriptive Statistics*

#### 3.1.1. Mortality-Culling Frequency of Dairy Calves and Replacement Heifers

The combined mortality-culling rate of female dairy calves and replacement heifers was 21.2%, in which the mortality-culling rate from day 3 to 60, 61 to 365, and 366 to first calving was 5.5%, 7.4%, and 8.7%, respectively. The variability in mortality-culling within each life stage over time is shown in Figure 1. From 2006 to 2008, the mortality-culling rate of dairy calves and replacement heifers increased from 15.2% to 25.9% (an increase rate of 70.4%).

#### 3.1.2. Death/Culling Age of Dairy Calves and Replacement Heifers

Descriptive statistics of death/culling age for 18,077 dairy calves and replacement heifers born from 2006 to 2018 were calculated. The average death/culling age was 399 days; the median 296 days and the lower and upper quartile were equal to 84 and 658 days, respectively. The death/culling age did not follow a normal distribution as the large majority of calves died early in life (Figure 2). The highest mortality-culling risk on dairy calves was within the first 100 days after birth. Mortality-culling tended to decrease with the increase of animal age. The mean and median of death/culling age over time are presented in Figure 3. Over the past 13 years, there was a large difference on death/culling age of dairy calves and replacement heifers and the average death/culling age fluctuated around 400 days. Furthermore, the variation range of median death/culling age was larger than the means among different years, and the maximum difference of median death/culling age was 337 days (between 2006 and 2012).

**Figure 1.** Mortality-culling rates of dairy calves and replacement heifers in different life stages over time.

**Figure 2.** Histogram of death/culling age (months) on dairy calves and replacement heifers.

**Figure 3.** Average death/culling age (days) and number of dairy calves and replacement heifers in different death years.

#### *3.2. Death*/*Culling Reasons of Dairy Calves and Replacement Heifers*

The overall death proportion of each reason category in period from day 3 to first calving in different years is presented in Figure 4. "Unknown reason" was not included in Figure 4, which was reached 8.99%–43.78% over these years. Diseases related to the digestive, respiratory and circulatory systems, and reproductive disorders were the main causes of death. These categories accounted for 58.6% of the known causes of death. Diarrhea, pneumonia, and infertility (based on non-return rate) were the main specific reasons of mortality. Over the past 13 years, the death proportion due to diseases related to the respiratory and circulatory systems and reproductive disorders gradually increased, in contrast with both infectious diseases and other diseases that showed a decrease trend. The average involuntary culling or death ages of dairy calves and replacement heifers based on different death reason categories are presented in Table 1. The individuals with digestive system diseases, diseases

of respiratory or circulatory systems, and death without clear reason were culled in early life (mean: up to 226.6 days; median: up to 101.0 days). As expected, the average death age of individuals with reproductive disorders (mean: 937.4; median: 891.0) was greater compared to the other categories (mean range: 165.1–476.7 days; median range: 84.0–450.5 days).


**Table 1.** The death age (days) of dairy calves and replacement heifers caused by different reason categories.

**Figure 4.** The death proportions of dairy calves and replacement heifers caused by different reason categories over time.1 The category of "unknown reason" was not included in Figure 4.

#### *3.3. Analyses of the Factors Influencing Survivability of Dairy Calves and Replacement Heifers*

According to the Wald test (Chi-square), both herd-birth year and birth season significantly (*p* < 0.01) influenced the mortality of dairy calves and replacement heifers during the stages of 3–60 days (Sur1), 61–365 days (Sur2), and from 366 days to first calving (Sur3). The dam parity significantly influenced Sur1 and Sur2 (*p* < 0.01), and did not significantly influence Sur3 (*p* = 0.19). However, dam calving ease score did not significantly impact any of the 3 survival traits. The results of the binomial

logistic regression on Sur1, Sur2, and Sur3 in dairy calves and replacement heifers are presented in Table 2.

The dairy calves and replacement heifers born in Spring had the lowest mortality risk in any of the 3 life stages. Across the 3 life stages (Sur1, Sur2, and Sur3), the mortality risk of dairy calves and replacement heifers born in Fall was between 1.13 and 1.53 times greater than those animals born in the Spring season. The calf birth season had larger impact on survivability of animals during 3–365 days (Sur1 and Sur2) compared to 366 days to first calving. In terms of dam parity, calves born from second parity cows had the lowest culling risk in any of the 3 life stages. From 3 to 60 days, calves born from first parity cows had the highest culling risk, i.e., 1.16 times greater than those animals born from second parity cows. However, the dairy calves and replacement heifers born from cows with 3 or more parities had the highest mortality risk during 61–365 and 366–first calving. Animals born by hard pull or surgery had the highest mortality risk, which was not significant compared with unassisted calves.


**Table 2.** Associations between different levels of birth season, dam parity and calving ease score on the odds ratio of mortality 1.

<sup>1</sup> The Spring season, second parity and calving ease score 1 were the base classes of birth season, dam parity and dam calving ease score, respectively. The results of herd-birth year are not shown.

#### **4. Discussion**

The involuntary culling and mortality rates of dairy calves and replacement heifers have been reported to vary across countries and dairy cattle populations. In the population used for the current study, dairy calves were usually weaned at 2 months of age, and the pre-weaning mortality rate was 5.5%, which is within the range reported in the literature. For instance, the calf mortality within the first month of life were 3.1%–3.4% in Danish [15] and UK [2] Holstein populations, while in the US, the mortality of pre-weaning Holstein calves was 7.8% [7]. The mortality rate of 12.9% for dairy calves up to yearling age is also within the ranges reported in the literature for worldwide dairy populations (3.7%–22.5%) [16,17]. Approximately 21.2% of dairy calves and heifers failed to reach first calving, which is substantially higher compared to other reports (e.g., 14.5%) [18]. Furthermore, there was an unfavorable increase trend on mortality of dairy calves and replacement heifers over time in Chinese Holstein population. The calf and replacement heifer survival between day 3 and the start of productive life should be given more attention, especially for genetically select animals with better genetic merit for survival traits.

Many factors have caused mortality of dairy calves and replacement heifers, including calf-related diseases, heifer fertility disorders and farm management factors. In this study, censored and voluntary culling records were removed from the datasets. Therefore, the mortality rates reported here represent involuntary culling in Chinese Holstein calves and replacement heifers. Due to poor data management and insufficient attention paid on data recording, culling/death reason were not always available for each animal, especially in early records. In general, the 2 most frequent causes of mortality are digestive [19,20] and respiratory diseases [21], in which diarrhea and pneumonia accounts for the majority of death cases [9,20,22]. In this study, diseases associated with the digestive, respiratory and circulatory systems were the main culling reasons, which is consistent with the findings reported in other studies. In the US Holstein cattle population, scours, diarrhea, and other digestive problems were the key causes of pre-weaning calf mortality, followed by respiratory diseases. For weaned calves, respiratory disease was the largest mortality reason in the US population [9]. Pritchard et al. [1] reported that a large number of heifers were culled due to been considered unsuitable as breeding replacement, failure to conceive and other reproduction disorders, which is in agreement with our findings. The median death age of calves or replacement heifers due to digestive system diseases, diseases of respiratory or circulatory systems, and reproductive disorders were 84, 101, and 891 days, respectively. Furthermore, the main causes of mortality were different over these years in calves and replacement heifers. The fertility recession and more attention on epidemic prevention may respectively result in increase/decrease trends of culling/death proportion of reproductive disorders/infectious diseases over these years.

Considering the impacts of management differences across herds and a likely interaction with birth year, the effect of herd was included in the statistical model as a combined effect (herd-birth year) with birth year of the calf in current study. Herd-birth year significantly impacted all survival traits consistent with Norberg et al. [8]. Birth season and dam parity significantly impacted all survival traits, which is in agreement with results reported by Norberg et al. [8] and Gulliksen et al. [16]. During Summer and Fall, animals can be under heat stress in the main dairy farming areas in China (including the herds in the current study). The calves that experience maternal heat stress during late gestation have been reported to have reduced survival rate before puberty [23]. Dairy producers can plan the calving accordingly in order to reduce calving mortality rates and/or implement other mitigation approaches. Furthermore, Henderson et al. [24] and Ring et al. [25] reported that dam calving ease score was an important risk factor of mortality, especially within the first 182 days of life. According to them, the calves and heifers born from increased calving ease score were more likely to die in early life stages. This is likely due to the stress suffered by calves during birth. The dam calving ease score had no statistically significant impact on the survival traits analyzed here, which may be related a small data size in current study. These information from risk factors will help farmers to reduce mortality rate of calves by implementing better management practices in their herds. In addition, the influence of the risk factors identified here will be important effects to be included in the statistical models for genetic evaluation for survival traits in dairy calves and replacement heifers.

Survival traits defined at different life periods during replacement heifer development may enable selection against certain diseases commonly prevalent during those life stages [1]. Three survival traits were defined in this study aiming to describe the likelihood of death (or survival) during pre-weaning period, day 61 to yearling and yearling to the first calving. Survival trait at early life of the calf (Sur1) may enable indirect selection against diseases related to the digestive, respiratory, and circulatory systems, which were the two most frequent mortality reason categories. The survival traits at different life stages, defined in this study, can be used to genetically improve the mortality rates of dairy calf and replacement heifers, and the results from the current study laid the foundation for establishing the statistical models for genetic evaluations. The next step will be the estimation of genetic parameters (heritability and genetic correlations) for the survival traits defined in this study.

#### **5. Conclusions**

The combined mortality rate of dairy calves and replacement heifers in Chinese Holstein cattle was 21.2% and an unfavorable trend on dairy calves and replacement heifer' mortality was observed. Diseases related with digestive (e.g., diarrhea), respiratory (e.g., pneumonia) and circulatory systems, and reproductive disorders (infertility based on non-return rate) were the main death reason categories. Herd-birth year, birth season, and parity of dam had significant effects on the survival traits of dairy calves and replacement heifers. Survival traits in dairy cattle from birth to first calving are important breeding goals to be incorporated into dairy genetic selection schemes.

**Author Contributions:** Data curation, H.L. and Y.D.; Formal analysis, H.Z.; Methodology, H.Z. and Y.C.; Project administration, Y.W.; Resources, G.D. and L.L.; Supervision, Y.W.; Visualization, R.S.; Writing—original draft, H.Z.; Writing—review & editing, L.B. and Y.W.

**Funding:** This research was funded by Modern Agro-industry Technology Research System (CARS-36); Beijing Dairy Industry Innovation Team (BAIC06-2019, Beijing, China); Beijing Sciences and Technology Program (D171100002417001); the Program for Changjiang Scholar and Innovation Research Team in University (IRT\_15R62); and Postgraduate Internationalization Training Promotion Project of CAU (31051521, Beijing, China).

**Conflicts of Interest:** The authors declare no conflict of interest. The funding providers had no role in the design, execution, interpretation, or writing of the study.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **The E**ff**ect of Heat Stress on Autophagy and Apoptosis of Rumen, Abomasum, Duodenum, Liver and Kidney Cells in Calves**

### **Ruina Zhai 1,**†**, Xusheng Dong 2,**†**, Lei Feng 2, Shengli Li 3,\* and Zhiyong Hu 2,\***


Received: 17 September 2019; Accepted: 21 October 2019; Published: 22 October 2019

**Simple Summary:** Heat stress causes significant negative responses in the dairy industry. The objective of this study was to assess the effect of heat stress on the autophagy and apoptosis of the rumen, abomasum, duodenum, liver and kidney in calves. The results showed that heat stress could increase the autophagy and apoptosis of the kidney, duodenum and abomasum. However, heat stress had no effect on the autophagy and apoptosis of the liver. In cows, most studies of autophagy and apoptosis have only focused on mammary remodeling. Our results provide new knowledge regarding autophagy and autophagy in calf heat stress management.

**Abstract:** The objective of this study was to assess the effect of heat stress on the autophagy and apoptosis of the rumen, abomasum, duodenum, liver and kidney in calves. Two groups of Holstein male calves were selected with similar birth weights and health conditions. Heat stress (HT): Six calves (birth weight 42.2 ± 2.3) were raised from July 15 to August 19. Cooling (CL): Six calves (birth weight 41.5 ± 3.1 kg) were raised from April 10 to May 15. All the calves were euthanized following captive bolt gun stunning at 35 d of age. The expression of protein 1 light chain 3-II (LC3-II) and caspase3 in the rumen, abomasum, duodenum, liver and kidney were determined by western blotting. In addition, other possible relevant serum biochemical parameters were evaluated. Significant differences were observed in alkaline phosphatase (ALP), albumin (ALB) and glucose (Glu). The results showed that heat stress could increase the autophagy and apoptosis of the kidney, duodenum and abomasum. However, heat stress had no effect on the autophagy and apoptosis of the liver. Additionally, the expression of caspase-3 in the rumen in HT was significantly lower than that in CL. In conclusion, the effects of heat stress on autophagy and apoptosis are organ-specific. The results provide knowledge regarding autophagy and autophagy in calf heat stress management.

**Keywords:** autophagy; apoptosis; heat stress; calf

#### **1. Introduction**

Heat stress causes significant negative responses in the dairy industry. Heat stress is caused by excessive temperature conditions that cannot be compensated for by the temperature regulation mechanism of cows. The temperature regulation ability is weaker for young calves than for adult cows, with an upper end of about 29 ◦C, and heat stress is considered to occur at temperatures greater than 32 ◦C and 60% humidity [1]. Extensive research has shown the diminution of liver enzyme activities and kidney functions in cows during heat stress [2,3]. In rats and pigs, heat stress apparently promotes intestinal mucosal damage due to reduced intestinal blood flow and tissue hyperthermia [4,5]. The reduced blood flow of the digestive tract is probably harmful to the barrier function integrity of rumen [6].

Autophagy and apoptosis are universal mechanisms that regulate gut homeostasis and reduce digestive tract damage [7,8]. When cells are under stress, autophagy and apoptosis are activated [9]. Autophagy is a specific protein degradation process that has been recognized as an important mechanism for cell survival under stress conditions [10,11]. The protein 1 light chain 3-II (LC3-II) is a useful marker of autophagic membranes and is essential for the expansion of the early autophagosome during cellular house-keeping and autophagic cell death [12,13]. In elegans, autophagy-related genes are transcriptionally upregulated in response to heat shock [14]. Heat stress also triggers autophagy in different types of cells such as human alveolar basal epithelial cells and rat cardiomyocytes [15]. In vivo apoptosis and autophagy are two forms of physiological and conserved programmed cell death [16]. Apoptosis is characterized by a series of morphological changes, including plasma membrane blebbing, nuclear condensation, and fragmentation, all of which lead to the formation of apoptotic bodies [9]. In general, autophagy is activated first and maintains cell homeostasis [17]. When stress is prolonged or exceeds a threshold, apoptosis is activated [9,18]. In cows, heat stress can induce the apoptosis of granulosa cells, as evidenced by the activation of caspase-3 [19]. Caspase-3 is an executioner caspase which plays an important role in apoptosis [9].

In cows, most studies of autophagy and apoptosis have only focused on mammary remodeling [20–22]. Thus far, very little attention has been paid to the role of autophagy and apoptosis in calves. The objective of this study was to assess the effect of heat stress on the autophagy and apoptosis of the rumen, abomasum, duodenum, liver and kidney in calves. We hypothesized that autophagy and autophagy would be stimulated to relieve heat stress. We hope to provide knowledge regarding autophagy and autophagy in calf heat stress management.

#### **2. Materials and Methods**

The study was conducted at the Shandong high-speed modern dairy farm in Ji Ning, Shandong, China in 2018. Animal care and use were approved and conducted under established standards of the Ethics Committee on animals of Shandong Agricultural University (SDAUA-2018-012).

Two groups of Holstein male calves were selected with similar birth weights and health conditions. Heat stress (HT): Six calves (birth weight 42.2 ± 2.3) were raised from July 15 to August 19. Cooling (CL): Six calves (birth weight 41.5 ± 3.1 kg) were raised from April 10 to May 15. Calves were individually housed in 1.5 × 3.4 m pens inside a naturally ventilated barn with free-choice water and solid feed. The HT calves were housed in the same pens and were only provided with shade. The relative humidity and air temperature of each pen were recorded at 7 days before euthanasia. The temperature humidity index (THI) of the HT and CL groups were calculated as described previously [23]. In the HT group, the average THI was 85.08, whereas the average the THI was 63.49 in the CL group. All the calves were provided with 4 L of colostrum within 2 hours from birth. From the next day, calves were fed 6 L of whole milk once daily until being euthanized. The ingredient and nutrient composition of the calf starter is given in Table 1.

The rectal temperature and respiratory rate of each calf were recorded 3 times per day. At 35 ± 2 d of age, fifteen milliliters of blood were collected from the caudal vein using 20 ml syringes. The samples were collected in the procoagulant tube, centrifuged at 1000× *g* for 15 min, and then the serum was collected in a 1.5 ml Eppendorf tube and stored at −20 ◦C. All the calves were euthanized following captive bolt gun stunning at 35 ± 2 d of age. After the opening of the body cavity, the samples of the rumen, abomasum, duodenum (entire wall from 6 cm distal to the pylorus), liver and left kidney were washed with normal saline. Then, these samples were frozen in liquid nitrogen and stored at −80 ◦C until western blotting was performed.


**Table 1.** Ingredient and nutrient composition of the experimental calves' starter.

DM: dry matter; CP: crude protein; NDF: neutral detergent fiber; ADF: acid detergent fiber; ME: metabolizable energy.<sup>1</sup> Premix contained (mg/kg): vitamin A, 4,035; vitamin D, 1,740; vitamin E, 39; Fe, 18; Zn, 37; Cu, 10.6; Mn, 15.3; Co, 0.12; I, 0.47; and Se, 0.35.

Briefly, the tissue samples blocks were washed with PBS (Solarbio, P1020-500ml, Beijing, China), cut into small pieces, homogenized in PBS at 4 ◦C using a Servicebio KZ-II homogenizer, kept on ice for 0.5 h, oscillated to ensure complete tissue cracking every 5 min, and then centrifuged (3000 × g, 10 min, 4 ◦C). Protein concentration was detected by a bicinchoninic acid (BCA) Protein Assay Kit (G2026, Servicebio, Wuhan, China). A Laemmli sample buffer (Bio-Rad, 1610737, Shanghai, China) was used to dilute the sample and then boiled for 5 min. Immunoblotting was performed as previously described [24,25]. The LC3 (Sigma-Aldrich, L8918, Shanghai, China), caspase-3 (Sigma-Aldrich, C8487, Shanghai, China) and β-actin (Sigma-Aldrich, A2066, Shanghai, China) antibodies and appropriate secondary antibodies (Servicebio, GB23303, Wuhan, China) were applied according to the manufacturer's guidelines. The chemiluminescence of the bands of interest was detected with a digital G: Box imager (Syngene, Frederick, MD, USA). Image J software (National Institutes of Health, Bethesda, MD, USA) was used to quantify the band density. Alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total protein (TP), albumin (ALB), glucose (Glu) and total cholesterol (TCHO) of the serum were determined with an automatic biochemical analyzer (Type 7020, Hitachi, Tokyo, Japan).

The data were analyzed with a completely randomized design using a one-way ANOVA of SAS 8.2 (SAS Institute Inc., Cary, NC, USA). The individual calf was considered as the experimental unit. The means were compared using Duncan's multiple range test. Significance was declared at *p* < 0.05.

#### **3. Results**

In the HT group, the average THI was 85.08, whereas the average THI was 63.49 in the CL group. It could also be seen that the rectal temperature (39.36 ± 0.26 vs. 38.31 ± 0.19 ◦C, respectively; *p* < 0.01) and respiratory rate (55.17 ± 3.49 vs. 34.17 ± 2.48 breaths per minute; respectively; *p* < 0.01) of the HT calves were significantly higher than the CL calves. Those data indicated that the HT calves were exposed to heat stress, while the CL calves were not subjected to heat stress.

No differences were observed in the concentrations of ALT, AST, TP and TCHO in plasma of two groups (Table 2). ALP and ALB in the CL group were significantly higher than that in the HT group (*p* < 0.05), while the CL calves had a lower amount of serum Glu (*p* < 0.05). Compared with the CL calves, the HT calves had a lower expression of LC3-II in the kidney (*p* < 0.05) and tended to have a lower expression in the duodenum, abomasum and rumen (Figure 1). The expressions of caspase-3 in the kidney, duodenum and abomasum were elevated in the HT calves relative to the CL calves (*p* < 0.01). No significant differences were found in caspase-3 expression of the liver between the two groups (Figure 2). Interestingly, the expression of caspase-3 in the rumen in the HT group was significantly lower than that of the CT group.

**Figure 1.** Effects of the temperature humidity index (THI) on the microtubule-associated protein 1 light chain 3-II (LC3-II) expression of the liver, kidney, duodenum, abomasum and rumen in the HT and CL calves. Treatment was as follows: (1) HT: Calves were fed from July 15 to August 19; (2) CL: Calves were fed from April 10 to May 15. Insets depict representative blots. Values represent means ± standard deviation Response from statistical result, *p* < 0.05. β-Actin was used to normalize the expression of target proteins. The letters below the bar graph indicate different organs. Different letters above the bar indicate differences between different groups (*p* < 0.05).

**Table 2.** Serum concentrations of alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total protein (TP), albumin (ALB), glucose (Glu) and total cholesterol (TCHO) of the calves in the CL (cooling) and HT (heat stress) groups.


**Figure 2.** Effects of the THI on the caspase3 expression of the liver, kidney, duodenum, abomasum and rumen in the HT and CL calves. Treatment was as follows: (1) HT: Calves were fed from July 15 to August 19; (2) CL: Calves were fed from April 10 to May 15. Insets depict representative blots. Values represent means ± standard deviation. Response from statistical result, *p* < 0.05. β-Actin was used to normalize the expression of target proteins. The letters below the bar graph indicate different organs. Different letters above the bar indicate differences between different groups (*p* < 0.05).

#### **4. Discussion**

In our study, the THI, rectal temperature and respiratory rate indicated that the HT calves were exposed to significant environmental heat stress, while the CL calves did not suffer from heat stress. The current study found that the HT group had a higher concentration of Glu than the CL group. This result agrees with an earlier study, which showed that Glu tended to have a higher concentration in an HT group [26]. However, a previous study found that pancreatic insulin response to Glu stimulation and the concentration of insulin in calves were not affected by the heat stress [27,28]. It seems possible that the high concentration of serum Glu was connected to the impaired cell metabolism and osmolarity caused by heat stress without the change of insulin. It has been suggested that ALP plays a vital role in bone mineralization and hepatobiliary diseases [29,30]. The low concentration of ALP in the HT calves may impact the development of bone and hepatobiliary. ALB may be able to regulate oncotic pressure and modulate inflammatory or immunological responses [31]. In our study, the concentration of serum ALB was lower in the HT calves than the CL calves. Our result agrees with a previous study, indicating that heat stress could decrease the concentration of serum ALB, which perhaps, in turn, could impact oncotic pressure [31].

We analyzed the expression of LC3-II and caspase-3 of the HT and CL calves to assess the effect of the THI on autophagy and apoptosis in calves. In our study, we found that different organs have different levels of autophagy and apoptosis. All the organs in our study except liver were affected by heat stress. In the duodenum and abomasum, especially the kidney, the level of autophagy and apoptosis were increased by heat stress. A previous study found that heat stress could induce autophagy in hepatocellular carcinoma [32]. In the skeletal muscle of *Sus scrofa*, the markers of autophagosome formation and autophagic activation were increased by heat stress [33]. These results were similar to those of our study. The enhanced autophagy may be a self-protection mechanism of organs under heat stress.

Hyperosmolarity, which is one of the consequences of heat stress, could activate several mediator systems that may cause renal injury [34]. Calves extensively sweating results in a serious loss of water and salt, which could lead to an increase of urine-specific gravity and osmolarity [34]. In our study, the high concentration of Glu in the HT calves may have been due to the increase of osmotic pressure. Additionally, the activity of aldose reductase, which can convert Glu into sorbitol and is increased by hyper osmolarity [35]. Sorbitol can protect kidney cells from the hyperosmotic environments under the conditions of plasma hyperosmolarity and dehydration [36,37]. Previous studies have found that hyperosmotic stress could induce apoptosis and suppressed mammalian target of rapamycin complex 1 (mTORC1) which could inhibit autophagy [38,39]. However, sorbitol dehydrogenase could convert sorbitol into fructose. In the small intestine, the metabolism of fructose is associated with local inflammation and increased intestinal permeability [40,41]. These physiological duodenal changes probably caused the high level of apoptosis in the duodenum seen in our study.

One interesting finding was that the level of autophagy and apoptosis in the liver were not affected by heat stress. Autophagy, which is a response to stressful conditions of the liver, could eliminate damaged mitochondria and accumulated lipid droplets in liver [42]. Both autophagy and apoptosis play important roles in liver injury [42]. It seems possible that the liver has a strong regulatory ability to reduce the damage caused by heat stress. Additionally, the apoptosis level was decreased by heat stress in the rumen. It seems that the effects of heat stress on autophagy and apoptosis are organ-specific. Further work is required to evaluate the effect of osmotic pressure on autophagy and apoptosis. In addition, further study should focus on the effects of heat stress on the liver.

#### **5. Conclusions**

In conclusion, heat stress could increase the level of autophagy and apoptosis of the kidney, duodenum and abomasum. However, heat stress has no effect on the autophagy and apoptosis of the liver. Heat stress decreased serum ALP and ALB and increased Glu concentration. In conclusion, the effects of heat stress on autophagy and apoptosis are organ-specific. These results provide knowledge regarding autophagy and autophagy in calf heat stress management.

**Author Contributions:** Z.H. and S.L. conceived and designed the experiments; X.D. and R.Z. performed the experiments; X.D. and L.F. analyzed the data; X.D. and R.Z. wrote the paper.

**Funding:** This research was funded by the National Natural Science Foundation of China, grant number 31772624, National Key Research and Development Program of China, grant number 2018YFD0501600.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
